American Journal of Obstetrics & Gynecology (AJOG) - Sep05

Founded in 1920
TABLE OF CONTENTS September 2005
EDITORS’ CHOICE
Pediatric presence at cesarean section: Justified or not? 599
Adrienne Gordon, MBChB, MRCP, FRACP,
Elizabeth Jane Mckechnie, MBBS, MRCP,
Heather Jeffery, MBBS, MRCP, FRACP, MPH, PhD
Sydney, Australia
The need for resuscitation is not significantly greater in infants delivered electively by
cesarean section under regional anesthesia compared with those delivered vaginally
provided there are no additional risk factors.
Commentary
Many hospitals require the presence of a pediatrician for all cesarean sections. This practice impacts the
cost of health care, but the practice has largely evolved as dogma with little care taken to break down the
need for the pediatrician’s presence based on the reason for the operative delivery. This thoughtful report
by Gordon et al finds no increased need for resuscitation in Cesarean sections done under regional
anesthesia except when the indication for the operation is for ‘‘fetal distress’’ or malpresentation. These
findings would logically lead to the conclusion that there is no more need for a pediatrician in the
operating room than in an otherwise uncomplicated spontaneous vaginal delivery.
High-dose methadone maintenance in pregnancy: Maternal and neonatal outcomes 606

John J. McCarthy, MD, Martin H. Leamon, MD, Michael S. Parr, MD,
Barbara Anania, PsyD
Sacramento and Davis, CA
Pregnant women who received a mean of 132 mg/day of methadone had less drug use
than a group of pregnant women who received 62 mg/day; there were no differences in
treatment for neonatal abstinence.
Commentary
The papers by McCarthy et al and Jansson et al in this issue provide contrasting information about the
benefits and risks of methadone maintenance in pregnancy. McCarthy et al found that higher dosages of
methadone were associated with less illicit drug use in mothers without increasing significant neonatal
withdrawal symptoms. Jansson et al caution however that methadone may have neurobehavioral and
cardiorespiratory effects on the fetus beyond withdrawal.
Contents continued on page 5A
The American Journal of Obstetrics and Gynecology (ISSN 0002-9378) is published monthly (twelve issues per year), by Elsevier
Inc., 360 Park Avenue South, New York, NY 10010-1710. Business Office: 1600 John F. Kennedy Blvd., Suite 1800, Philadelphia, PA
19103. Editorial Office: 360 Park Avenue South, New York, NY 10010. Accounting and Circulation Offices: 6277 Sea Harbor Drive,
Orlando, FL 32887-4800. Periodicals postage paid at Orlando, FL 32862 and additional mailing offices. POSTMASTER: Send
address changes to The American Journal of Obstetrics and Gynecology, Elsevier Periodicals Customer Service, 6277 Sea Harbor
Drive, Orlando, FL 32887-4800.
2A September 2005
TABLE OF CONTENTS continued
Fetal response to maternal methadone administration 611

Lauren M. Jansson, MD, Janet DiPietro, PhD, Andrea Elko, PA-C
Baltimore, MD
Maternal methadone administration is associated with significant changes in fetal
heart rate and movement that are independent of maternal effects.
Commentary
The papers by McCarthy et al and Jansson et al in this issue provide contrasting information about the
benefits and risks of methadone maintenance in pregnancy. McCarthy et al found that higher dosages of
methadone were associated with less illicit drug use in mothers without increasing significant neonatal
withdrawal symptoms. Jansson et al caution however that methadone may have neurobehavioral and
cardiorespiratory effects on the fetus beyond withdrawal.
RESEARCH METHODS: STATE OF THE SCIENCE

A new series of articles on research methodology in clinical medicine 618
George A. Macones, MD, MSCE, Series Editor
Philadelphia, PA
Issues in clinical trial design: Stopping a trial early and the large and simple trial 619
Elizabeth A. Thom, PhD, Mark A. Klebanoff, MD, MPH
Rockville and Bethesda, MD
We review the decision process of a Data and Safety Monitoring Committee in
recommending early termination of a clinical trial, and discuss the ‘‘large and simple
clinical trial’’ design.
CLINICAL OPINION
Research agenda for preterm birth: Recommendations from the March of Dimes 626
Nancy S. Green, MD, Karla Damus, RN, PhD, Joe Leigh Simpson, MD,
Jay Iams, MD, E. Albert Reece, MD, PhD, MBA, Calvin J. Hobel, MD,
Irwin R. Merkatz, MD, Michael F. Greene, MD, Richard H. Schwarz, MD,
and the March of Dimes Scientific Advisory Committee on Prematurity
White Plains, Bronx, and Brooklyn, NY, Houston, TX, Columbus, OH,
Little Rock, AR, Los Angeles, CA, and Boston, MA
A national research agenda to address the etiology and prevention of preterm
birth is proposed by The Scientific Advisory Committee on Prematurity of the
March of Dimes.
REVIEW ARTICLE
Headache as a side effect of combination estrogen-progestin oral contraceptives: 636
A systematic review
Elizabeth W. Loder, MD, Dawn C. Buse, PhD, Joan R. Golub, MD
Boston, MA
There is limited evidence to support the common belief that combination oral
contraceptives have an important influence on headache in most women.
September 2005 5A
GENERAL OBSTETRICS AND GYNECOLOGY: GYNECOLOGY

Colposcopic and histopathologic evaluation of women participating in 650
population-based screening for human papillomavirus deoxyribonucleic
acid persistence
Kristina Elfgren, MD, PhD, Eva Rylander, MD, PhD, Thomas Rådberg, MD,
PhD, Björn Strander, MD, Anders Strand, MD, PhD, Kirsti Paajanen, MD,
Inga Sjöberg, MD, PhD, Walter Ryd, MD, PhD, Ilvars Silins, MD, PhD,
Joakim Dillner, MD, PhD, for the Swedescreen Study Group
Stockholm, Gothenburg, Uppsala, Malmö, and Umeå, Sweden
Among women with a normal Papanicolaou smear participating in population-
based cervical screening, human papillomavirus deoxyribonucleic acid persistence
had a positive predictive value for detection of cervical intraepithelial neoplasia 2
or 3 of 29%.
The effect of bright light therapy on depression associated with premenstrual 658
dysphoric disorder
Catherine Krasnik, PhD, Victor M. Montori, MD, Gordon H. Guyatt, MD,
Diane Heels-Ansdell, MSc, Jason W. Busse, DC, for the Medically Unexplained
Syndromes Study Group
Hamilton, Ontario, Canada, and Rochester, MN
This meta-analysis suggests that the role of bright light therapy in the management of
premenstrual dysphoric disorder remains uncertain.
Tumor-specific p53 sequences in blood and peritoneal fluid of women with 662
epithelial ovarian cancer
Elizabeth M. Swisher, MD, Melissa Wollan, BS, Sarita M. Mahtani, BS,
Julia B. Willner, MD, Rochelle Garcia, MD, Barbara A. Goff, MD,
Mary-Claire King, PhD
Seattle, WA
Tumor-specific p53 DNA sequences may be identified in free DNA from blood and
peritoneal fluid of women with ovarian cancer.
Melanoma, thyroid, cervical, and colon cancer risk after use of fertility drugs 668
Michelle D. Althuis, PhD, Bert Scoccia, MD, Emmet J. Lamb, MD,
Kamran S. Moghissi, MD, Carolyn L. Westhoff, MD, Jerome E. Mabie, BS,
Louise A. Brinton, PhD
Bethesda and Rockville, MD, Chicago, IL, Stanford, CA, Detroit, MI,
and New York, NY
Fertility drugs do not increase risk of melanoma or thyroid, cervical, or colon cancer.
First glimpse of the functional benefits of clitoral hood piercings 675

Vaughn S. Millner, PhD, Bernard H. Eichold II, MD, DR PH,
Thomasina H. Sharpe, MD, Sherwood C. Lynn Jr, MD
Mobile, AL
This exploratory, descriptive study provides the first empirical glimpse of the
relationship between female genital piercing and sexual functioning.
Arcus tendineus fascia pelvis: A further understanding 677

Todd S. Albright, DO, Alan P. Gehrich, MD, Gary D. Davis, MD,
Farzaneh L. Sabi, MD, Jerome L. Buller, MD
Bethesda, MD
An average length for the arcus tendineus fascia pelvis is obtained, and associations to
fascial attachments, cadaver height, and pelvis type are explored.
6A September 2005
Characterization of vaginal microflora of healthy, nonpregnant women by 682
chaperonin-60 sequence-based methods
Janet E. Hill, PhD, Swee Han Goh, PhD, Deborah M. Money, MD, FRCSC,
Melissa Doyle, Andra Li, BSc, William L. Crosby, PhD, Matthew Links, BSc,
Amy Leung, Debbie Chan, Sean M. Hemmingsen, PhD
Saskatoon, Saskatchewan, and Vancouver, British Columbia, Canada
The application of cpn60 sequence-based methods to profiling vaginal microflora led
to the identification of significant diversity within traditionally identified taxa and
novel vaginal organisms.
A national probability survey of American Medical Association gynecologists and 693

primary care physicians concerning menopause
Betsy Singh, PhD, Xiao-Dong Liu, PhD, Claudia Der-Martirosian, PhD,
Mary Hardy, MD, Vijay Singh, BS, Neil Shepard, MA, Sonal Gandhi, MD,
Raheleh Khorsan, MA
Whittier and Los Angeles, CA, and Bowling Green, OH
Physicians’ views on various treatment options available to menopausal women
and their interaction with menopausal women during their office visits are
presented.
GENERAL OBSTETRICS AND GYNECOLOGY: OBSTETRICS

A randomized trial of amnioreduction versus septostomy in the treatment of 701
twin-twin transfusion syndrome
Kenneth J. Moise Jr, MD, Karen Dorman, MS, Georgine Lamvu, MD, MPH,
George R. Saade, MD, Nicholas M. Fisk, MD, PhD, Jan E. Dickinson, MD,
R.D. Wilson, MD, Alain Gagnon, MD, Michael A. Belfort, MD,
Richard O. O’Shaughnessy, MD, Usha Chitkara, MD, Sonia S. Hassan, MD,
Anthony Johnson, DO, Anthony Sciscione, DO, Daniel Skupski, MD
Chapel Hill, NC, Galveston, TX, London, United Kingdom, Mumbai, India,
Vancouver, British Columbia, Canada, Salt Lake City, UT, Columbus, OH,
Stanford, CA, Ann Arbor, MI, Newark, DE, and New York, NY
Amnioreduction and septostomy are associated with similar rates of survival of a least
1 twin fetus in cases of severe twin-to-twin transfusion syndrome.
Chorioamnionitis with a fetal inflammatory response is associated with higher 708

neonatal mortality, morbidity, and resource use than chorioamnionitis displaying
a maternal inflammatory response only
Jacqueline Lau, BSc, Fergall Magee, MD, FCRCP(C), Zhenguo Qiu, PhD,
Jill Houbé, MD, FRCPC, MPhil, Peter Von Dadelszen, MBChB, MRCOG,
FRCSC, DPhil, Shoo K. Lee, MBBS, FRCPC, PhD
Vancouver, British Columbia, Canada
Chorioamnionitis with a fetal inflammatory response is associated with higher
neonatal mortality, morbidity, and resource use than when only a maternal
inflammatory response is present.
Early discharge from obstetrics-pediatrics at the Hospital de Valme, with 714

domiciliary follow-up
José Antonio Sainz Bueno, Marı́a Ruiz Romano, Rogelio Garrido Teruel,
Antonio Gutiérrez Benjumea, Ana Fernández Palacı́n, Carmen Almeida
González, Manuel Caballero Manzano
Sevilla, Spain
Early discharge is not implicated in increased maternal or neonatal diseases.
September 2005 7A
Impact of maternal-fetal surgery for myelomeningocele on the progression of 727

ventriculomegaly in utero
Amy Adelberg, MD, Angela Blotzer, BS, Gary Koch, PhD, Rachael Moise,
Nancy Chescheir, MD, Kenneth J. Moise Jr, MD, Honor Wolfe, MD
Chapel Hill, NC
In utero repair of myelomeningocele does not affect the rate of progression of fetal
ventriculomegaly when compared with control fetuses with myelomeningocele
undergoing postnatal repair.
Isolated fetal pyelectasis and chromosomal abnormalities 732

Claudio Coco, MD, Philippe Jeanty, MD, PhD
Nashville, TN, and Rome, Italy
In a review of 12,672 unselected patients, karyotyping of fetuses with isolated
pyelectasis of greater than 4 mm is not justified.
Endometrial microbial colonization and plasma cell endometritis after spontaneous 739
or indicated preterm versus term delivery
William W. Andrews, PhD, MD, Robert L. Goldenberg, MD, John C.
Hauth, MD, Suzanne P. Cliver, BA, Michael Conner, MD, Alice R. Goepfert, MD
Birmingham, AL
Microbial colonization of the endometrium and plasma cell endometritis are similar
3 months after spontaneous or indicated preterm or term births.
Variation in microbiologic profiles among pregnant women with bacterial vaginosis 746
Leonardo Pereira, MD, Jennifer Culhane, PhD, MPH,
Kelly McCollum, MPH, Kathy Agnew, BS, Paul Nyirjesy, MD
Portland, OR, Philadelphia, PA, and Seattle, WA
The presence of Mobiluncus is associated with black race, clue cells on wet mount,
and a positive amine odor after KOH preparation in pregnant women with
bacterial vaginosis.
Use of DNA hybridization to detect vaginal pathogens associated with bacterial 752
vaginosis among asymptomatic pregnant women
Kim A. Boggess, MD, Thomas N. Trevett, MD, Phoebus N.
Madianos, DDS, PhD, Lorna Rabe, BS, Sharon L. Hillier, PhD,
James Beck, PhD, Steven Offenbacher, DDS, PhD
Chapel Hill, NC, and Pittsburgh, PA
Microbial DNA hybridization may be a useful method to study bacterial vaginosis
during pregnancy.
Is zygosity or chorionicity the main determinant of fetal outcome 757

in twin pregnancies?
Stephen G. M. Carroll, MD, Linda Tyfield, PhD, Louise Reeve, PhD,
Helen Porter, MD, Peter Soothill, MD, Phillipa Mm Kyle, MD
Bristol, United Kingdom
Fetal outcome in twin pregnancies is related to chorionicity rather than zygosity.
Sonographic myometrial thickness predicts the latency interval of women with 762
preterm premature rupture of the membranes and oligohydramnios
Catalin S. Buhimschi, MD, Irina A. Buhimschi, MD, Errol R. Norwitz, MD,
PhD, Anna K. Sfakianaki, MD, Benjamin Hamar, MD, Joshua A. Copel, MD,
George R. Saade, MD, Carl P. Weiner, MD
New Haven, CT, Galveston, TX, and Baltimore, MD
Myometrial thickness is directly correlated with latency interval in non-laboring
women with PPROM.
8A September 2005
Use of over-the-counter medications during pregnancy 771
Martha M. Werler, ScD, Allen A. Mitchell, MD,
Sonia Hernandez-Diaz, MD, DrPH, Margaret A. Honein, PhD,
and the National Birth Defects Prevention Study
Boston, MA, and Atlanta, GA
Findings show that over-the-counter medications are used by most pregnant women.
Comparison of the TDx-FLM II and lecithin to sphingomyelin ratio assays in 778
predicting fetal lung maturity
Tamina Winn-McMillan, MD, Brad S. Karon, MD, PhD
Las Vegas, NV
We retrospectively analyzed results of 218 consecutive paired TDx-FLM II and L/S
ratio tests, and compared the utility of the two tests in predicting fetal lung maturity.
Increasing maternal parity predicts neonatal adiposity: 783
Pune Maternal Nutrition Study
Niranjan P. Joshi, MD, DNB, Smita R. Kulkarni, MSc,
Chittaranjan S. Yajnik, MD, FRCP, Charudatta V. Joglekar, MS,
Shobha Rao, PhD, Kurus J. Coyaji, MD, Himangi G. Lubree, MSc,
Sonali S. Rege, MSc, Caroline H. D. Fall, DM, FRCP, FRCPCH
Pune, Maharashtra, India, and Southampton, United Kingdom
Increasing parity predicts higher adiposity in the offspring of thinner
rural Indian mothers.
Misoprostol induces cervical nitric oxide release in pregnant, but not in 790
nonpregnant, women
Mervi Väisänen-Tommiska, MD, Tomi S. Mikkola, MD, PhD,
Olavi Ylikorkala, MD, PhD
Helsinki, Finland
Vaginal misoprostol stimulates cervical NO release in early and late pregnancy,
but not in nonpregnant women.
N-acetyl-transferase phenotype and risk for preeclampsia 797
Petra L. M. Zusterzeel, René H. M. te Morsche, Maarten T. M. Raijmakers,
Eva Maria Roes, Wilbert H. M. Peters, Régine P. M. Steegers-Theunissen,
Eric A. P. Steegers
Nijmegen and Rotterdam, The Netherlands
Women with the fast acetylating N-acetyltransferase phenotype, which may result in
altered N-acetyltransferase detoxification capacity, may have enhanced susceptibility
to preeclampsia.
Profound hypotension and associated electrocardiographic changes during 803
prolonged cord occlusion in the near term fetal sheep
Bert Wibbens, MD, Jenny A. Westgate, MD, PhD, Laura Bennet, PhD, Vincent
Roelfsema, Harmen H. De Haan, MD, PhD, Christian J. Hunter, MD, PhD,
Alistair J. Gunn, MBChB, PhD
Auckland, New Zealand, Groningen and Zwolle, The Netherlands,
and Loma Linda, CA
During profound asphyxia fetal T/QRS ratio and ST segment height reach their
greatest elevation shortly after the start of asphyxia and reduce in amplitude as
hypotension develops.
Hyperemesis gravidarium: Epidemiologic findings from a large cohort 811
Jennifer L. Bailit, MD, MPH
Cleveland, OH
Hyperemesis of pregnancy complicates 473 of 100,000 live births. Fetal and
neonatal death rates are similar to the general population.
September 2005 9A
The involvement of Rho-associated kinases in agonist-dependent contractions of 815

human maternal and placental arteries at term gestation
Mark Wareing, PhD, Maureen O’Hara, Fella Seghier, MSc,
Philip N. Baker, DM, Michael J. Taggart, PhD
Manchester, Great Britain
Pharmacologic inhibition of rho kinase reduces contractility of intact, and
Ca2+-sensitization of force of permeabilized human maternal and placental
arteries from pregnant women at term.
Novel peptides prevent alcohol-induced spatial learning deficits 825

and proinflammatory cytokine release in a mouse model of fetal
alcohol syndrome
Joy Vink, BA, Jonathan Auth, MD, Daniel T. Abebe, BS,
Douglas E. Brenneman, PhD, Catherine Y. Spong, MD
Bethesda, MD
Novel peptides prevent alcohol-induced spatial learning deficits and increases in
proinflammatory cytokines (interleukin-6 and tumor necrosis factor–a) in a mouse
model of fetal alcohol syndrome.
Neutrophils from pregnant women produce thromboxane and tumor necrosis 830
factor-a in response to linoleic acid and oxidative stress
John E. Vaughan, PhD, Scott W. Walsh, PhD
Richmond, VA
Neutrophils from normal pregnant women produced thromboxane and tumor
necrosis factor-a when exposed to conditions present in women with preeclampsia.
Heat shock protein-70 and 4-hydroxy-2-nonenal adducts in human 836

placental villous tissue of normotensive, preeclamptic and intrauterine
growth restricted pregnancies
Michael D. Hnat, DO, Juliana W. Meadows, PhD, Diane E. Brockman, MS,
Brad Pitzer, BS, Fiona Lyall, MD, Leslie Myatt, PhD
Cincinnati, OH, and Glasgow, UK
Presence of total heat shock protein 70 and 4-hydroxy-2-nonenal adducts in villous
tissue revealed no significant difference between samples from uncomplicated and
pathologic pregnancies with preeclampsia and fetal growth restriction.
Cloning and cellular expression of aquaporin 9 in ovine fetal membranes 841

Shengbiao Wang, Jiexiong Chen, Bing Huang, Michael G. Ross
Torrance, San Juan Capistrano, and Orange, CA
Cloning and demonstration of aquaporin 9 expression in ovine amnion and
allantois are consistent with intramembranous water absorption into fetal,
not maternal vasculature.
Limited differentiation to neurons and astroglia from neural stem cells in the 849
cortex and striatum after ischemia/hypoxia in the neonatal rat brain
Tomoaki Ikeda, MD, Masanori Iwai, MD, Takeshi Hayashi, MD,
Isao Nagano, MD, Mikio Shogi, MD, Tsuyomu Ikenoue, MD, Koji Abe, MD
Miyazaki and Okayama, Japan
A rat model of neonatal hypoxic/ischemic encephalopathy showed poor levels
of differentiation to neurons and astroglia from newly proliferated neural
stem cells.
10A September 2005

Intrauterine therapy of goitrous hypothyroidism in a boy with a new compound 857
heterozygous mutation (Y453D and C800R) in the thyroid peroxidase gene.
A long-term follow-up
Kirsten Börgel, MD, Joachim Pohlenz, MD, Wolfgang Holzgreve, MD,
Jurgen H. Bramswig, MD
Münster and Mainz, Germany, and Basel, Switzerland
Spontaneous delivery was possible after intrauterine L-thyroxine therapy improved
thyroid size and thyroid function in a fetus with goitrous hypothyroidism caused by
TPO gene defect.
CLASSIC PAGES
Syndrome of hemolysis, elevated liver enzymes, and low platelet count: 859
A severe consequence of hypertension in pregnancy
Louis Weinstein, MD
Philadelphia, PA
An excerpt from the American Journal of Obstetrics and Gynecology

1982;142:159-67, with a commentary by Lawrence D. Longo, MD,
followed by It has been a great ride: The history of HELLP syndrome,
by Louis Weinstein, MD.
CASE REPORTS
Recombinant human activated protein C treatment of septic shock 864
syndrome in a patient at 18th week of gestation: A case report
László Medve, MD, István Kis Csitári, MD, Zsolt Molnár, MD, PhD,
Ádám László, MD, PhD
Salgótarján, Pécs, and Budapest, Hungary
Septic shock syndrome in a pregnant patient at the 18th week of gestation was
successfully treated with recombinant human-activated protein C.
Clinicopathologic features of six cases of primary cervical lymphoma 866

John K. Chan, MD, Vera Loizzi, MD, Alessandra Magistris, MD,
Mark I. Hunter, MD, Joanne Rutgers, MD, Philip J. DiSaia, MD,
Michael L. Berman, MD
Stanford, Orange, and Long Beach, CA
Cervical lymphoma is a rare disease. Most patients present with localized stage IE
disease and will usually respond to various combinations of surgery, chemotherapy,
and radiotherapy.
Port site ischemic necrosis: An unforeseen complication of laparoscopic surgery 873

R. Oliver, MBBS, MRCOG, A. Coker, MBBS, MRCOG,
Cert Lap Sur (RCOG), Cert BSCCP
Romford, United Kingdom
This case report of necrosis of tissue at the site of a laparoscopy port insertion
highlights an unforeseen and previously unreported complication of
laparoscopic surgery.
Spontaneous epidural hematoma of the spine in pregnancy 875

Ashley S. Case, MD, Patrick S. Ramsey, MD
Birmingham, AL
We report a case of spontaneous epidural hematoma of the spine that complicated a
term pregnancy.
September 2005 11A

Incisional hernia on the 5-mm trocar port site and subsequent wall endometriosis on 878
the same site: A case report
Rodolfo Sirito, MD, Andrea Puppo, MD, Maria Grazia Centurioni, MD,
Claudio Gustavino, MD
Genova, Italy
One year after the removal of an endometrial cyst and, a hernia developed at the
trocar site; 2 years later, a wall endometriosis and another endometrial ovarian
cyst developed.
Prenatal diagnosis of amniotic sheets by magnetic resonance imaging 881

Kiyoshi Kato, MD, Tanri Shiozawa, MD, Takashi Ashida, MD,
Nobuya Unno, MD, Ikuo Konishi, MD
Nagano, Japan
Magnetic resonance imaging is useful for the diagnosis of pregnancies complicated by
amniotic sheets.
Posterior reversible leukoencephalopathy in a case of postpartum eclampsia 885

Maryam Parisaei, MRCOG, Iris Derwig, MBBS, Jeannie Yoon, MRCOG,
Katrina J. Erskine, MRCOG, Paul R. Jarman, PhD
London, United Kingdom
An atypical presentation of eclampsia.
Twin-to-twin transfusion syndrome at 11 weeks of gestation 887

Marieke Sueters, MD, Johanna M. Middeldorp, MD, Dick Oepkes, PhD,
Enrico Lopriore, MD, Frank P. H. A. Vandenbussche, PhD
Leiden, The Netherlands
Presumed vascular connections on the surface of a monochorionic twin placenta
are a possible cause of fetal co-twin death in the first trimester of pregnancy.
Spontaneous closure of the hymen during pregnancy 889

M. A. Onan, MD, A. B. Turp, MD, C. Taskiran, MD, C. Ozogul, MD,
O. Himmetoglu, MD
Besevler, Ankara, Turkey
Case study of spontaneous formation of an imperforate hymen during pregnancy in
the absence of previous regional surgeries.
Small bowel obstruction due to adhesive disease observed after uterine 892
fibroid embolization
Jay Goldberg, MD, MS, Kristyne Boyle, MD, Monica Choi, CRNP, PhD,
Narhari Panchal, MD
Philadelphia, PA, and Buffalo, NY
Small bowel obstruction caused by adhesive disease should be included
in the differential of patients with abdominal pain having a history of
uterine fibroid embolization.
LETTERS TO THE EDITORS

Selective fetocide reverses preeclampsia in discordant twins 894
Francois Audibert, MD, Laurent J. Saloman, MD, René Frydman, MD
Montreal, Quebec, Canada, and Clamart, France
Reply 894
Kent D. Heyborne, MD, Richard P. Porreco, MD
Englewood, CO
12A September 2005

Is circulating extracellular VEGF increased in preeclamptic women? 895
Simone Ferrero, MD
Genoa, Italy
Cytokines, preeclampsia, and uterine denervation? 896

M. J. Quinn, MD
Salford, UK
Reply 896
Gareth C. McKeeman, Joy E. S. Ardill, Carolyn M. Caldwell, Neil McClure
Belfast, Northern Ireland
Accidental fetal lacerations during cesarean delivery: Experience in an Italian 897

level III university hospital
Koji Nishijima, MD, Ken-ichi Shukunami, MD, Fumikazu Kotsuji, MD, PhD
Fukui, Japan
Reply 898
Salvatore Dessole, MD, Giampiero Capobianco, MD, PhD, Erich Cosmi, MD
Sassari, Italy
CORRECTION
Metabolites of progesterone and the pregnane X receptor: A novel pathway 899
regulating uternine contractibility in pregnancy?
Mitchell et al
READER SERVICES
Information for Readers 16a
Professional and educational opportunities 17a
Change of address 667
ELECTRONIC PAGES
(to view these pages, please visit our Web site at www.ajog.org)
2004 Board Certificates e1
September 2005 13A

American Journal of Obstetrics and Gynecology (2005) 193, 599–605
www.ajog.org
EDITORS’ CHOICE
Pediatric presence at cesarean section: Justified or not?

Adrienne Gordon, MBChB, MRCP, FRACP, Elizabeth Jane Mckechnie, MBBS, MRCP,
Heather Jeffery, MBBS, MRCP, FRACP, MPH, PhD
Department of Neonatal Medicine, Royal Prince Alfred Hospital, Sydney, Australia
Received for publication November 28, 2004; revised March 23, 2005; accepted June 24, 2005
KEY WORDS Objectives: This study was undertaken to determine the incidence and type of resuscitation
Cesarean section required for infants delivered by both elective and emergency cesarean section relative to
Anesthesia spontaneous vaginal delivery.
Resuscitation Study design: A hospital-based cohort study from 1990 to 2002. Information was extracted from
Apgar a prospectively collected database on term (R37 weeks) singleton infants delivered by cesarean
section and spontaneous vaginal delivery. Analysis was performed on type of cesarean section,
type of anesthetic, fetal presentation, and evidence of fetal distress. Outcomes assessed were
resuscitation and Apgar scores.
Results: There were 44,938 eligible deliveries. There was no significant difference in need for
resuscitation between infants born by elective cesarean section under regional anesthetic
compared with spontaneous vaginal delivery (c2 = 0.998; df = 1; P = .318). General anesthesia,
fetal distress, and noncephalic presentation increase the need for resuscitation.
Conclusion: An advanced skills practitioner does not need to be present at elective cesarean
sections under regional anesthesia provided there are no additional risk factors.
Ó 2005 Mosby, Inc. All rights reserved.
The rate of cesarean sections is increasing in most independently by 2 reviewers and scored for 4 subscales
developed country hospitals placing extra demands on of a Quality Index for randomized and nonrandomized
currently strained workforces. International guidelines studies.16,17 Despite the evidence, many hospitals still
on neonatal resuscitation and Pediatric and Obstetric require an advanced skills practitioner (most commonly
Colleges state that an appropriately trained practitioner the pediatrician) to be present at all cesarean deliveries.
should be present at all births by cesarean section and an We undertook this study to assess the need for resusci-
advanced skills practitioner present for high risk deliv- tation in a tertiary hospital that has this policy to
eries.1-3 Evidence suggests that term infants born after facilitate implementation of evidence-based practice and
elective cesarean section (CS) under regional anesthetic appropriately allocate staff resources.
have no greater need for resuscitation than those born
vaginally.4-15 A summary of the evidence from the
published randomized, cohort and case-control studies Methods
is attached as Appendix A. All studies were reviewed
Royal Prince Alfred Hospital (RPAH), Sydney, Aus-
tralia, is the major obstetric tertiary referral center for
Reprints not available from the author. Central Sydney Area Health Service (CSAHS). It covers
0002-9378/$ - see front matter Ó 2005 Mosby, Inc. All rights reserved.
doi:10.1016/j.ajog.2005.06.013
600 Gordon, Mckechnie, and Jeffery
Table I Total cohortddemographic, presentation, and resuscitation details

Emergency CS Elective CS
SVD Regional General Regional General
Total no. of deliveries 35,753 3,694 1,004 3,258 1,229
Demographic details*
Mean maternal age (y) 30.2 (5.4) 31.7 (5.2) 30.9 (5.8) 33.5 (5.1) 32.7 (5.6)
Mean birth weight (g) 3,430 (464) 3,420 (522) 3,429 (605) 3,428 (536) 3,427 (514)
Mean gestation (wk) 39.4 (1.2) 39.4 (1.3) 39.3 (1.4) 39.3 (1.1) 39.4 (1.1)
Presentation
Cephalic 35,661 3,230 839 2340 965
Breech 13 392 106 814 207
Other 79 72 59 104 57
Apgar !6 at 1 min 2,636 (7.4%) 528 (14.3%) 380 (37.8%) 122 (3.7%) 264 (21.5%)
Total requiring resuscitation 2,830 (7.9%) 530 (14.3%) 436 (43.3%) 274 (8.4%) 409 (33.3%)
Bag and mask ventilation 2,286 375 268 249 361
ETT and IPPR 284 92 123 13 39
ETT 233 55 32 11 8
CPR 27 8 13 1 1
ETT and IPPR, Eendotracheal tube insertion plus positive pressure ventilation; ETT, endotracheal tube insertion; CPR, cardiopulmonary resuscitation.
* Mean (SD).
an inner city, multicultural population of approxi- and positive pressure ventilation, endotracheal tube, and
mately 500,000 people.18 Within the CSAHS popula- cardiopulmonary resuscitation. The Apgar score, al-
tion, 39.6 % of residents are born overseas and 41.3% though known to be subjective, was recorded consis-
speak a language other than English at home, compared tently for all deliveries throughout the study period. An
with 18.7% in NSW as a whole.18 Of mothers giving Apgar score of less than 6 was used as an indicator of
birth, 41.4% were born in a non-English speaking the need for resuscitation and not as a marker for
country, with Chinese, Arabic, Greek, and Vietnamese subsequent neonatal outcome.
the most commonly spoken languages.18 During the Statistical analysis was performed with SPSS (Statis-
study period, the average number of live births per year tical Package for Social Sciences) version 11.1 (SPSS,
ranged from 3,591 to 4,969. The Obstetric Information Inc, Chicago, Ill). Independent proportions were com-
Service (OIS) computer database was initiated at RPAH pared with the use of the c2 test.
in 1990. Information on all deliveries is entered pro-
spectively by the attending practitioner. Information
collected includes demographic data, fetal and maternal Results
condition before delivery, type of delivery, and fetal
outcomes. Ethics approval was not sought because all There were a total of 44,938 eligible singleton term
data were obtained from this de-identified established deliveries during the study period. This comprised 79%
database used for audit purposes. of all deliveries at RPAH. There was a trend to increas-
The OIS was analyzed for a 13-year period, from ing CS rates throughout the 13 years with an average rate
January 1, 1990, to December 31, 2002. Data were of 21.2% (range 18.3%-27.8%). There were 35,753
extracted on all singleton term deliveries (R37 weeks) spontaneous vaginal deliveries, 4698 emergency CS,
born by vaginal delivery and CS. Multiple, preterm, and and 4,487 elective CS. The demographic, presentation,
instrumental deliveries were excluded because of the and resuscitation details are shown in Table I. Women
mandatory requirement for an advanced skills practi- delivered by elective CS were older than the emergency
tioner. CS was defined as emergency or elective by the CS or vaginal delivery groups. There were no significant
presence or absence of labor. Information was collected differences in mean birth weight or gestational age
on type of anesthetic (regional or general), fetal presen- between the groups.
tation, and the presence of fetal distress. Fetal distress Emergency versus elective CS
was diagnosed if 1 or more of the following was present:
fetal bradycardia, fetal tachycardia, meconium stained As expected significantly more infants required resusci-
liquor, late decelerations, or fetal scalp pH less than 7.2. tation when delivered by emergency CS when compared
The main outcome measures assessed were resuscitation with an elective procedure (966 [20.6%] vs 683 [15.2%]
and Apgar scores. Resuscitation was classified into 4 c2 = 44.4, df = 1, P ! .0001). The number of infants
categories: bag and mask ventilation, endotracheal tube with Apgar scores less than 6 at 1 minute are consistent
Gordon, Mckechnie, and Jeffery 601
Table II SVD and elective CSdeffect of fetal distress on resuscitation

SVD Elective CS - regional P value*
No fetal distress Total no. 21,547 3,119
No. (%) requiring resuscitation 908 (4.2%) 242 (7.8%) !.0001
No. (%) requiring active resuscitation 50 (0.2%) 9 (0.3%) NS
Apgar !6 at 1 min 1,012 (4.7%) 104 (3.3%) .001
Fetal distress Total no. 14,206 139
No. (%) requiring resuscitation 1,922 (13.5%) 32 (23%) .001
No. (%) requiring active resuscitation 494 (3.5%) 16 (11.5%) !.0001
Apgar !6 at 1 min 1,624 (11.4%) 18 (12.9%) NS
NS, Not significant (P O .05).
* Compared with vaginal deliveries.
with this finding (908 [19.3%] vs 386 [8.6%] c2 = 218.1, Cephalic versus noncephalic presentation
df =1, P ! .0001].
Vaginal and elective CS deliveries without fetal distress
were next examined with regard to fetal presentation.
General versus regional anesthesia
The results are shown in Table III. There was no
For both emergency and elective CS significantly more significant difference in need for resuscitation between
infants require resuscitation when delivered under gen- vaginal deliveries and elective CS deliveries under re-
eral anesthetic. For emergency CS, 436 infants (43.3%) gional anesthetic with cephalic presentation in the
required resuscitation after general anesthesia compared absence of fetal distress (906 [4.2%] vs 100 [4.5%]
with 530 (14.3%) under regional anesthesia (c2 = 408.7, c2 = 0.362, df = 1, P = .547). Noncephalic presenta-
df = 1, P ! .0001). For elective CS, 409 infants (33.3%) tion, significantly increased the need for resuscitation in
required resuscitation after general anesthetic compared elective CS deliveries (100 [4.5%] vs 142 [16.0%]
with 274 (8.4%) under regional anesthesia. The Apgar c2 = 117.2, df = 1, P ! .0001).
scores were generally consistent with the need for
resuscitation with the exception of elective CS under Comment
regional anesthesia. An unexpectedly small number of
infants had a low Apgar score compared with the num- This study was performed to assess the need for resus-
ber of infants requiring resuscitation (122 [3.7%] vs 274 citation at CS with particular regard to type of anes-
[8.4%]). The above results clearly demonstrate an in- thetic, elective or emergency procedures, presence of
creased need for resuscitation with emergency CS and fetal distress, and fetal presentation. This was primarily
general anesthetic. After these findings, the remaining to determine whether the need for an advanced skills
results compare elective CS under regional anesthetic practitioner was justified for all CS. Previous studies
with vaginal deliveries. and international guidelines have indicated that the
need for resuscitation in term infants delivered by
Fetal distress versus no fetal distress elective CS under regional anesthesia is not significantly
different to those delivered vaginally.1-15 Many of these
Vaginal deliveries and elective CS were compared with studies, however, have not assessed all relevant factors
regard to fetal distress. The results are shown in Table II. contributing to the need for resuscitation or have not
Fetal distress significantly increased the need for resus- had a control group of vaginal deliveries. This repre-
citation in both groups. In infants without fetal distress, sents the largest hospital-based study on this subject and
those born by elective CS under regional anesthetic were confirms the findings of the only other large population-
more likely to require resuscitation (242 [7.8%] vs 908 based cohort study by Parsons et al4 in Tasmania. In
[4.2%] c2 = 77.0, df = 1, P ! .0001). However, there addition, we were able to assess the type of resuscitation
were significantly more infants with a low Apgar score in required, the presence or absence of fetal distress, and
the vaginal delivery group compared with the elective CS whether the CS was an emergency or elective procedure.
group (1012 [4.7%] vs 104 [3.3%] c2 = 11.7, df = 1, The results clearly demonstrate the increased need for
P = .001]. The need for active resuscitation (intubation resuscitation for infants delivered after emergency CS,
and/or cardiac massage) was not significantly different general anesthesia, and fetal distress.
between the elective CS regional anesthesia and vaginal The crude rates for resuscitation between those
delivery groups in the absence of fetal distress (9 [0.3%] infants delivered vaginally and those delivered by elec-
vs 50 [0.2%] c2 = 0.365, df = 1, P = .546). tive CS under regional anesthetic were not significantly
Table III SVD and elective CSdeffect of presentation on resuscitation

Elective CSdno fetal distress
Presentation SVDdno fetal distress Regional General
Cephalic Total no. deliveries 21,497 2,230 870
No. requiring resuscitation 906 (4.2%) 100 (4.5%) 217 (25%)
Breech Total no. deliveries 9 791 199
No. requiring resuscitation 1 (11%) 127 (16%) 109 (55%)
Other Total no. deliveries 41 98 53
No. requiring resuscitation 1 (2.4%) 15 (15%) 27 (51%)
different consistent with published data. It is imperative, 19%) if presentation was noncephalic within an elective
however, particularly when planning hospital guidelines regional anesthetic CS group but did not analyze the
and offering practice recommendations to compare CS elective CS findings with respect to fetal distress.
relative to an appropriate control group. This control There are several factors that may influence the find-
group consisted of vaginal deliveries with no fetal ings of this study. The hospital-based population within a
distress, as these deliveries would not routinely call an tertiary referral center will be of higher risk than a wider
advanced skills practitioner. When we compared these population-based cohort. This population will, however,
groups, we found a significant difference between the be similar to that of many tertiary centers with rising CS
need for any type of resuscitation. However, when the rates that need to allocate staff and resources appropri-
Apgar score at 1 minute was assessed as an indication of ately. We did not analyze the significance of ethnicity on
the need for resuscitation, significantly more infants in need for resuscitation between the 2 groups; however, the
the control group had low Apgar scores. Also, when previous literature has not indicated that this would make
fetal presentation was taken into account, there was no a significant difference within the context of delivery and
significant difference in need for resuscitation between anesthetic type.4-15 An important factor that may influ-
the 2 groups with respect to cephalic presentation. ence the interpretation of our findings is the classification
Furthermore, the need for active resuscitation, ie, intu- of emergency or elective CS as being the presence or
bation and/or cardiac massage, was not significantly absence of labor. Although the absence of labor is a
different between the groups. We would therefore sug- common classification for elective CS, it does not accu-
gest that the increased number of infants requiring bag rately define true planned cesarean deliveries in low-risk
and mask resuscitation at elective CS under regional patients. It is feasible that some infants delivered ‘‘elec-
anesthetic may be secondary to the very fact that an tively’’ as defined by the absence of labor may be at
advanced skills practitioner was present. Another pos- increased risk of resuscitation secondary to other factors
sibility is that the larger number of infants with non- such as maternal illness, antepartum hemorrhage, or
vertex presentation in the elective CS group may have abnormal placental Doppler flow studies. We were un-
required more bag and mask resuscitation, although the able to assess all the above factors via the OIS database;
initial Apgar scores were not significantly different however, the exclusion of such cases from our elective CS
between the 2 groups making the former suggestion group would favor our interpretation of the results.
more likely. Competency at bag and mask resuscitation
is a recommended standard requirement for all those
Conclusions
practitioners involved in CS. In the absence of fetal
distress, the extra need for an advanced skills practi- The results of this study suggest that an advanced skills
tioner competent at intubation and cardiac massage practitioner need not be present at uncomplicated elec-
would not have been required in the elective CS group tive CS under regional anesthesia provided there are no
any more often than the low-risk vaginal deliveries. other risk factors, namely, fetal distress and noncephalic
Importantly, we assessed fetal presentation within the presentation. Conversely, an advanced skills practi-
regional elective CS group to assess whether noncephalic tioner is required at emergency CS, CS under general
presentation affected the need for resuscitation in the anesthesia, and in the presence of fetal distress and
absence of fetal distress. Surprisingly, we found a 4-fold noncephalic presentation.
increased need for resuscitation in those infants with a
noncephalic presentation. This may relate to a techni- Acknowledgments
cally more difficult delivery if presentation is non-
cephalic, despite no fetal distress before delivery. This We acknowledge the help of Dr Phillip Beeby, Neona-
finding has also been shown by Ng et al7 who demon- tologist, for data retrieval from the Obstetric Informa-
strated a similar increased risk to this study (5% vs tion Systems Database.
References 11. Annibale DJ, Hulsey TC, Wagner CL, Southgate WM. Compar-
ative neonatal morbidity of abdominal and vaginal deliveries after
1. International Guidelines for Neonatal Resuscitation. An excerpt uncomplicated pregnancies. Arch Pediatr Adolesc Med 1995;
from the Guidelines 2000 for cardiopulmonary resuscitation and 149:862-7.
emergency cardiovascular care: International Consensus on Sci- 12. Kolatat T, Somboonnanonda A, Lertakyamanee J, Chinachot T,
ence. Pediatrics 2000;106:pe29. Tritrakarn T, Muangkasem J. Effects of general and regional
2. RACP Health Policy Unit, Pediatric Policy: Pediatrician atten- anaesthesia on the neonate (a prospective randomized trial). J Med
dance at caesarean sections. Available at: http://www.racp.edu.au/ Assoc Thai 1999;82:40-5.
hpu/paed/caesarean/print.htm. Accessed on June 3, 2004. 13. Posen R, Friedlich P, Chan L, Miller D. Relationship between fetal
3. Clinical Guideline 13. Caesarean Section. National Institute for monitoring and resuscitative needs: fetal distress versus routine
Clinical Excellence UK. National Collaborating Centre for Wom- caesarean deliveries. J Perinatol 2000;3:101-4.
en’s and Children’s Health April 2004. Available at: http://www. 14. Burt R, Vaughan L, Daling JR. Evaluating the risks of caesarean
nice.org.uk. Accessed on August 6, 2004. section: low apgar score in repeat C Section and vaginal deliveries.
4. Parsons SJ, Sonneveld S, Nolan T. Is a pediatrician needed at all Am J Public Health 1988;78:1312-4.
caesarean sections? J Pediatr Child Health 1998;34:214-44. 15. Zagorzycki M, Brinkmann CR. The effect of general and epidural
5. Primhak RA, Herber SM, Hincup G, Milner RD. Which deliv- anaesthesia upon neonatal apgar scores in repeat caesarean
eries require pediatricians in attendance? BMJ (Clin Res) section. Obstet Gynecol 1982;155:641-5.
1984;289:16-8. 16. Downs SH, Black N. The feasibility of creating a checklist for the
6. Hogston P. Is a pediatrician required at caesarean section? Eur J assessment of the methodological quality both of randomised and
Obstet Gynecol Reprod Biol 1987;26:91-3. non-randomised studies of healthcare interventions. J Epidemiol
7. Ng PC, Wong MY, Nelson EAS. Pediatrician attendance at Community Health 1998;52:377-84.
caesarean section. Eur J Pediatr 1995;154:672-5. 17. Glasziou P, Irwig L, Bain C, Colditz G. How to review the
8. Press S, Tellechea C, Pregen S. Caesarean delivery of full term evidence: systematic identification and review of the scientific
infants: identification of those at high risk for requiring resuscita- literature. NHMRC, endorsed November 1999. Commonwealth
tion. J Pediatr 1985;106:477-9. of Australia 2000.
9. Jacob J, Pfenniger J. Caesarean deliveries: when is a pediatrician 18. Wen LM, Manson A, Poulos R, Rissel C. A 2001 demographic
necessary? Obstet Gynecol 1997;89:217-20. profile of the Central Sydney Area Health Service. Camperdown,
CSAHS Division of Population Health, 2003. Available at: http://
10. Levine EM, Ghai V, Barton JJ, Strom CM. Pediatrician atten-
www.cs.nsw.gov.au/pophealth/dph/census2001report_complete.pdf.
dance at caesarean delivery: necessary or not? Obstet Gynecol
Accessed on November 12, 2004.
1999;93:338-40.
Appendix A
Review of published literature in descending order of level of evidence by study type* and quality indexy rating
Quality
Author and No. of ratingy Outcome
year Study type Country Study subjects patients (x /27) assessed Results Comments
Kolatat Randomized Thailand Uncomplicated 341 17 Apgar scores NACS Lower Apgar scores Anesthetic changed
1999 trial pregnant women (neurologic and in GA group No in 39 patients and
undergoing CS at adaptive capac- difference in NACS not documented
term randomized ity scores). between groups. whether intention-
to general, epi- to-treat analysis
dural or spinal used.
anesthesia.
Parsons et al Statewide Australia All singleton 64,739 20 Need for intubation Low intubation rate Large population-
1998 population- deliveries R37 Apgar scores. and lower inci- based study of
based cohort wks’ gestation dence of Apgar !4 prospectively
study and/or R2,500 g in repeat CS under collected data.
in state over epidural group
10-y period compared with
1980-1989. vaginal
delivery.
Gordon 2004 Hospital-based Australia All singleton 44,938 20 Need and type of Lower incidence of Large hospital-based
(current cohort study deliveries R37 resuscitation Apgar !6 and no study of prospec-
study) wks’ gestation Apgar scores. significant differ- tively collected
born by CS or ence in need for data. Assessment
vaginal delivery resuscitation in of elective or
over 13-year elective CS under emergency CS,
period regional group fetal distress and
1990-2002. compared with anesthetic on need
vaginal delivery. for resuscitation
compared with
vaginal deliveries.
Annibale et al Cohort study in United States Low-risk population 11,702 19 Need and type of Higher incidence of CS not classified by
1995 2 hospitals at term with CS resuscitation low apgar scores, indication,
and vaginal Apgar scores bag and mask presence of fetal
deliveries. Respiratory ventilation and distress or type of
support, NICU intubation in all anesthesia.
admission. CS.
Posen et al Retrospective United States R37 wks’ gestation 499 19 Need for Increased risk of No control group of
2000 hospital- and delivered by ventilation, resuscitation vaginal
based cohort CS over 3-year circulatory if fetal distress deliveries.
study period. Exclusion support, fluid present.
of maternal and resuscitation or
fetal conditions medications
felt to increase Apgar scores
fetal risk. NICU admission.
Levine et al Hospital-based United States R2,500 g infants 12,923 18 Apgar scores No higher incidence Did not assess need
1999 cohort study delivered by CS of low Apgar score for or type of
or spontaneous in infants delivered resuscitation.
vaginal delivery by CS under
over a 5-year epidural for
period. nonfetal indication
compared with
SVD.
Ng et al Hospital-based UK Term singleton 520 18 Need and type of Increased need for No control group of
1995 cohort study deliveries born resuscitation active resuscita- vaginal deliveries.
by CS over a Apgar scores tion with GA.
1-year period. NICU admission. Increased need for
active resuscita-
tion with
fetal distress and
noncephalic
presentation in
both regional and
general anesthetic
groups.
Appendix A
(Continued)
Quality
Author and No. of ratingy Outcome
year Study type Country Study subjects patients (x /27) assessed Results Comments
Zogorzycki and Retrospective United States Elective repeat CS 195 15 Apgar scores No difference Nonrandomized
Brinkmann hospital- deliveries with between Apgar study of interven-
1982 based cohort infants O2,000 g scores in general tion-based on
study over a 1-year and epidural anes- patient preference.
period. thetic groups. Need for resuscita-
tion not docu-
mented. No control
group of vaginal
deliveries.
Primhak et al Retrospective UK O37 wk’ gestation 1,781 13 Apgar scores Increased For the 1 year
1984 hospital- vaginal, instru- incidence of low period only 1781
based cohort mental and Apgar score if fetal records from the
study operative distress and total of 2,086 term
deliveries over instrumental or deliveries were
1-year operative delivery. available. Need
period. However, if no and type of
fetal distress inci- resuscitation not
dence of low 1-min analyzed.
Apgar score similar
in general and
epidural groups.
Press et al Prospective United States Term infants 377 12 Type of Increased need for No control group of
1984 hospital- delivered by CS resuscitation intubation if vaginal deliveries.
based cohort to private fetal distress and/
study patients over a or fetopelvic
2-year period. disproportion.
Hogston Retrospective UK Consecutive 460 11 Need for In the 55 CS with no Numbers too small
1987 hospital- deliveries R37 resuscitation fetal distress and to compare type
based wks’ gestation cephalic presenta- of anesthetic. No
cohort study tion similar rates period or popula-
of resuscitation as tion documented
vaginal deliveries. for the study
subjects.
Burt et al Statewide case- United States Singleton Cases 21 Delivery type in Infants of repeat CS Type of anesthetic
1988 control study uncomplicated 1,030 cases and con- more likely to have for the repeat CS
births in state Con- trols low Apgar score. not analyzed.
over 3-year trols Fetal distress not
period delivered 998 analyzed.
by repeat CS or
vaginal delivery.
Cases- 5 min
Apgar scores of
0-6. Controls-
5-min Apgar
scores of 7-10.
Jacob and Retrospective United States Singleton term Cases 15 Need and type of Need for active Types of populations
Pfenniger case-control newborns (37-42 834 resuscitation resuscitation from each hospital
1997 study in 2 wks’ gestation) Con- Apgar scores similar for repeat not described.
hospitals over 2-year trols CS and low-risk
period. Cases–CS 834 vaginal deliveries.
deliveries with Increased need for
exclusion of resuscitation if
maternal or fetal general anesthesia.
risk factors
Controls–low-risk
vaginal
deliveries.
* Descending order of level of evidence as per NHMRC How to review the evidence 2000.17
y
Quality Index published by Downs and Black J Epidemiol Community Health 1998;52:377-384.16
www.ajog.org
High-dose methadone maintenance in pregnancy:

Maternal and neonatal outcomes
John J. McCarthy, MD,a,* Martin H. Leamon, MD,b Michael S. Parr, MD,c
Barbara Anania, PsyDa
Bi-Valley Medical Clinic, Sacramento, CA, a Department of Psychiatry and Behavioral Science,
University of California, Davis, CA, b and private practice, Sacramento, CA c
Received for publication September 22, 2004; revised March 11, 2005; accepted March 30, 2005
KEY WORDS Objective: This study assesses the effect of higher doses of methadone during pregnancy on
Methadone maternal and fetal outcomes.
Pregnancy Study design: We retrospectively reviewed clinical data for 81 mothers who received methadone
Neonatal abstinence and their 81 offspring. The cohort was divided into high-dose (R100 mg) and low-dose (!100
syndrome mg) groups.
Drug abuse Results: There were no differences in the rate of medication treatment for neonatal abstinence
symptoms or days of infant hospitalization between the high-dose (mean, 132 mg) and low-dose
(mean, 62 mg) groups. Despite longer histories of opiate abuse, the high-dose group had less illicit
drug use at delivery. The whole cohort, which received an average of 101 mg/d, had an 81% rate
of negative toxicology screens at delivery.
Conclusion: High doses of methadone were not associated with increased risks of neonatal
abstinence symptoms but had a positive effect on maternal drug abuse. Arbitrarily limiting
methadone dose as a way of minimizing the risks of neonatal abstinence symptoms may be
unwarranted.
Methadone maintenance treatment in opiate-addicted nence symptoms can occur in gastrointestinal, metabolic,
pregnant women reduces maternal morbidity and mor- and neurologic domains. Mild symptoms may not re-
tality rates and promotes fetal stability and growth, quire medication treatment, although moderate or severe
compared with mothers who use heroin.1,2 Methadone is symptoms usually require medication-assisted with-
associated with better compliance with obstetric care and drawal and 3-5 weeks of hospital monitoring.
better preparation for parenting responsibilities.3 The There are conflicting studies on whether the higher
baby, however, is at risk for symptoms of neonatal methadone doses that are often needed to eliminate
abstinence syndrome (NAS) that is associated poten- maternal withdrawal symptoms and drug abuse may
tially with withdrawal from methadone at birth. Absti- increase the level of fetal pharmacologic dependence,
potentially leading to more severe NAS.4-9 Berghella
* Reprint requests: John J. McCarthy, MD, 2100 Capitol Ave,
et al,10 in a retrospective review of 100 women maintai-
Sacramento, CA 95816. ned on methadone during pregnancy, found no difference
E-mail: jmccarthy@bivalley.com in severity, duration, or treatment of NAS between
doi:10.1016/j.ajog.2005.03.072
McCarthy et al 607
infants of mothers who received !80 mg/d of meth- The study was approved by the University of
adone and those who received R80 mg/d. In contrast, California, Davis, Institutional Review Board. SPSS
Dashe et al,11 in their retrospective review of 70 women software (version 11.5; SPSS Inc, Chicago, Ill) was used
(mean dose, 20 mg/d) who were withdrawn or tapered for all analyses, and probability values O.05 were
just before delivery, found significant correlations be- selected for statistical significance. Data were analyzed
tween methadone dose and NAS. with independent samples (between-subjects) 2-tailed
Therapeutic response to methadone is dose related. t-tests, chi-squared analyses, and Mann-Whitney tests.
Higher doses are associated with the better treatment
outcomes in nonpregnant patients, and federal guide-
lines recommend increasing methadone doses in preg- Results
nant patients with withdrawal symptoms.12-15 Pregnant
patients have required 50 to 150 mg/d to suppress There were 94 admissions to the pregnancy program
withdrawal symptoms.16 from February 1999 to May 2003. Thirteen subjects
We retrospectively reviewed mothers and infants in a were excluded: 4 women miscarried; 3 women decided to
specialized methadone maintenance pregnancy program terminate pregnancy; 2 women left treatment; 2 patients
with individualized dosing to assess whether higher requested to taper off methadone, and 2 patients had
doses of methadone were associated with adverse neo- unavailable outcome information. Eight women had
natal outcomes. 2 pregnancies during the study; each pregnancy was
considered a separate admission. Data were analyzed for
81 admissions and 81 offspring.
Material and methods The study group was 64% white, 25% Mexican/
Hispanic, 6% African-American, 4% Asian, and 1%
The study’s narcotic treatment program maintains an other. The average maternal age on admission was
active census of approximately 1100 methadone main- 32 G 6.4 years. The average age of first opiate use was
tenance patients in a California metropolitan area 23 G 5.6 years, and the average years of use was 10 G 6.5
with a population of 1.5 million. It is the only special- years. Twenty-five admissions had conceived while on
ized provider of pregnancy services for opiate-addicted methadone maintenance. All others (n = 56 women) were
women in the area. Women in the program are assigned acutely addicted to heroin (n = 49 women), prescrip-
to a specially trained counselor, are all linked with tion opiates (n = 5 women), or opium (n = 2 women).
obstetric care, and give written consent for providers to Seventy-seven percent of the women were cigarette
share information. All of the women participate in and smokers, with 28 % of the smokers using O1 pack/day.
receive a psychiatric assessment, supportive psychother- Polydrug abuse (alcohol, cocaine, methamphetamine, or
apy, 1 hour of individual drug treatment counseling per marijuana) was reported by 38% of the women on admis-
week, and participate in a weekly support group for sion. Seventy-eight percent (n = 1188/1528 specimens) of
both pregnant and early postpartum patients. All pa- all maternal urine toxicology screens before delivery were
tients provide random weekly urine drug screens. As negative for illicit drugs.
part of the clinic’s on-going quality assurance program, The average maternal methadone dose at delivery was
maternal and infant data are collected from program 101 mg/d (range, 14-190 mg/d). Trough serum metha-
entry until 1 month after delivery. done levels were obtained at different gestational ages on
Patients were maintained on divided doses of meth- only 59 of 81 women during pregnancy because of the
adone, given twice or occassionally 3 times a day, difficulty of peripheral venous access in heroin injectors.
because the sustained plasma levels that are achieved The mean trough serum level was 146 ng/mL (median,
with split dosing are associated with fewer withdrawal 115 G 101.5 ng/mL; range, 20-478 ng/mL). Forty-six
symptoms and less illicit drug use during pregnancy.17,18 percent of mothers nursed their babies. The Figure shows
As a quality control measure, methadone trough serum the number of babies who were treated for NAS at each
levels are measured after women reach stable methadone maternal dose range.
dosing and are repeated in patients who require unusu- The infants had a mean gestational age at delivery of
ally high doses. Although there is a therapeutic range for 37.3 weeks and a mean birth weight of 2792 g. No major
methadone trough levels in nonpregnant patients of 150 developmental abnormalities were noted. Eighty-one
to 600 ng/mL,19 there is no attempt to achieve ‘‘target’’ percent of infant toxicology screens (n = 66/81 screens)
serum levels. Methadone doses are clinically adjusted, at the time of delivery were negative for illicit drugs. The
without arbitrary limits, in response to illicit opiate use, 15 positive screens detected opiates (n = 4 women), am-
withdrawal symptoms, or side effects. phetamines (n = 9 women), cocaine (n = 4 women),
NAS was evaluated with an objective scoring sys- diazepam (n = 2 women), marijuana (n = 1 woman),
tem,20 and treatment of the infant was initiated clinically and alcohol (n = 1 woman). Six infants tested posi-
when repeated scores were in the 6 to 8 range. tive for 2 drugs. Thirty-seven babies (46%) required
608 McCarthy et al
Figure Number of infants who were treated for NAS, by maternal dose range.
medication for treatment of NAS symptoms. Infants groups in maternal age, age of onset of drug use, or time
were treated with paregoric (n = 20 infants), phenobar- in treatment, although the high-dose group had signif-
bital (n = 10 infants), both paregoric and phenobarbital icantly longer histories of opiate abuse (mean, 11.6 years
(n = 4 infants), methadone (n = 1 infant), ativan vs 7.8 years in the low-dose group; t = 2.6 G 66.6;
(n = 1 infant), and both paregoric and ativan (n = 1 P ! .05). Chi-squared analyses showed no significant
infant). differences between groups in ethnicity, polydrug use
Because of custody issues, length of stay information history, and smoking history.
was not available on 10 infants. The median length of The Table shows infant outcome data by maternal
stay for the 71 infants on whom data were available was dose group. Chi-squared analyses revealed that the higher
10.0 days (range, 1-105 days). There was no significant dose group had significantly less drug use at delivery: 11%
correlation between maternal dose and length of stay of infant toxicology screens were positive for illicit drugs
(Pearson correlation co-efficient, .066; P = .586). When in the high-dose group versus 27% positive screens in the
divided into NAS-treated (n = 37 infants) and un- low-dose group (P = .05). There were no significant
treated (n = 44 infants) groups, the untreated babies differences in the incidence of treated NAS between
spent a median of 3 days (range, 1-44 days) in the infants of high- and low-dose methadone mothers; 51%
hospital, while babies who were treated for NAS spent a of the high-dose babies and 49% of the low-dose babies
median of 25 days (range, 8-105 days). We observed no required treatment. Mann-Whitney tests for non-normal
cases of post-hospitalization NAS in untreated babies distributions revealed no significant differences in gesta-
during the 1-month postpartum period. tional age (U = 735; N1 = 36; N2 = 45; P = .47), birth
To assess whether higher doses resulted in increased weight (U = 775; N1 = 36; N2 = 45; P = .74), or days
risks of NAS, the cohort was divided into 2 dose groups: of infant hospitalization (U = 600; N1 = 31; N2 = 40;
mothers who were treated with !100 mg of methadone P = .81) between high- and low-dose groups.
(n = 36 mothers) and mothers who were treated with
R100 mg (n = 45 mothers). The cut-off of 100 mg for Comment
the groups was chosen to achieve approximately equal
cohort size. Comparison of maternal dose groups This retrospective records review of methadone-main-
revealed a mean dose in the R100-mg group of 132 tained pregnant women and their offspring found no
mg and 62 mg in the !100-mg group. Independent evidence of an increased incidence of adverse outcomes in
samples t-tests showed no significant differences between babies who were exposed to higher, clinically determined
McCarthy et al 609
Table Infant outcomes by maternal methadone dose group

Maternal methadone dose group
All (n = 81) R100 mg/d (n = 45) !100 mg/d (n = 36) P value
Maternal dose (mg/d)* 101 G 42.4 132 G 24.1 62 G 24.3 .08y
Gestational age (wk)* 37.3 G 3.1 37.1 G 2.9 37.2 G 3.3 .47z
Birth weight (g)* 2792 G 694.6 2795 G 693 2787 G 687 .74z
Infants treated for NAS (n) 37 (46%) 19 (51%) 18 (49%) .32x
Positive infant toxicologic result (n) 15 (18%) 5 (11%) 10 (27%) .05x
* Data are given as mean G SD.
y
Independent samples t-test.
z
Mann-Whitney test.
x
Chi-squared test.
methadone doses. The rate of treatment for NAS and low cord methadone concentrations at delivery were
length of infant hospitalization was similar for both high- associated with more severe NAS. These studies under-
dose (mean, 132 mg/d) and low-dose (mean, 62 mg/d) score the importance of infant variables in the determi-
groups that were studied. Our results extend the findings nation of the risks of NAS. Furthermore, almost one half
of Berghella et al10 to higher average dose ranges. of our mothers nursed their babies. Methadone levels in
Importantly, our high-dose group had significantly milk are small and normally not sufficient to prevent
less detected illicit drug use at delivery, even though this NAS.24 However, Ballard25 found that frequent small
group had significantly longer histories of addiction. feedings in the neonatal period were associated with
Berghella et al10 found a trend toward less drug use reduced symptoms of NAS. Finally, it is speculative, but
at doses of R80 mg. Our study suggests that, as in the more stable serum levels that are achieved by split
nonpregnant populations, higher doses of methadone do doses may have some protective effect against NAS.
lead to less drug use.13 Any theoretic goal of reducing Mothers who receive inadequate doses of methadone or
NAS by using low doses or tapering schedules may single-dose regimes often experience repeated episodes of
well be off-set by the adverse effects of more illicit drug withdrawal, which could possibly sensitize the fetus to
use. For example, Brown et al7 reported on a low-dose the withdrawal state. Further study of the effect of split
methadone-treated pregnant population (41% were doses of methadone on NAS is warranted.
maintained on !50 mg) in which 84% of newborn The role of non-opiate fetal drug exposure (alcohol,
infants tested positive for illicit drugs at the time of cocaine, amphetamine, and benzodiazepine) in effecting the
delivery, which led them to question the efficacy of expression of NAS has not been studied systematically and
methadone treatment. remains a potential confounder in our study, as in others.4-9
The dose range (14-190 mg/d) in our cohort was quite Maternal recovery from illicit drug abuse is critical
wide, possibly reflecting individual differences in meth- for the long-term health and safety of both the mother
adone metabolism.21 Accelerated methadone metabo- and the child. The use of adequate doses of methadone
lism and decreases in methadone bioavailability occur during pregnancy in a specialized program such as the
during pregnancy.22,23 Consistent with these metabolic one described in this study can increase the likelihood of
effects, the mean maternal methadone serum level dur- the mother achieving recovery early in treatment. Con-
ing pregnancy in this study was in the low range for tinued methadone maintenance after delivery may fur-
methadone, despite the high average dose. ther reduce the risks of maternal relapse during the
The overall 46% rate of treated NAS for the infants is critical and often stressful period of parenting a new-
comparable to, or better than, studies in which lower born child.
doses were used. Doberczak et al5 reported a 78% rate
of treated NAS, where the average maternal dose was 50 Acknowledgments
mg/d. The overall rate of treatment in the Dashe et al11
study (median, 20 mg/d) was 46%. We thank Genelle Smith, MSW, for her significant con-
We used treated NAS as an outcome measure. We did tributions to patient care and data collection and Kimberly
not assess other variables that might affect the severity S. Tyda, MA, for her help with data analysis.
of NAS because our study relied on readily available
measures that were used in routine clinical practice.
Doberczak et al5 found that the severity of NAS was References
related to the decline of the neonatal plasma methadone 1. Kaltenbach KA, Berghella V, Finnegan L. Opioid dependence
level from day 1 to day 4 of life. Kushel et al9 confirmed during pregnancy: effects and management. Clin Obstet Gynaecol
this finding and further found that both low maternal and 1998;25:139-51.
610 McCarthy et al
2. Kandell SR, Doberczak TM, Jantunen M, Stein J. The metha- 4, 2005. Available at: http://odp.od.nih.gov/cosensus/cons/108/
done-maintained pregnancy. Clin Perinatol 1999;26:173-83. 108_statement.htm.
3. Wilson GS, Desmond MM, Wait RB. Follow-up of methadone- 15. Kaltenbach K, Silverman N, Wepner RJ. Methadone maintenance
treated and untreated narcotic-dependent women and their infants: during pregnancy. In: Parrino MW, editor. State methadone
health, developmental, and social implications. J Pediatr 1981; treatment guidelines, Center for Substance Abuse Treatment.
98:716-22. Rockville (MD): US Department of Health and Human Services;
4. Doberczak TM, Kandall SR, Wilets I. Neonatal opiate abstinence 1992. p. 85-93.
syndrome in term and preterm infants. J Pediatr 1991;118:933-7. 16. Drozdick J 3rd, Berghella V, Hill M, Kaltenbach K. Methadone
5. Doberczak TM, Kandall SR, Friedmann P. Relationship between trough levels in pregnancy. Am J Obstet Gynecol 2002;187:
maternal methadone dosage, maternal-neonatal methadone levels, 1184-8.
and neonatal withdrawal. J Obstet Gynaecol 1993;81:936-40. 17. Swift RM, Dudley M, DePetrillo P, Camara P, Griffiths W.
6. Madden JD, Chappel JN, Zuspan F, Gumpel J, Mejia A, Davis R. Altered methadone pharmacokinetics in pregnancy: implications
Observation and treatment of neonatal narcotic withdrawal. Am J for dosing. J Subst Abuse 1989;1:453-60.
Obstet Gynecol 1977;127:199-201. 18. DePetrillo PB, Rice J. Methadone dosing and pregnancy: impact
7. Brown HL, Britton KA, Mahaffey D, Brizendine E, Hiett AK, on program compliance. Int J Addiction 1995;30:207-17.
Turnquest MA. Methadone maintenance in pregnancy: a reap- 19. Dole V. Implications of methadone maintenance for theories of
praisal. Am J Obstet Gynecol 1998;179:459-63. narcotic addiction. JAMA 1988;260:3025-9.
8. Blinnick G, Jerez E, Wallach RC. Methadone maintenance, 20. Finnegan LP, Kron RE, Connaughton JF, Emidl JP. Assessment
pregnancy, and progeny. JAMA 1973;225:477-9. and treatment of abstinence in the infant of the drug-dependent
9. Kushel CA, Austerberry L, Cornwell M, Couch R, Rowley RSH. mother. Int J Clin Pharmacol Biopharmacol 1975;12:19-32.
Can methadone concentrations predict the severity of withdrawal 21. Eap CB, Buclin T, Baumann P. Interindividual variability of the
in infants at risk of neonatal abstinence syndrome? Arch Dis Child clinical pharmacokinetics of methadone: implications for the
Fetal Neonatal Ed 2004;89:F390-3. treatment of opioid dependence. Clin Pharmacokinet 2002;
10. Berghella V, Lim PJ, Cherpes J, Chennat J, Kaltenbach K. 41:1153-93.
Maternal methadone dose and neonatal withdrawal. Am J Obstet 22. Pond SM, Kreek MJ, Tong TG, Raghunath J, Benowitz NL.
Gynecol 2003;189:312-7. Altered methadone pharmacokinetics in methadone-maintained
11. Dashe JS, Sheffield JS, Olscher DA, Todd SJ, Jackson GL, Wendel pregnant women. J Pharmacol Exp Ther 1985;233:1-6.
GD. Relationship between maternal methadone dosage and neo- 23. Jarvis MA, Wu-Pong S, Kniseley JS, Schnoll SH. Alterations in
natal withdrawal. Obstet Gynecol 2002;100:1244-9. methadone metabolism during late pregnancy. J Addict Dis
12. Leavitt SG, Shinderman M, Maxwell S, Eap CB, Paris P. When 1999;18:51-61.
enough is not enough: new perspectives on optimal methadone 24. Jansson LM, Velez M, Harrow C. Methadone maintenance and
maintenance dose. J Mt Sinai Hosp NY 2000;67:404-11. lactation: a review of the literature and current management
13. D’Aunno T, Pollack HA. Changes in methadone treatment prac- guidelines. J Hum Lact 2004;20:62-70.
tices. JAMA 2002;288:850-6. 25. Ballard JL. Treatment of neonatal abstinence syndrome with
14. National Institutes of Health Consensus statement 108: Effective breast milk containing methadone. J Perinatol Neonat Nurs
medical treatment of opiate addiction 1997. Last accessed: May 2002;15:76-85.
www.ajog.org
Fetal response to maternal methadone administration

Lauren M. Jansson, MD,a,* Janet DiPietro, PhD,b Andrea Elko, PA-Cc
Department of Pediatrics, Johns Hopkins University School of Medicine,a Department of Population and Family
Health Sciences, John Hopkins Bloomberg School of Public Health,b and Department of Obstetrics
and Gynecology, Johns Hopkins Bayview Medical Center, Baltimore, MD c
Received for publication June 12, 2004; revised November 5, 2004; accepted February 9, 2005
KEY WORDS Objective: The purpose of this study was to investigate the effect of methadone on fetal
Methadone neurobehavioral functions and maternal physiologic indicators.
Fetal development Study design: Forty women attending a substance abuse treatment facility with otherwise
Substance abuse uncomplicated pregnancies were evaluated at peak and trough methadone levels. Fetal measures
Pregnancy included heart rate, variability, periodic accelerations/decelerations, motor activity, and fetal
Fetal heart rate movement-heart rate coupling. Maternal measures included maternal heart period, variability,
Fetal movement electrodermal skin conductance, respiration, and respiratory sinus arrhythmia (RSA). Repeated
measure analysis of variance was used to evaluate within-subject changes.
Results: At peak methadone, fetal heart rate was slower, less variable, and displayed fewer
accelerations. Fetuses displayed less motor activity, and the integration between heart rate and
motor activity was attenuated. Maternal heart rate and skin conductance were unchanged, but
methadone administration was associated with lower respiratory rate and RSA, an indicator of
parasympathetic tone.
Conclusion: Maternal methadone administration has significant effects on fetal behavioral
functions that are independent of maternal effects.
The problem of chemical dependency during preg- because there are well known serious health conse-
nancy has plagued the health care system since the quences for both mother and child. In 2000, heroin was
1970s. During the 1990s, reports of illicit drug use by the leading illicit drug among all treatment admissions in
pregnant women hovered between 9% and 10%.1,2 the US, and 40% of participants received methadone;
Opiate dependency during pregnancy presents a partic- approximately one third of these participants were
ularly vexing problem for the health care community women.3 Abrupt cessation of opiates in gravid women
is not advised because of adverse fetal consequences.2,4
Methadone is currently the only pharmacologic
This work is supported by NIH grants K08 DA00495 awarded to treatment offered for this group in the US, but its use
the first author, and R01 HD27592 awarded to the second author. is somewhat controversial. Methadone administration
* Reprint requests: Lauren M. Jansson, MD, Johns Hopkins has well-documented positive effects for the pregnant
University, Johns Hopkins Bayview Medical Center, The Center for
Addiction and Pregnancy, 4940 Eastern Avenue, D5, Baltimore, MD
opiate-dependent woman. Compared with heroin use,
21224. methadone maintenance during pregnancy is associated
E-mail: ljansson@jhmi.edu with more prenatal care, improved fetal growth, reduced
doi:10.1016/j.ajog.2005.02.075
612 Jansson, DiPietro, and Elko
fetal mortality, and increased chances of discharging the cations, such as hypertension and gestational diabetes.
infant into the care of his/her parents upon hospital The research was approved by the governing institu-
discharge.5 Conversely, methadone use during preg- tional review board, and informed consent was obtained
nancy is associated with a significant risk of neonatal from all participants.
abstinence syndrome (NAS). Approximately 75% of
opiate-exposed infants develop NAS after birth,6 and Procedures
methadone-exposed infants display more profound Data collection involved 2 60-minute fetal monitoring
NAS symptoms than heroin-exposed infants.5,7,8 When sessions on one day during the 36th week of gestation.
comparing methadone with heroin-exposed infant out- Estimation of gestational age was based on a sonogram
comes, methadone-exposed infants are heavier at birth,9 performed between 16 and 20 weeks’ gestation in all
but weigh less than non–drug-exposed infants.10 cases, and was confirmed by postnatal assessment. The
Methadone crosses the placenta readily, reaching initial test session took place at trough maternal meth-
concentrations in the umbilical cord of one fourth the adone level, typically occurring at 1 hour before oral
maternal serum.11 A number of studies have found less dose, near 09:30; the second, at peak methadone, 2½
reactivity in nonstress testing in methadone-exposed hours after oral methadone, near 13:30. Peak plasma
fetuses after maternal methadone administration when methadone levels in individuals chronically treated with
compared with controls.12-15 Although consistent in oral methadone occur between 2 to 4 hours after an oral
their findings, these reports are variously limited by daily dose.17 Methadone was administered once daily,
reliance on visual inspection of fetal tracings, subther- at the conclusion of the first monitoring session
apeutic maternal methadone dosing during pregnancy, (M dose = 78.4 mg, sd 17.8, range 40 to 115 mg). Metha-
relatively small sample sizes, and potential confounding done administration was confirmed by pupillometer
because of recent, or failure to evaluate for, illicit drug photograph, which revealed pupillary constriction at
or alcohol use. peak (vs trough) methadone in all cases.
The present study was designed to ascertain the Women were asked to eat a meal and refrain from
maternal and fetal effects of therapeutic methadone caffeine and/or cigarettes 2 hours before each session in
administration by simultaneously monitoring fetal an effort to control for maternal glycemic status and the
neurobehavioral indicators and maternal physiologic effects of caffeine and nicotine on the fetus. Recent
responses at peak and trough methadone levels. Assess- cigarette smoking was evaluated at both sessions by
ment of subtle indicators of fetal functioning provides breathalyzer testing for carbon monoxide (CO) using
a window into the developing nervous system and the Micro Smokerlyser (Bedfont Scientific Limited,
quantification of fetal response to episodic methadone Kent, UK). Carbon monoxide levels in parts per million
exposure. were recorded; values were not adjusted for background
levels. Recent alcohol use was similarly evaluated using
Material and methods a Digitox-D.O.T. (Sound Off, Inc, Hudsonville, MI)
screening breath tester, which measures blood alcohol
Study population content to 0.02%, or the equivalent of 1 drink. Urine
was obtained for toxicology at both sessions. Partic-
Forty-two methadone-maintained women were selected ipants’ perceived symptomatology indicative of opiate
from a population of women engaged in substance withdrawal was surveyed at each test session using the
abuse treatment at an urban, comprehensive substance Subjective Opiate Withdrawal Scale (SOWS).18
abuse treatment program for pregnant and postpartum
substance abusing women and their children. The pro- Primary outcome measures
gram has been described elsewhere.16 Subjects were
selected based on recommendation from their coun- Fetal neurobehavioral assessment
selors for compliance to program standards and staff Fetal data were collected using a Toitu (MT325, Tokyo,
confidence in their abstinence from illicit substances and Japan) fetal actocardiograph. This monitor detects fetal
alcohol during treatment, but were not different demo- movement (FM) and fetal heart rate (FHR) using a
graphically (race, parity age), or by substance use single-wide array transabdominal Doppler transducer,
history from program participant means for women and processes this signal through a series of filters. The
meeting criteria for methadone maintenance. At the actograph detects fetal movements by preserving the
start of treatment, all women met DSM IV-R criteria for remaining signal after band-passing frequency compo-
opiate dependence, and federal guidelines for metha- nents of the Doppler signal that are associated with
done maintenance. To minimize confounds based on FHR and maternal somatic activity. Reliability studies
pregnancy risk factors, inclusion was limited to women comparing actograph-based vs ultrasound-visualized
with uncomplicated singleton pregnancies who were fetal movements have found the performance of this
HIV negative and free of significant pregnancy compli- monitor to be highly accurate in detecting both fetal
Jansson, DiPietro, and Elko 613
motor activity and quiescence.19 Fetal data were col- deviation of successive heart periods). Both cardiac
lected from the output port of the monitor and digitized measures are influenced by both neural and non-neural
at 1000 Hz through an internal A/D board using factors, so spectral analysis was used to extract the
streaming software. Data were analyzed offline using variation in heart period within the high frequency
customized software (James Long Company, Caroga oscillations coincident with spontaneous breathing
Lake, NY). Digitized heart rate data underwent error (James Long Company), yielding a measure of respira-
rejection procedures based on moving averages of tory sinus arrhythmia (RSA; msec2). RSA is mediated
acceptable values as needed. Fetal variables included 4 through the sinoatrial node, and thus, is frequently used
cardiac measures: fetal heart rate and variability (root as an indicator of neural activation of parasympathetic
mean squared) computed for each 1-minute epoch processes and vagal tone.20 Maternal variables were
averaged over time, and accelerations and decelerations. averaged during the recording.
Accelerations were identified when FHR values attained
Data analysis
10 bpm above baseline for R15 seconds; decelerations
Univariate repeated measure analysis of variance was
defined as %15 bpm below baseline for R15 seconds.
Fetal movement measures were based on the acto- used to evaluate changes in fetal and maternal values
from trough to peak maternal methadone levels. Change
graph signal, which ranges from 0 to 100 in arbitrary
scores from peak to trough in each measure were
units. A movement bout was considered to begin when
computed and Pearson correlations were used to deter-
the first spike of the actograph attained an amplitude of
mine whether methadone dosage was related to the
15 units, and ended when there was a cessation of 15
magnitude of fetal and maternal baseline values and
unit signals for at least 10 seconds. The number of
responses.
movement bouts was counted, and the duration of each
movement was measured. Total motor activity was
computed as the number of movement bouts multiplied
by the mean movement duration (seconds) divided by Results
3600 (number of seconds per 60-minute recording),
yielding the total time the fetus spent moving. FM- Participants tended to be multigravid (75%), near 30
FHR coupling was calculated as the percentage of fetal years of age (mean = 29.4 years, sd = 5.7), varied in
movements associated with excursions in FHR R5 bpm race and ethnicity (40% African American, 57.5%
over baseline within 5 seconds before the start of a Caucasian, 2.5% Hispanic), and unmarried (92.5%).
movement or within 15 seconds after the start of a Prescribed substances included psychotropic medica-
movement. tions, primarily fluoxetine and sertraline (40%), asthma
medications (5%), and antibiotics (12.5%). Most (85%)
Maternal physiologic assessment smoked cigarettes regularly. Two women were retro-
Maternal physiologic signals were amplified using a spectively removed from study inclusion as a result of
multichannel, electrically isolated bioamplifier. Electro- positive urine toxicology at the time of testing, leaving a
cardiogram (ECG) was recorded from 3 carbon fiber sample of 40 women.
disposable electrodes in triangulated placement (right Half (52.5%) of the fetuses were male, and 92.5%
mid subclavicle, left mid axillary thorax, and upper left (37) were born at term. Of the remainder, 2 were born in
thigh for ground lead). Electrodermal activity (ie, skin the 36th week of gestation, and there was 1 fetal death at
conductance) was monitored from 2 silver-silver chlo- term from unspecified causes. This fetus was not out-
ride electrodes with a gelled skin contact area placed on standing in terms of location within the distribution of
the distal phalanxes of the first and index fingers of the any fetal or maternal parameter measured in the study,
nondominant hand affixed with adhesive collars to limit so was retained in the final sample. Birth weights of the
gel contact to a 1-cm diameter circle and Velcro. surviving neonates were appropriate for gestational age
Maternal data were time synchronized and analyzed (M birth weight = 3078.50 g, sd 415.50 g); there was no
in conjunction with fetal data. ECG data underwent evidence of growth retardation for any infant. All 5-
R-wave detection, manual editing for artifact, and minute Apgar scores were O7. Infants were without
interbeat interval computation. Skin conductance was significant neonatal morbidity aside from NAS, which
measured by administering a constant 0.5-V root mean occurred with treatable significance in 51.4%.
square 30 Hz AC excitation signal, and detecting the Fetal parameters
current flow. Skin conductance was scaled from 0 to 25
microsiemens and detrended to remove the mean, There were highly significant changes in all aspects of
thereby amplifying the signal to noise ratio (ÿ2.5 to fetal neurobehavioral functioning from trough to peak
C2.5 microsiemens). Maternal ECG values were quan- methadone; data are presented in Table I. At peak
tified as heart period (ie, interval between R-waves in methadone, fetuses displayed significantly slower fetal
milliseconds), and heart period variability (standard heart rate, reduced variability, and fewer heart rate
decline, women at trough levels were experiencing

Table I Fetal neurobehavioral measures at trough and peak
methadone (n = 40) very little to no withdrawal symptomatology.
Trough Peak
Methadone dosage effects
Mean (SD) Mean (SD) F
Heart rate (bpm) 136.47 (7.56) 128.29 (5.22) 52.09* Maternal methadone doses ranged from 40 to 115 mg
Heart rate 5.93 (1.26) 3.72 (1.05) 100.33* daily. Correlation coefficients were computed between
variability dosage and change scores from trough to peak for each
Number of 3.63 (3.03) 0.47 (0.75) 48.21* fetal variable. To control for the Law of Initial Values,
accelerations trough levels were partialled from each correlation.
Movement bouts 66.80 (16.45) 63.55 (18.94) 0.79 Methadone dosage was significantly associated with
Duration of 26.88 (15.38) 13.69 (9.05) 26.17*
greater changes from trough to peak only for fetal heart
movements
Total motor 1627.81 (670.33) 880.05 (588.85) 36.66*
rate variability (r (38) = ÿ.34, P ! .05), indicating that
activity those receiving more methadone showed greater changes
FM-FHR coupling 22% (8%) 16% (9%) 18.01* in variability. Correlations for methadone dosage and
(%) mean fetal measures at peak (not change scores) indi-
* P ! .0001. cated the following: higher dosages were associated with
lower heart rate (r (38) = ÿ.32, P ! .05), greater
variability (r (38) = .37, P ! .05), and a trend to-
accelerations. Decelerations were too infrequent to be wards more fetal heart rate accelerations (r (38) = .29,
analyzed: 3 were observed at trough, and 1 at peak. P ! .10). No significant associations were detected with
Although the number of movement bouts was un- fetal motor variables or FM-FHR coupling. Maternal
changed, the duration of each movement and the resul- measures were unrelated to dosage.
tant total amount of fetal motor activity were both
significantly reduced, approximately by half, at peak Effects of other substances
maternal methadone. In addition, the degree of coupling No appreciable levels of alcohol use were detected via
between fetal movements and fetal heart rate (FM-FHR alcohol screening by breathalyzer testing for any subject
coupling) was significantly lower at peak. at either visit. Women were asked to refrain from
Although these quantitative differences are statisti- smoking, and carbon monoxide testing indicated no
cally robust, mean values do not adequately capture the substantial recent cigarette use. Trough values
degree to which methadone exerted an effect on fetal (M = 6.80 ppm, sd 4.8) did not differ from peak values
functioning. The Figure presents representative data at (M = 7.5, sd 5.3), F (1,39 = 1.27, P O .10).
peak and trough from a single fetus. To better define the
totality of the effect of methadone, a clinician expert in
fetal actocardiograph tracings who was blinded to Comment
condition was asked to judge under which condition
(trough or peak) the fetus looked ‘‘less well.’’ The rater Administration of methadone to pregnant women is
identified the peak methadone tracing as the one in associated with profound effects on fetal neurobehav-
which the fetus looked less well in all 40 cases. ioral functioning. These changes were evidenced despite
minor effects on indicators of physiologic functioning in
Maternal measures pregnant women, suggesting that methadone may be
mediating these fetal effects directly. The effects were
Despite the significant differences in fetal measures, pervasive and evidenced in fetal heart rate, motor
methadone administration generated few effects on activity, and their interrelation. Fetal neurobehavior
maternal physiologic parameters (Table II). Maternal serves as a window to the developing nervous system,
heart period and variability were not significantly dif- and its disruption implies threat to the neural develop-
ferent at trough vs peak methadone, nor was skin ment of the fetus.21
conductance. Maternal RSA was significantly lower at In opiate-exposed newborns, evidence of neonatal
peak methadone, indicating lower parasympathetic behavioral disorganization does not stipulate that neu-
tone. Respiratory period was slightly higher at peak rologic harm has been done during pregnancy, but
methadone. Maternal reports of withdrawal symptom- rather may reflect more transient effects of methadone
atology as determined by the Subjective Opiate With- administration or withdrawal. Support for this view is
drawal Scale declined from trough (M = 5.0, sd 7.4) to provided by the lack of association between the severity
peak (M = 2.5, sd 4.1), F (1, 38) = 8.84, P ! .01. of neonatal abstinence and later developmental out-
However, SOWS values below middle-teen scores reflect come.22 The goal of this study was to evaluate the
minimal withdrawal discomfort; thus, despite this transient effects of maternal methadone administration
Figure Representative fetal heart rate and fetal movement tracing at trough (A) and peak (B) maternal methadone level.
obscured by the nature of the design, subsequent

Table II Maternal physiologic measures at trough and peak
methadone (n = 40) investigations would benefit from inclusion of a non-
opiate-dependent control group evaluated at the same
Trough Peak
times of day.
Mean (SD) Mean (SD) F
Although our results indicate that methadone main-
Heart period 884.41 (152.10) 881.07 (155.48) 0.16 tenance is clearly not innocuous to the fetus, the broader
Heart period 32.68 (30.57) 32.17 (30.50) 0.08 issue, the use of methadone as the treatment of choice
variability
for pregnant opiate addicts, is a complex one. This
Respiratory sinus 61.32 (29.09) 51.30 (30.48) 5.28*
arrhythmia
strategy has persisted despite evidence that methadone
Skin conductance 6.19 (4.63) 5.76 (3.72) 1.33 generates effects on the neonate that are as deleterious, if
Respiratory period 4.43 (1.24) 4.69 (1.43) 4.82* not more so, than heroin, on neonatal abstinence
* P ! .05.
symptoms. The benefits of methadone use during preg-
nancy have been well documented for the mother, but
not conclusively so for the infant. This study evaluated
single-dose therapy, the most common approach to
on fetal behavior. The long-term sequelae of daily, methadone administration. Findings suggest that while
repeated depression of motor activity and heart rate this may be the most convenient management strategy
are unknown, but potentially present a distinct mecha- for the mother, other approaches may be less disruptive
nism through which methadone may exert an effect on to the fetus. Split-dosing administration has been shown
development beyond neurotoxic effects of the substance to decrease fetal effects in a small study of 7 methadone
itself. Without a control group, we are unable to maintained women at 26 to 37 weeks’ gestation.32
evaluate the neurobehavioral toxicity of methadone Fetuses of women on a split dose displayed less decrease
exposure. However, 36-week FM-FHR coupling trough in ultrasound-observed fetal body movements and
values from fetuses in this sample, when compared breathing rate in contrast to those receiving a single
to previously published values23 from a comparable dose. More intensive investigation is needed to establish
population of socioeconomically disadvantaged, but the dosing strategy that provides least disruption to the
nondrug-using women from the same city, differ by fetus, while offering the opiate-dependent woman the
more than a standard deviation (22% vs 38%). This multiple benefits of methadone therapy. In addition,
suggests persistent effects of methadone exposure even evaluation of the degree to which new treatments on the
when methadone levels are at their nadir. horizon for use during pregnancy, including buprenor-
A limitation of this study was that we were unable to phine, are more or less disruptive to fetal neurobehav-
counterbalance dosage (ie, have half of the participants ioral development than methadone should be
undergo peak methadone at 10:30, and the other half at conducted concurrently with efficacy studies for their
13:30) to control for potential circadian influences and use in pregnant women.
the effect of meals. This was due to the strict dosing and
programmatic guidelines for methadone outpatient
treatment. To achieve appropriate peak and trough
levels at comparable times would require providing
Acknowledgments
women with take home doses for a prolonged period The authors thank the subjects, Kathleen Costigan, RN,
in order to achieve pharmacologic stabilization, which is MPH, Vickie Walters, LCSW-C, Tammy Roberts,
prohibited by federal regulations. Moreover, we would Martha Velez, MD, and Hendrée Jones, PhD, for their
be unable to ascertain compliance in timing of dose help in conducting this research. This research was funded
administration. However, based on existing literature by K08-DA00495, NIDA, provided to the first author.
and our experience in fetal assessment, there is little
reason to expect that either the 4-hour span between
trough and peak or the transition from breakfast to
lunch would generate the striking effects observed here. References
Although circadian rhythms in fetal behaviors from
1. Gfroerer J. National household survey on drug abuse: population
morning to night have been documented, the prepon- estimates 1994, DHHS Publication No. 95-3063. Rockville, Mary-
derance of literature finds minimal, if any, alterations in land: Office of Applied Studies; 1995.
fetal behavior and heart rate during daylight hours.24-28 2. King J. Substance abuse in pregnancy: a bigger problem than you
Second, despite common clinical perception to the think. Post Med 1997;102:135-50.
contrary, there is no empirical support for the notion 3. Substance Abuse and Mental Health Services Administration.
Planned methadone treatment for heroin admissions. Drug and
that maternal food intake or blood glucose levels affect Alcohol Services Information System. The DASIS Report, 2003.
fetal behavior or heart rate.28-31 Thus, while we are Retrieved from: http://www.samsha.gov/oas/2k3/methadoneHtx/
confident that interpretation of these results is not methadoneHtx.htm. Accessed Jan 21, 2004.
4. Zuspan FP, Gumpel JA, Mejia-Zelaya A, Madden J, Davis R. 19. DiPietro J, Costigan K, Pressman E. Fetal movement detection:
Fetal stress from methadone withdrawal. Am J Obstet Gynecol comparison of the Toitu actograph with ultrasound from 20 weeks
1975;122:43-6. gestation. J Matern Fetal Med 1999;8:237-42.
5. Kandall SR, Albin RS, Gartner LM, Lee KS, Eidelman A, 20. Berntson G, Cacioppo J, Quigley K. Respiratory sinus arrhythmia:
Lowinson J, et al. The narcotic dependent mother: fetal and autonomic origins, physiological mechanisms, and psychophysio-
neonatal consequences. Early Hum Develop 1977;1:159-69. logical implications. Psychophys 1993;30:183-96.
6. Kandall SR. Treatment options for drug-exposed infants. NIDA 21. Nijhuis J. Fetal behavior. Neurobiol Aging 2003;S41-6.
Res Monogr 1995;149:78-99. 22. Kaltenbach K, Finnegan L. Neonatal abstinence syndrome, phar-
7. Rajegowda BK, Glass L, Evans H, Maso G, Swartz DP, Leblanc macotherapy and developmental outcome. Neurobehav Toxicol
W, et al. Methadone withdrawal in newborn infants. J Peds Teratol 1986;8:353-5.
1972;81:532-4. 23. DiPietro JA, Costigan KA, Shupe AK, Pressman EK. Fetal
8. Wilson G, Desmond M, Wait R. Follow-up of methadone treated neurobehavioral development: associations with socioeconomic
and untreated narcotic dependent women and their infants: class and fetal sex. Devel Psychobiol 1998;33:79-91.
health, development, and social implications. J Peds 1981;98: 24. Nijhuis IJ, ten Hof J, Mulder EJ, Nijhuis JG, Narayan H, Taylor
716-22. DJ, et al. Fetal heart rate in relation to its variation in normal and
9. Connaughton JF, Reeser D, Schutt J, Finnegan LP. Perinatal growth retarded fetuses. Eur J Obstet Gynecol Reprod Biol
addiction: outcome and management. Am J Obstet Gynecol 2000;89:27-33.
1977;729:679-86. 25. ten Hof J, Nijhuis IJ, Mulder EJ, Nijhuis JG, Narayan H, Taylor
10. Kaltenbach K, Finnegan L. Perinatal and developmental outcome DJ, et al. Longitudinal study of fetal body movements: nomo-
of infants exposed to methadone in-utero. Neurotoxicol Teratol grams, intrafetal consistency and relationships with episodes of
1987;9:311-3. heart rate patterns A and B. Pediatr Res 2002;52:568-75.
11. Harper RG, Solish G, Feingold E, Gersten-Woolf NB, Sokal MM. 26. Koenen SV, Franx A, Faber JAJ, Mulder EJH. Blunting of the
Maternal ingested methadone, body fluid methadone, and the maternal diurnal blood pressure profile but unaltered fetal diurnal
neonatal withdrawal syndrome. Am J Obstet Gynecol 1977;129: rhythms. Prenat Neonat Med 2000;5:48-55.
417-24. 27. Arduini D, Giannini F, Rizzo G, Valensise H, Cesa F. Circadian
12. Archie CL, Lee MI, Sokol RJ, Norman G. The effects of rhythms of fetal activity in early pregnancy. In J Gynecol Obstet
methadone treatment on the reactivity of the nonstress test. Obstet 1996;8:43-50.
Gynecol 1989;74:254-5. 28. Patrick J, Campbell K, Carmichael L, Natale R, Richardson B.
13. Anyaegbunam A, Tran T, Jadali D, Randolph G, Mikhail MS. Patterns of gross fetal body movements over 24-hour observation
Assessment of fetal well-being in methadone-maintained pregnan- intervals during the last 10 weeks of pregnancy. Am J Obstet
cies: abnormal non-stress tests. Gyn Ob Inv 1997;43:25-8. Gynecol 1982;142:363-71.
14. Levine A, Rebarber A. Methadone maintenance treatment and the 29. Serra-Serra V, Camara R, Sarrion P, Jareno M, Cervera J, Bellver
nonstress test. J Perinatol 1995;15:229-31. J, et al. Effects of prandial glycemic changes on objective fetal
15. Cejtin H, Mills A, Swift E. Effect of methadone on the biophysical heart rate parameters. Acta Obstet Gynecol Scand 2000;79:953-7.
profile. J Repro Med 1996;41:819-22. 30. Lunshof S, Boer K, Howeler S, Hulsman C, Ruijter J, Wolf H,
16. Jansson LM, Svikis D, Lee J, Paluzzi P, Rutigliano P, Hackerman et al. Maternal food intake has no effect on fetal heart rate
F. Pregnancy and addiction: a comprehensive care model. J Subst parameters. Prenat Neonat Med 1999;4:457-60.
Abuse Treat 1996;13:321-9. 31. Bocking A, Adamson L, Carmichael L, Patrick J, Probert C. Effect
17. Kreek M. Plasma and urine levels of methadone. NYS J Med of intravenous glucose injection on human maternal and fetal heart
1973;23:2773-7. rate at term. Am J Obstet Gynecol 1984;148:414-20.
18. Handelsman L, Cochrane KJ, Aronson MJ, Ness RA, Rubinstein 32. Wittman B, Segal S. A comparison of the effects of the single- and
KJ, Kanof PD. Two new rating scales for opiate withdrawal. Am J split-dose methadone administration on the fetus: ultrasound
Drug Alcohol Abuse 1987;13:293-308. evaluation. Int J Add 1991;26:213-8.
American Journal of Obstetrics and Gynecology (2005) 193, 618
www.ajog.org
RESEARCH METHODS: STATE OF THE SCIENCE
A new series of articles on research methodology

in clinical medicine
George A. Macones, MD, MSCE,* Series Editor
Department of Obstetrics and Gynecology, Department of Biostatistics and Epidemiology, University of Pennsylvania,
School of Medicine, Philadelphia, PA
Received for publication December 11, 2004; revised March 7, 2005; accepted May 22, 2005
It is a pleasure to introduce a new series that will be years, as concerns regarding the protection of human
appearing in the American Journal of Obstetrics and subjects in research have increased. In fact, any National
Gynecology - Research Methods: State of the Science. Institutes of Health (NIH)-funded clinical must have
With the recent explosion of research methods - clinical, provisions made for monitoring data on an interim basis.
basic, and translational - physicians and investigators are One question unanswered by the manuscript by Thom
faced with an increasingly complex body of research and Klebanoff is whether there is need for data safety
literature in obstetrics and gynecology. The goal of this and monitoring of smaller clinical trials, such as those
series is to provide in depth reviews and discussions of commonly performed at a single site. For the Maternal-
emerging themes and controversies in research and will Fetal Medicine Units (MFMU) Network, which is
include manuscripts focused on both basic and clinical supported by NIH, funds are available for this aspect
research topics. Manuscripts from this series will be of clinical research. In addition, there are certainly
appearing intermittently in the journal over the next ‘‘economies of scale,’’ such that the same DSMC can
several years and will be solicited and written by experts in assess all of the ongoing clinical trials within the MFMU
these specific content areas of research. It is our hope and Network. But the situation is very different for a single-
belief that this series will aid both clinicians and re- site, investigator initiated study. In this case, the devel-
searchers in their understanding of the literature related to opment of a plan for interim analysis and composing a
obstetrics and gynecology. DSMC can be daunting. Still, to ensure adequate pro-
In this edition of the Journal, Thom and Klebanoff tection of human subjects in our research studies, we do
eloquently discuss 2 critical issues related to the design and have an obligation to assess efficacy and safety of our
execution of randomized clinical trials. The first is the role clinical trials. On a smaller scale, this can be accom-
of Data Safety and Monitoring Committees (DSMC) in plished by consultation with a statistician before com-
the context of clinical trials and how decisions are made to mencing with the study and construction of a DSMC
stop studies. This topic has gained in importance in recent using clinicians and researchers at local institutions.
* Reprint requests: George A. Macones, MD, MSCE, Division of

Maternal-Fetal Medicine, 2000 Courtyard Building, 3400 Spruce
Street, Philadelphia, PA, 19104.
E-mail: gmacones@mail.obgyn.upenn.edu
doi:10.1016/j.ajog.2005.04.059
www.ajog.org
Issues in clinical trial design: Stopping a trial early

and the large and simple trial
Elizabeth A. Thom, PhD,a Mark A. Klebanoff, MD, MPHb
The Biostatistics Center, George Washington University,a Rockville, MD; National Institute of Child Health and
Human Development, National Institutes of Health, Department of Health and Human Services,b Bethesda, MD
Received for publication December 11, 2004; accepted April 22, 2005
KEY WORDS During the conduct of a clinical trial, a primary function of the Data Safety and Monitoring
Research methods Committee is to select the trial conduct and the accumulating data to determine whether the trial
Clinical trial should continue or be discontinued earlier than planned. Reasons for early discontinuation of a
Biostatistics trial include: evidence of benefit, evidence of harm, and evidence of futility. More than 1 of these
Study monitoring elements will often be present. These principles will be illustrated with examples from National
Institute of Child Health and Human Development–Maternal-Fetal Medicine Units clinical
trials. The ‘‘large and simple clinical trial’’ is a study design rarely undertaken in the United States
but commonly used elsewhere. The principles of this type of trial will be introduced and
contrasted with those of the ‘‘conventional clinical trial.’’
This essay, one in a series on research methods, will (NIH)-sponsored trials that entail potential risks to
discuss 2 issues in the design and analysis of randomized participants. The role of the DSMC is often thought
clinical trials. The issues are (1) when a Data and Safety by investigators to be restricted to monitoring the safety
Monitoring Committee (DSMC) should recommend of the subjects enrolled in the trial. In fact, these
early termination of a randomized trial and (2) the committees play a broader role during the development
complexity of the study protocol itself. Although these and conduct of randomized trials. As an example, the
topics are not directly related, both can lead to contro- National Institute of Child Health and Human Devel-
versy when clinicians decide whether to incorporate trial opment (NICHD) Division of Intramural Research
results into practice. delineates the following as functions of DSMCs for
their clinical trials: (1) to review new intervention
protocols to suggest modifications in design, reporting
Stopping a trial early and analysis relating to safety and efficacy endpoints;
(2) to review and provide guidance regarding major
It is generally acknowledged that most randomized trials study design changes; (3) to monitor endpoint and
should have an independent DSMC. Indeed, such over- toxicity data, based on a specific and predetermined
sight is required for all National Institutes of Health schedule; (4) to make recommendations concerning
continuation, termination, or modification of ongoing
Supported by National Institutes of Health, grant no. U01-HD36801. studies; and (5) to assist in resolving problems that may
Reprints not available from the authors. arise during conduct of the study.1
doi:10.1016/j.ajog.2005.05.061
620 Thom and Klebanoff
Here we focus on the role of the DSMC in a clinical of women at high risk,3 the stratum of women with
trial after patient recruitment has started, and in par- insulin-dependent diabetes was predicted to have a
ticular on how the committee approaches the decision to preeclampsia rate of 10% in the placebo group. After
stop a trial at an interim point. For the purposes of this 3 interim analyses, it was observed that the rate of
discussion, we will consider a randomized trial designed preeclampsia in the placebo group was consistently
to test a new treatment against a control treatment such greater than 20%, and the sample size for this cohort
as placebo to determine whether the new treatment was reduced from 980 to 450.
reduces the risk of an adverse medical outcome, desig-
nated as the primary outcome. The DSMC meets during Evidence of benefit
the course of the trial to review the emerging datadnot
only safety data such as adverse events, but also the If the new treatment shows a significant beneficial effect
frequencies of primary and secondary outcomes. at an interim analysis it would be unethical not to stop
At each interim review, the DSMC may recommend the trial so that the new treatment may be made
continuing the trial as originally designed, or continuing available. However, it is important to define in advance
with a modification of the protocol. Alternatively, for what is meant by a ‘‘significant benefit’’ and to consider
several reasons the committee may recommend stopping the effect of conducting multiple interim data analyses
the trial before the estimated sample size has been on the type I error. Using a rough analogy, imagine a
reached, or before the expected follow-up period is 1-mile race between 2 horses. We agree before starting
complete. First, it is possible that the original sample that if 1 of the horses is ahead by at least 1 length at
size was overestimated. Second, the new treatment may the end of the race, then that horse is truly faster. If the
have a beneficial or a detrimental effect that is already margin of victory is less than that, then we ascribe the
apparent. Third, the trial may be deemed to be futile; in victory to luck. Now suppose that to spare the horses,
other words, it is unlikely that an answer to the primary we also decide to stop the race early if at some point one
question will be attained in a reasonable time, if ever. of them has such a large lead that the other would be
This might be due to a previously unrecognized flaw in very unlikely to close to within a length by the end of the
the trial design, or to inadequate patient recruitment. race, let alone pull ahead. There may be times during the
Fourth, there may be external influences such as chang- race when 1 horse has a 1-length lead, but how confident
ing practice or newly learned results from other studies. are we that this lead will be maintained? To stop the race
Fifth, the question under study may no longer be after a quarter mile, 1 of the horses would have to be far
relevant; the new treatment may even be ‘‘obsolete.’’ behind and almost have to break a leg leaving the
We discuss these reasons for early discontinuation of starting gate for us to be confident that the ultimate
recruitment in more detail below, with examples derived victory margin for the other would be by at least
mainly from the NICHD’s Maternal-Fetal Medicine 1 length. At the half mile mark, the lead could be less
Units (MFMU) Network, a multicenter study group overwhelming, but an interim lead of only 1 length
that conducts trials to answer timely clinical questions in would still not be enough to convince most of us that the
obstetrics and perinatal medicine. Of the 13 randomized final margin will be at least 1 length, or that the lagging
clinical trials completed by the MFMU Network to horse could not catch up. We require even less of a lead
date, 8 have been stopped before the estimated sample at 3/4 mile, but we still acknowledge a possibility that
size was reached. Of the 8 trials, 2 were discontinued for the trailing horse might catch up.
futility, and 1 for evident benefit. The rest were more Interim analyses are very similar to this analogy. At the
complicated. Some of these have been previously end of a trial, a treatment is declared to be of significant
discussed2 and some are presented here. benefit if the calculated test statistic exceeds a prespecified
statistical boundary or critical value, generally analogous
Reduction in sample size to the usual Z-value of 1.96 (P = .05) for a single test. For
interim analyses, it is not acceptable to stop the trial
The DSMC may determine that sufficient power to show merely because the test statistic exceeds 1.96. This is
equivalence between the treatments, or to exclude ben- because looking at the data more than once increases the
efit or harm of the new treatment, will be achieved chance of finding a significant difference when there is
before the full sample size has accrued. This can occur none (type I error). Therefore, for interim analyses, a
when the primary outcome occurs more frequently than more conservative boundary is useddone example is the
originally anticipated. In our experience this is unusual; Peto boundary of a Z-value of 3.5 (P = .0005).4 In the
it is more common to underestimate the required sample MFMU Network, we use a boundary that is unlikely to be
size when planning a trial. However, in one instance the crossed early in the trial unless there is a huge treatment
MFMU Network DSMC did recommend a reduction in difference, but becomes closer to 1.96 near the end of the
sample size. In a trial of low-dose aspirin to prevent trial; the exact value depending on the previous number of
preeclampsia among four separately powered strata looks at the data, and the percentage of the final sample
Thom and Klebanoff 621
size that is included in the analysis.5 In the recently at three previous interim analyses. In this case, we did
completed trial of progesterone versus placebo to prevent use a boundary for harm that was symmetric with the
recurrent preterm delivery,6 an interim analysis of 351 one used for benefit and so we would have required
delivered patients out of a projected 500 patient sample a P-value of .0002 to declare harm. In retrospect this
size revealed a preterm delivery rate of 54% in the placebo was much too conservative. Nevertheless, the committee
group and 37% in the progesterone treatment group, a recommended discontinuation of the trial because of the
significant Z-value of 3.0 (P = .0027). The boundary was trend toward harm. Also factored into their decision
at Z = 2.43, corresponding to a P-value of .0149. At the was the declining recruitment, which meant the trial
time the DSMC met, another 112 patients had already would take another 7 years to complete. In fact this
been randomized (463 total patients) but were not yet decision to stop this trial was based on a combination of
delivered. Although trials should not be stopped for reasons, evidence of both harm and futility.
benefit until the predefined boundary has been crossed,
DSMCs may opt to continue the trial once significance Futility and other reasons
has been reached to collect data on important but less
frequent secondary outcomes. In the case of this proges- Interim analysis may reveal a neutral effect or even a
terone trial, the primary outcome was delivery less than 37 trend toward a negative effect of the new treatment, and
weeks of gestation; but neonatal morbidity and mortality it may appear increasingly unlikely that a positive effect
were very important secondary outcomes. In fact, the will be demonstrated by the end of the trial. It is often
committee did consider completing recruitment of the useful to determine the conditional power of the ob-
total 500 patient cohort to examine neonatal outcome, served difference, calculated under the assumption that
but decided that it was unlikely that data derived from 37 the current trend in the data will continue until the end
patients remaining to be randomized warranted a delay in of the trial.9 Not surprisingly, if a negative or neutral
publishing the results, and recommended discontinuation trend is observed at interim analysis, the power to show
of study recruitment. a beneficial effect assuming the same trend in the
remainder will be slim to nonexistent. Caution should
Evidence of harm be exercised in discontinuing a trial on this basis alone,
however. There are many instances in the published
Interim data analysis may reveal a negative impact of literature, and even in the MFMU Network, where
the new treatment on the primary outcome, rather than trends in the data have changed over timedgoing from
a positive effect. Again, a statistical boundary is usually positive to negative or vice versa. In the trial of repeated
determined before recruitment starts such that if it is weekly versus single course steroids conducted by Guinn
crossed, the new treatment is deemed to be harmful. The et al,10 a sample size of 1000 women was originally
boundary may be symmetric with the boundary for planned to give 90% power to detect a one-third
benefit (ie, analogous to ÿ1.96) but generally a lesser reduction in a composite outcome of neonatal morbidity
degree of evidence is required to declare a negative trend and mortality, from an estimated 25% in the single-
than to declare a beneficial effect. In other words, course group to 16.5% in the weekly course group. An
DSMCs will usually require less evidence to stop study interim analysis after 308 women had been recruited,
recruitment when a new treatment might be harmful revealed very little difference between the 2 treatment
than when it might be beneficial. How asymmetric, ie, groups. The rate of composite neonatal morbidity/mor-
how small a negative effect is needed to cross the tality was 27% in the single-course group and 24% in
boundary, would depend on whether the new treatment the weekly course group, only an 11% reduction. The
is already in general use for this or another indication, group calculated the conditional power assuming this
and whether there is other evidence of benefit.7 How- trend in the remainder of the patients and of course,
ever, care should be taken not make the boundary for found it to be very low, less than 2%. They used this as a
determination of harm too easy to cross. justification to stop after 500 rather than the originally
Discontinuation of a study too early for harm may calculated 1000 patients had been enrolled. In fact, had
result in a new treatment that has promise with regard to they calculated the conditional power under the as-
secondary outcomes being ignored, or the trial results sumption of the positive trend that became evident in
being dismissed if the new treatment is already being the next cohort of recruited patients (which was actually
used. In an MFMU Network trial of metronidazole to quite close to the one third reduction assumed in the
prevent preterm delivery in women with asymptomatic sample size calculation) there would have been more
Trichomonas vaginalis,8 the DSMC reviewed the data than 75% power to show a difference by the time that
after 516 women (of the final sample size of 1900) had 1000 patients had accrued.11
been delivered. The preterm delivery rate in the active We usually recommend calculating conditional power
treatment group was higher than in the placebo group for more than one scenario, that is, not just assuming
(18.8% vs 11.5%, P = .02) as it had been consistently the observed trend to date, as Guinn et al10 did. We also
recommend calculating the conditional power assuming Summary

that there will be a positive effect for the remainder of
the trial. The positive effect that we frequently use is the In summary, data and safety monitoring committees are
same as that used for the original sample size calcula- charged with monitoring not only adverse events during
tion. As a hypothetical example, if a trial’s sample size the conduct of clinical trials, but also with monitoring
were chosen to detect a 20% reduction in the primary the trial’s progress, likelihood of coming to a successful
outcome for a new treatment compared with a control completion, and the need for modification or early
treatment, then we would calculate the conditional termination based on data derived from interim analy-
power assuming that for the remainder of the trial there ses. Although stopping rules for benefit and harm
is a 20% reduction in the primary outcome in the new should be specified before starting the trial, in practice,
treatment group. This can be described as ‘‘calculating the decision to stop a trial is usually not clear-cut, and
the conditional power under the alternative hypothesis.’’ cannot be reached solely on statistical grounds but
If the conditional power under the alternative hypoth- rather on a number of complex statistical and clinical
esis is low, there can be considerably more confidence considerations,13 including changes in prevailing prac-
about stopping the trial. tice or new data available in the literature, and invari-
In the case of Guinn et al,10 external evidence, albeit ably incorporates some subjective considerations. When
retrospective studies, had shown some evidence of harm a trial is stopped early, we believe that it is important for
and this, in addition to the results of the conditional the readers of the report of the trial to have the reasons
power analysis, was a factor in the decision. It is also given, and especially to have the importance of each
likely, as has been the case in several of the MFMU reason clarified, if there is more than 1.
Network trials, that it was not just these considerations
but the uphill battle to keep recruitment going for The large and simple clinical trial
another number of years that led to the decision to
terminate the trial earlier than first planned. As generally conducted in the United States, clinical
The MFMU Network also initiated a randomized trials are detailed, complex and expensive undertakings,
trial of repeated versus single dose prenatal corticoste- limiting the number of that can be attempted. Due to the
roids,12 using a similar design to that of Guinn et al.10 An vigorous advocacy of statistician Richard Peto, much of
interim analysis was conducted on the data from 282 ran- the rest of the world employs a different approachdthe
domized patients, the planned sample size being 2400. Large and Simple Clinical Trial.14-18 These different
The DSMC recommended discontinuation of the trial at approaches to clinical trial development have inherent
this interim review. The committee was concerned re- benefits and problems. In this discussion, we compare
garding safety as infants in the repeat course group 2 studies evaluating interventions to improve perinatal
tended to be smaller, which was concordant with external outcomesdthe NIHCD MFMU Network Trial of 17-
evidence of harm in recent publications. However, there alpha hydroxyprogesterone caproate to prevent preterm
was no difference between the groups in head circumfer- birth,6 and the UK-based ORACLE studies19,20 of
ence, the parameter of most concern. The committee maternal antibiotic therapy to improve neonatal out-
noted the slight suggestion of benefit in some of the come to illustrate these different philosophies.
pulmonary parameters, but absolutely no trend for The progesterone study was conducted in 19 US
benefit in the primary outcome of neonatal morbidity academic medical centers that had competed success-
and mortality. The committee also had concerns regard- fully for membership in the MFMU Network. As
ing the ongoing relevance of the question being studied, described previously, 463 women were enrolled in this
as the regimen of repeated steroids had been abandoned trial. The study was restricted to women who had a
by many obstetricians since the start of recruitment. In previous spontaneous preterm birth and were within a
addition, recruitment was slow, although it had more relatively narrow gestational age window in the current
than doubled since the year before. It was estimated that pregnancy. The study design required that the prior
it could take another six years to finish recruitment. spontaneous preterm birth used to determine eligibility
Finally, the outcome rate in the placebo group was be confirmed by reviewing the original medical records,
somewhat less than expected, implying that a larger and the gestational age at randomization in the current
sample size would be required, and this was also taken pregnancy had to be confirmed by sonography before
into account in their decision. Nevertheless, because inclusion. There was an extensive list of conditions in the
only 12% of the final sample size was included in the current pregnancy that rendered a woman ineligible.
interim analysis cohort, there was still reasonable power Treatment was intensivedweekly injections of 17-alpha
to show a difference under the alternative hypothesis. hydroxyprogesterone caproate or placebo; if a woman
This last case had elements of benefit, harm and futility missed her weekly clinic appointment a research nurse
and is an excellent illustration of the difficulty in visited her at home to administer the injection. Weekly
conducting and monitoring clinical trials. saliva specimens for assay of steroid hormones were
obtained from all women. The study required up to 13 often uncommon but important outcomes such as fetal/
different forms (some of which were used on multiple neonatal death, chronic lung disease, or brain damage.
occasions) for each woman enrolled; the primary study Surrogates such as preterm birth are relegated to
outcome was the prevention of birth before 37 weeks of secondary outcome status. The large and simple trial is
gestation.6 Infant outcomes were evaluated secondarily. based on the belief that as long as proper randomization
In contrast, the ORACLE studies were conducted in is used, the outcome is important and assessed in an
161 volunteer sites and enrolled 8710 women in multiple unbiased manner, and an intent-to-treat analysis is
countries. The eligibility criteria for the ORACLE I followed, large numbers are more important than highly
trial19 included a pregnancy less than 37 weeks of standardized data quality.21
gestation complicated by preterm premature rupture of Yusuf et al15 advance 3 major underlying principles to
membranes (preterm PROM), and managing clinician guide development of the large and simple trial. First,
uncertainty regarding the need to prescribe antibiotics. identification of effective treatments is likely to be relevant
The eligibility criteria for the companion ORACLE II if the disease being treated is common and the condition
trial20 included a pregnancy at less than 37 weeks of being prevented is important. In perinatology, important
gestation in suspected or definite labor with intact mem- outcomes include maternal, fetal and/or neonatal death;
branes, and the managing clinician was uncertain as to or serious, long-term neonatal outcomes such as chronic
whether to prescribe antibiotics. No definitions were lung disease or cerebral palsy. Except in populations at
provided for how membrane rupture or labor was to be very high-risk, these outcomes are uncommon.
diagnosed, how gestational age was to be ascertained, or Prolongation of a pregnancy and reduction of the
why the clinician did not believe antibiotics to be either frequency of preterm birth are merely surrogates for
indicated or not indicated. The lists of exclusions for both these more important outcomes. Second, ‘‘miracle treat-
trials were short: antibiotics already being prescribed, ments’’ like penicillin come along perhaps once in a
immediate delivery desirable or unstoppable, fetus not lifetime. The real difference between 2 treatments, or
sufficiently preterm to cause concern (all of these as between treatment and no treatment, will probably not
determined by the enrolling clinician), and specific con- be large. However, even moderate reductions in impor-
traindications (such as allergy) to the antibiotics being tant outcomes (say, from 5% to 4%) may be worth
used. The treatment was not intensivedwomen were achieving. Third, although some readily identifiable
given packs of pills to be taken for 10 days or until subgroups of patients might respond more favorably
delivery. The study used only 3 single-sided data than others to a given treatment, the effect of treatment
formsd1 collected at randomization, 1 at delivery, and is unlikely to be reversed in different subgroups. In other
1 at death or hospital discharge of the neonate. The words, treatments might be more helpful in some preg-
primary study outcome was not merely prevention of nancies than in others, but it is unlikely that a treatment
preterm birth, but rather fetal/neonatal death, major will help some while harming others. Peto et al17 suggest
adverse outcome (such as chronic lung disease) or major that when the overall study result indicates no benefit,
cerebral abnormality on ultrasound, all determined by any benefit observed in a particular subgroup of partic-
the time of hospital discharge of the newborn infant. ipants is probably a false-positive result. When the over-
all result indicates a benefit, failure to find a benefit in a
particular subgroup is probably a false-negative result.17
Philosophy of the large and simple trial If one accepts these principles (and not everyone
does), then the need for and design of large and simple
As demonstrated previously, the philosophical under- trials is clear. The rarity of important outcomes and the
pinnings of the conventional trial and the large and likely modest benefits that one can expect of treatment
simple trial are fundamentally different. Peto et al14-18 mean that clinical trials must be large to achieve
have written extensively over the years about how large adequate power. The need to accrue large numbers of
and simple trials should be conducted and why they are patients means that common conditions must be stud-
valid. The 2 types of trials are compared in the Table. In ied, and that the study protocol must be simple,
contrast with the conventional trial, the large and simple collecting only minimal prognostic and outcome data
trial uses few specific inclusion or exclusion criteria, from each individual patient. These principles also imply
relying instead on each individual clinician’s level of that inclusion and exclusion criteria can be minimal,
uncertainty. It collects minimal data, puts minimal usually limited to clinician uncertainty regarding the
burden on the enrolling clinicians and participating need for treatment and the absence of a few absolute
women, and uses a study protocol that is very simple contraindications to use of the treatment. The underly-
and easy to follow. Large and simple trials enroll many ing assumption that a beneficial treatment will benefit
more participants than conventional trials; and their everyone also supports broad inclusion and minimal
primary outcomes, rather than being surrogate measures exclusion criteria. The broad, nonspecific inclusion crite-
such as preterm birth or prolongation of pregnancy, are ria and simple treatments are also suggested to improve
Table Comparison of Conventional and Large and Simple Clinical Trials

Conventional Trial Large and Simple Trial
Number of patients Small to moderate Large
Number of clinics Few (usually !20) Many (up to hundreds)
Support for clinics Clinics compete for award Mainly volunteer, minimal compensation
Who benefits from Specific subgroups benefit. Everyone benefits or no one benefits.
treatment Therefore detailed inclusion Therefore few inclusion and exclusion
and exclusion criteria needed. criteria needed. Main criterion: ‘clinician
uncertainty regarding need for treatment’
Treatment protocol Complex, defined in detail Simple, minimally defined
Reliance on routine Low. Often requires special High
clinical care research clinics
Quality control Extensive Minimal
Number of forms Many Few
Biospecimens saved Often Rarely
Study outcome Surrogate conditions (preterm birth, Severe conditions (fetal, neonatal, maternal
preeclampsia, etc) death; convulsions, stroke, cerebral palsy, etc.)
Advantages Entry criteria well defined- applicability clear Severe outcomes studied
Extensive quality control Patients more like those in routine practice
Extensive data and specimen collection Simple, practical protocol
can help explain study results
Disadvantages Complex protocol might not be practical Entry criteria vague- applicability to any
in routine setting given patient uncertain
Surrogate outcomes studied Little quality control
Patients often highly selected Minimal data and specimen collection-
difficult to explain study results
generalizability of the trial’s results, as they might mimic In their textbook of clinical trials, Friedman et al21
actual practice more closely. Peto et al17 have noted that provide a balanced discussion of the advantages and
‘‘There is no need to collect large amounts of data on disadvantages of large and simple trials. They note that
prognostic features, nor for extensive ‘quality’ review, this design works best when the intervention is simple,
since detailed analyses of such data add little to the the condition being treated is easily diagnosed, both the
statistical power of the trial, and are generally counter- outcome and important prognostic variables can be
productive by discouraging participation.’’ Because the determined without the need for specialized tests, and
outcomes are ‘‘hard’’ endpoints such as death, complex the follow-up time is relatively short. When these
diagnostic algorithms are not necessary. conditions cannot be met, the large and simple trial
may not be feasible. Although large and simple clinical
Arguments against large and simple trials trials in perinatal medicine are rarely done in the United
States, they are the most common type of clinical trial in
Some of the most cogent arguments against the use of the rest of the world.
the large and simple trial were made by the late internist In practice, the complexity and size of a trial depend
and clinical epidemiologist Alvan Feinstein22 who drew on the specific question being answered and the setting
the distinction between evaluating the average perfor- where the trial will be conducted. Although there have
mance of 2 treatments and evaluating treatments for the been exceptions, new treatments that are complicated,
care of patients. Feinstein complained that lumping expensive, or acutely risky should probably be tested in
together heterogeneous groups of patients sharing noth- a relatively small number of centers where skilled,
ing more than a broad, unverified definition of some specifically trained personnel can be assembled, and
disease and clinician uncertainty of the need for treat- protocol adherence and participant safety can be closely
ment may produce results with excellent statistical monitored. The cost and recruitment difficulties of such
stability but which will be relatively worthless when a study will probably mandate a small-to-moderate
clinicians try to apply the results in practice. In other sample size, and therefore a surrogate outcome. Con-
words, the large and simple trial may gloss over impor- versely, trials relying on volunteer physicians (as might
tant clinical distinctions, and as such would not be be the case in countries where funds to establish the
helpful to clinicians in deciding how to treat any elaborate structure of a conventional trial are not
particular patient. available) are probably best served when they evaluate
relatively simple, inexpensive, low-risk interventions and controlled trials in the NICHD MFMU Network. Sem Perinatol
use a protocol that does not interfere with routine 2003;27:253-60.
3. Caritis S, Sibai B, Hauth J, Lindheimer M, Klebanoff M, Thom E,
clinical care. A more severe primary outcome is well- et al, for the NICHD Maternal Fetal Medicine Units Network.
advised to maintain clinical credibility in the face of a Low dose aspirin to prevent preeclampsia in women at high risk.
relatively nonspecific study protocol; fortunately, the N Eng J Med 1998;338:701-5.
low cost and burden of such a study makes the required 4. Peto R, Pike MC, Armitage P, Breslow NE, Cox DR, Howard SV,
large sample size feasible. et al. Design and analysis of randomized clinical trials requiring
prolonged observation of each patient: I, introduction and design.
Most MFMU Network trials follow the middle road Br J Cancer 1976;34:585-612.
of a ‘‘larger and less complex’’ protocol. A relatively few 5. Lan KKG, DeMets DL. Discrete sequential boundaries for clinical
critical entry criteria, such as history of preterm birth in trials. Biometrika 1983;70:659-63.
the progesterone trial,6 history of preeclampsia or chronic 6. Meis P, Klebanoff M, Thom E, Dombrowski M, Sibai B, Moawad
hypertension in the high-risk aspirin trial,3 the diagnoses A, et al, for the NICHD Maternal Fetal Medicine Units Network.
Prevention of recurrent preterm delivery by 17 alpha-hydroxypro-
of bacterial vaginosis and trichomoniasis in the BVTV gesterone caproate. N Eng J Med 2003;348:2379-85.
trials,8,23 and gestational age in virtually all trials, are 7. DeMets DL, Pocock SJ, Julian DG. The agonising negative trend
defined according to a specific and detailed protocol. in monitoring of clinical trials. Lancet 1999;354:1983-8.
Specific, detailed protocols are developed to diagnose the 8. Klebanoff MA, Carey JC, Hauth JC, Hillier SL, Nugent RP,
primary and often a few key secondary study outcomes; Thom EA, et al, for the NICHD Maternal Fetal Medicine Units
Network. Failure of metronidazole to prevent preterm birth
centrally done chart reviews are used as needed to verify among asymptomatic pregnant women with Trichomonas vagi-
these outcomes. Other measures are standardized only nalis. N Eng J Med 2001;345:487-93.
when deemed absolutely necessary. Wherever feasible, 9. Lan KKG, Wittes J. The B-value: a tool for monitoring data.
the remainder of patient care is left to routine practice, Biometrics 1988;44:579-85.
and relevant data are abstracted from the regular clinic 10. Guinn DA, Atkinson MW, Sullivan L, Lee MJ, MacGregor S,
Parilla B, et al. Single vs weekly courses of antenatal corticosteroids
chart. Albeit more complex than trials such as ORA- for women at risk of preterm delivery. JAMA 2001;286:1581-7.
CLE,19,20 this approach is considerably less complex than 11. Jenkins TM, Wapner RJ, Thom EA, Das AF, Spong CY. Are
that used by trials such as the multiple risk factor weekly courses of antenatal steroids beneficial or dangerous?
intervention trial.24 We believe this strikes a favorable JAMA 2002;287:187-8.
balance between cost, complexity, and clinical credibility. 12. Wapner R for the NICHD MFMU Network. A randomized trial
of single vs. weekly courses of corticosteroids. Am J Obstet
Which is the ‘‘correct’’ way to conduct a trial? In the Gynecol 2003;189:S56.
final analysis, for a clinical trial to be of value, its results 13. Pocock SJ. When to stop a clinical trial. BMJ 1992;305:235-40.
must influence practice. Clinicians might question the 14. Peto R. Clinical trial methodology. Biomedicine 1978;28:24-36.
validity of a large and simple trial because of the 15. Yusuf S, Collins R, Peto R. Why do we need some large, simple
vagueness of the entry criteria or the lack of quality randomized trials? Stat Med 1984;3:409-20.
16. Peto R, Collins R, Gray R. Large-scale randomized evidence:
control. Conversely, clinicians might question the rele- large, simple trials and overviews of trials. Ann NY Acad Sci
vance of a conventional trial because the detailed entry 1993;703:314-40.
criteria excluded the very patients whom that clinician 17. Gray R, Collins R, Peto R. Making randomized trials larger:
believes might have benefited from treatment, because a simple solution? Eur J Surg Oncol 1995;21:137-9.
the complex trial procedures could never be accom- 18. Peto R, Baigent C. Trials: the next 50 years. BMJ 1998;317:170-1.
19. Kenyon SL, Taylor DJ, Tarnow-Mordi W, ORACLE Collabora-
plished in the ‘‘real world,’’ or because the outcome of tive Group. Broad-spectrum antibiotics for preterm, prelabour
the study was not sufficiently severe to be worth rupture of fetal membranes: the ORACLE I randomised trial.
preventing in its own right. Ultimately, whether large Lancet 2001;357:979-88.
numbers of patients and severe study outcomes can 20. Kenyon SL, Taylor DJ, Tarnow-Mordi W, ORACLE Collabora-
compensate for reduced data collection and quality tive Group. Broad-spectrum antibiotics for spontaneous preterm
labour: the ORACLE II randomised trial. Lancet 2001;357:989-94.
control is a question of judgment, not of statistics. 21. Friedman LM, Furburg CD, DeMets DL. Fundamentals of clinical
trials. 3rd ed. New York: Springer-Verlag, Inc; 1998. p. 56-7.
References 22. Feinstein A. Reply to Yusuf. Stat Med 1984;3:421-2.
23. Carey JC, Klebanoff MA, Hauth JC, Hillier SL, Thom EA, Ernest
1. Policies and procedures for the NICHD Intramural Clinical Trial JM, et al. Metronidazole to prevent preterm birth in asymptomatic
Data and Safety Monitoring Committee. Available at http://icrp. women with bacterial vaginosis. N Engl J Med 2000;342:534-40.
nichd.nih.gov/dsmc_charter.html. Accessed on December 11, 2004. 24. The multiple risk factor intervention trial (MRFIT). A national
2. Thom E, Rouse D for the NICHD Maternal Fetal Medicine Units study of primary prevention of coronary heart disease. JAMA
Network. What we have learned about conducting randomized 1976;235:825-7.
www.ajog.org
CLINICAL OPINION
Research agenda for preterm birth: Recommendations

from the March of Dimes
Nancy S. Green, MD,a,b,* Karla Damus, RN, PhD,a,c Joe Leigh Simpson, MD,d
Jay Iams, MD,e E. Albert Reece, MD, PhD, MBA,f Calvin J. Hobel, MD,g
Irwin R. Merkatz, MD,c Michael F. Greene, MD,h Richard H. Schwarz, MD,i
and the March of Dimes Scientific Advisory Committee on Prematurity
March of Dimes, White Plains, NY a; Departments of Pediatrics and Cell Biology,b and Department of Obstetrics &
Gynecology and Women’s Health,c Albert Einstein College of Medicine, Bronx, NY; Department of Obstetrics and
Gynecology and Molecular and Human Genetics, Baylor College of Medicine, Houston, TXd; Department of
Obstetrics and Gynecology, Ohio State University, Columbus, OHe; Department of Obstetrics and Gynecology, Dean’s
Office, University of Arkansas for Medical Sciences, Little Rock, ARf; Departments of Obstetrics, Gynecology and
Pediatrics, Cedars-Sinai Medical Center, University of California Los Angeles School of Medicine, Los Angeles, CAg;
Department of Obstetrics and Gynecology and Reproductive Biology, Harvard Medical School, Boston, MAh;
Department of Obstetrics and Gynecology, Maimonides Medical Center, Brooklyn, NY i
Received for publication November 11, 2004; revised February 7, 2005; accepted February 18, 2005
KEY WORDS Preterm birth (PTB) is a common, serious, and costly health problem affecting nearly 1 in 8 births
Prematurity in the United States. Burdens from PTB are especially severe for the very preterm infant (!32
Research weeks’ gestation), comprising 2% of all US births. Successful prevention needs to include newly
Very preterm focused and adequately funded research, incorporating new technologies and recognition that
Common complex genetic, environmental, social, and behavioral factors interact in complex pathogeneses and
disorder multiple pathways leading to PTB. The March of Dimes Scientific Advisory Committee created
this prioritized research agenda, which is aimed at garnering serious attention and expanding
resources to make major inroads into the prevention of PTB, targeting six major, overlapping
categories: epidemiology, genetics, disparities, inflammation, biologic stress, and clinical trials.
Analogous to other common, complex disorders, progress in prevention will require incorpo-
rating multipronged risk reduction strategies that are based on sound scientific discovery, as well
as on effective translation into clinical care.
Preterm birth (PTB) is the major determinant of early black infant mortality (!365 days of life) in the United
childhood mortality and morbidity and is the leading States.1 As much as half of all pediatric neurodevelop-
cause of neonatal mortality (!28 days of life) and of mental problems can be ascribed to preterm birth.2
Moreover, the severity and incidence of adverse out-
* Reprint requests: Nancy Green, MD, Medical Director, March of
comes inversely correlates with gestational age, espe-
Dimes, 1275 Mamaroneck Ave, White Plains, NY 10605. cially birth before 32 completed weeks of gestation.
E-mail: ngreen@modimes.org Beyond the enormous impact on affected families, the
doi:10.1016/j.ajog.2005.02.106
Green et al 627
Table I Incidence of PTB and very PTB in the United States, 2002, by race/ethnicity
Total US Non-Hispanic Non-Hispanic Hispanic Non-Hispanic Non-Hispanic Native
Race/ethnicity (%) black (%) white (%) (%) Asian (%) American (%)
PTB rate (!37 wks) 12.1 17.7 11.0 11.6 10.4 13.0
Very PTB rate (!32 wks) 2.0 4.0 1.6 1.7 1.4 2.0
All race categories exclude Hispanic births. Source: National Center for Health Statistics.
economic costs are staggering. Although only 9% of all

Table II Data collection variables for PTB
infant hospitalizations in 2002 were related to prematu-
rity, the charges were $15.5 billion, representing nearly Plurality All prior pregnancy outcomes
half of all charges for infant hospitalization (unpub- Maternal race/ethnicity Interval between pregnancies
Maternal age Prenatal care (onset, timing/
lished data from March of Dimes Perinatal Data Center.
number of visits)
Agency for Healthcare Research and Quality, National
Maternal education Clinical course of pregnancy
Inpatient Sample, 2002.) Paternal race/ethnicity Screening and diagnostic tests
Despite decades of research and clinical efforts, the US Medical interventions
preterm birth rate (!37 completed weeks’ gestation) rose Paternal age Substance use (alcohol, tobacco,
to an all time high of 12.1% in 2002, a 29% increase over other drugs)
the previous 2 decades.3 Most of this increase has been Stress (domestic violence,
reported in moderately preterm births (32-36 weeks of catastrophic events, racism)
gestation), as the rate for very preterm births (!32 Mode of conception Induction /augmentation of
completed weeks’ gestation) has remained virtually stable labor
at about 2%.3 Contributing factors include the increasing Mode of delivery and indication
Paternity information Method(s) used to assess
frequency of births to women older than 35 years, and the
gestational age
use of infertility treatments,4 with their significant en-
Socioeconomic status Neonatal outcomes
hancement of multiple birth rates.5,6 Preterm births in All preexisting maternal medical
singleton pregnancies result most frequently from spon- conditions (eg, infection,
taneous preterm labor and preterm premature rupture of depression, oral health)
membranes (PPROM).7-10 In addition to spontaneous Prepregnancy weight, body
PTB, 20% to 30% of preterm births are considered mass index, and weight change
medically indicated to avoid or minimize maternal and/ in pregnancy
or fetal complications, such as intrauterine growth ab- History of preterm labor/PTB
normalities. Advances in maternal, fetal, and neonatal (spontaneous and iatrogenic)
management have led to an increased willingness to Micronutrient and supplement use
deliver high-risk pregnancies preterm.11,12 Patient and
provider preferences may also be promoting clinical
interventions that are further increasing PTB rates at Research has identified numerous risk factors for
later gestations, such as scheduled deliveries.13,14 spontaneous PTB, although accurate prediction and
Rates of PTB in the United States differ profoundly prevention remain elusive.2,8-10,21-25 A few of the major
among racial/ethnic groups, with the largest and most risk factors have biologic plausibility, such as multifetal
persistent disparities occurring between non-Hispanic gestation, a history of prior PTB or preterm labor and
white and non-Hispanic black births. In 2002, the rates certain cervical, uterine, and placental structural or
of preterm and very preterm births for Hispanics were physiologic abnormalities. Much research has focused
only modestly higher than for non-Hispanic white on the roles of infection and the resultant cascade of
infants, but respective rates for non-Hispanic black immunologic effects. However, environmental interac-
infants were 60% and 250% higher (Table I, NCHS tionsdbroadly defined to include behavioral factors and
data). Native Americans have intermediate rates, social underpinningdare also likely to play important
whereas Asian PTB rates are the lowest of all these but less well understood roles. Other factors include
groups. These disparities remain even after stratification race/ethnicity, maternal age, socioeconomic status, in-
by plurality and adjusting for possible confounders such terconceptional interval, maternal medical conditions,
as education and occupation, thus likely reflecting a maternal weight, nutritional status, substance abuse,
combination of genetic, environmental, and social paternal effects, stress, depression and other factors
factors.15-20 Racial/ethnic disparities in PTB are associ- (Table II). The precise mechanisms by which these risk
ated with persistent gaps in chronic health outcomes factors influence the preterm birth rate and the potential
throughout life.15-20 impact of their successful modification on PTB are
628 Green et al
gation should include consideration of the broad range of

factors and mechanisms that need to be prioritized by
frequency and by level of impact; and (2) that successful
interventions will likely need to simultaneously and/or
sequentially target multiple risk factors and will require
tailoring for individuals and/or specific communities.
This work will be aided by use of the tools of genomics,
proteomics, and genetic epidemiology and by an im-
proved understanding of pathophysiology in preterm
birth. As a common, complex disorder, PTB also meets
the criteria recently set forth for problems of high research
priority and of major public health significance.29
Consistent with this conceptual framework, March of
Dimes initiated its Perinatal Research Initiative (PERI)
in 1998 to fund multidisciplinary research targeting
the interactions between epidemiologic and biologic
Figure Complex Interactions of factors and pathways lead-
ing to PTB. Depiction of combinations of etiologic pathways bases of prematurity.37,38 Most recently, the Foundation
and interacting factors leading to PTB. The elements shown launched its National Prematurity Campaign in 2003 to
here are selectively chosen, unweighted, and not intended to stimulate public concern for prematurity and to reduce
represent known quantifiable contributions or relationships. the nation’s rate of PTB,39,40 with campaign partners of
the American College of Obstetricians and Gynecolo-
largely unknown. Unfortunately, most biomarkers as- gists (ACOG), American Academy of Pediatrics (AAP),
sessing risk of PTB have poor positive predictive value and Association of Women’s Health, Obstetrics and
for guiding clinical interventions.2,10,21-26 Neonatal Nursing (AWHONN). A principal aim of the
PTB is now best understood as the clinical endpoint campaign is to foster research on PTB through increased
for a number of potential causes and pathogeneses support from federal and other funding sources, includ-
(Figure). In this context, at least 4 major pathophysio- ing the March of Dimes.
logic pathways have been described toward the shared
outcome of PTB: inflammation/infection with its asso- Development of a research agenda
ciated maternal and fetal cytokine response; maternal/
fetal stress with generation of placental and fetal mem- A Scientific Advisory Committee (SAC) on Prematurity
brane-derived corticotrophin-releasing hormone, which was created to advise the National March of Dimes
in turn, enhances placental estrogen and fetal adrenal Campaign (membership listed in the Appendix). The SAC
cortisol production; abruption or decidual hemorrhage has interdisciplinary representation of national experts
with thrombin-induced protease expression and distur- in PTB research and treatment, from fields of obstetrics-
bances in uterine tone; and mechanical stretch due to gynecology, pediatrics, nursing and public health, with
multifetal pregnancy or polyhydramnios-induced abnor- representation from the 3 campaign partner organiza-
mal uterine and cervical distention.7 These pathophysio- tions. It was charged to develop a national research
logic pathways may occur independently but more agenda that included prioritized recommendations. After
commonly are present in various degrees of combination. deliberation, the SAC concluded that research efforts
Research is needed to define both the biologic pathways of should focus primarily on cause and prevention of very
PTB and successful clinical interventions. preterm birth (!32 weeks). The rationale was that the
In contemporary terms therefore, PTB is increasingly study of the most severely affected would likely be more
conceptualized as a ‘‘common, complex disorder,’’27,28 biologically distinct and thus most amenable to discovery,
defined as a condition stemming from heterogeneous com- and may translate into the greatest impact on child health.
posites of multiple gene-environment interactions.29-31 Four subcommittees were formed to address the interre-
The prototype of such disorders is cardiovascular disease, lated research categories of: basic science, clinical man-
which arises from multiple inherited factors, including agement, social and behavioral interactions, and racial/
metabolism of cholesterol, fat, homocysteine, and insulin, ethnic disparities. Committee processes included formal
acting often synergistically with modifying environmental reviews of extant publications from relevant disciplines,
and behavioral influences such as diet, weight, exercise, and based on the members’ experience as investigators and
smoking. The evidence supporting this conceptualization clinicians. External review of the resultant document was
of PTB includes findings of familial aggregation,32,33 then obtained from additional experts representing the
nonmendelian heritability,34 high rates of recurrence,24,35 Prematurity Campaign partner organizations ACOG,
and the existence of ethnic/racial disparities.15-17,36 For AAP, and AWHONN, as well as the National Institute
PTB, 2 corollaries thus follow: (1) that etiologic investi- of Child Health and Human Development (NICHD),
Green et al 629
Centers for Disease Control (CDC), and Society of

Table III Six main research targets
Maternal Fetal Medicine (SMFM).
The current report summarizes recommendations for 1. Contemporary epidemiologic studies of very PTBs
(%32 weeks’ gestation)
addressing specific research strategies, methodologies and
Data collection variables
approaches needed to stimulate and implement the pri-
Adequate sample size
oritized prematurity research agenda. These recommen- Biomarkers of PTB
dations emphasize the most promising experimental Epidemiologic overlap between PTB and other adverse birth
paradigms to elucidate the complex contributions of outcomes
genetic and environmental influences. Research on PTB 2. Genes and gene-environment interactions
needs to: (1) better define the etiologic mechanisms High-risk phenotypes
responsible for PTB; (2) identify biomarkers for PTB to New analytic techniques and bio-information systems
improve clinical risk assessment; (3) develop clinical Biobanking
interventions that lead to reduction in rates of PTB; and Animal models
(4) eliminate disparities in PTB among racial/ethnic 3. Racial/ethnic disparities
Risk factor analysis
groups in the United States. Genetics is a relatively new
Genetic factors
aspect of prematurity research that brings an exciting
Behavioral factors
array of possibilities for new avenues of investigation Health care delivery
through new technologies. Ultimately, translating the find- 4. Role of inflammatory responses in PTB
ings from genetic research into modification of risk for PTB Micro-organisms
for individuals and for communities will need to take into Infection
account the effects of complex environmental influences. Effects of antibiotics
By establishing a national emphasis on strategies that Role of the cervix
have proven successful for other complex common Immune pathways
disorders, the Foundation hopes to attract investigators Immune modulation
to this important arena, including some not previously 5. Stress responses and PTB
Biologic assessment of maternal and fetal stress
vested in prematurity research. The March of Dimes thus
Effects of stress on racial/ethnic disparities
seeks to inspire novel innovative approaches to address
6. Clinical trials
the challenges inherent in translating science into clinical Biomarkers
practice. A second March of Dimes report on prematurity Progesterone supplementation
will be forthcoming to address broader research topics Infection
within public, provider and health care delivery issues, Abruption and impaired uterine-placental blood flow
such as the impact of education, clinical management, Multifetal pregnancy
racial/ethnic disparities, stress, and behavioral modifica- Assisted reproductive technologies
tions, particularly as they impact the large group of Risk reduction strategies
moderately premature infants (32-36 weeks’ gestation).
early preterm births. Some of the latter were histor-
ically considered stillbirths, but now constitute the
Major recommendations source of the majority of perinatal and infant
Preface morbidity and mortality. Robust epidemiologic
studies should be conducted on heterogeneous pop-
Decreasing PTB will require enhanced efforts that ulations for which data are collected on all relevant
explore mechanisms of specific contributing genetic biologic and environmental risk factors and clinical
and environmental factors and their interactions, as management variables. These studies must be ade-
well as a better understanding of uterine and placental quately powered to establish sample sizes large
pathophysiology. These findings should lead to hypoth- enough to accommodate appropriate stratification,
esis-driven, controlled clinical trials for prediction and adjustment, and multivariate analytic techniques.
prevention of PTB, focussed on those births occurring Although cross-sectional and case-control studies
before 32 weeks’ gestation. are important, prospective cohort studies and cohort
The SAC recommends 6 main targets for research studies of sequential pregnancies should also be
initiatives (Table III): performed to generate positive and negative predic-
tive profiles of very preterm births.
1. Contemporary epidemiologic studies of very preterm
births (%32 weeks’ gestation). The traditional epi- a. Data collection variables. To provide uniformity,
demiology of PTB is rooted in decades-old clinical comparability, and to conduct appropriate adjust-
data, much of which was acquired outside of the ments and stratification that was based on prob-
United States9,41 and does not focus as heavily on able factors that afford risk or protection,
630 Green et al
epidemiologic studies of preterm births should col- gies permit comprehensive and pathway-based
lect at least the clinical variables listed in Table II. genomic/proteomic analyses without presupposed
b. Adequate sample size. Study designs should take identification of target genes and proteins.45,46
into account genetic and other heterogeneity, with Applications should be paired with collection
adequate case ascertainment and patient selection and storage of biologic specimens and bioinfor-
that meet specific inclusion and exclusion criteria. matics support. These tools should be applied in
Sample sizes should be specified to ensure that conjunction with epidemiologic analyses, family
studies have adequate power to address the studies, cohorts of severely affected subjects,
hypotheses in question. community-based intervention trials, and relevant
c. Biomarkers of PTB. Prediction of PTB should animal models.
include the identification of specific markers of c. Biobanking. Research studies on PTB should in-
risk and their incorporation into epidemiologic clude collection and storage of biologic specimens
analyses. Such markers include genetic polymor- with the minimum of collection of maternal blood/
phisms (see below), alterations in level of expres- DNA samples. Sampling should include other
sion of specific genes and susceptibility to adverse maternal samples (urine, vaginal secretions, pla-
effects from smoking, genital tract infection, fetal centa, uterine tissue), paternal samples, and fetal
fibronectin, cervical length, and contour and other samples (amniotic fluid, cord blood). Samples
influential variables. should be appropriately banked and made avail-
d. Epidemiologic overlap between PTB and other able to future investigators.
adverse birth outcomes. Other adverse perinatal d. Animal models. Better use of animal models is
outcomes commonly occur with PTB, such as needed to elucidate the underlying mechanisms of
birth defects42 and intra-uterine growth restric- preterm labor and PPROM and to identify prom-
tion.43,44 The association between PTB with other ising interventions. Rodent models are useful for
adverse birth outcomes need definition to delin- some areas of investigation, despite not being ideal
eate overlapping mechanisms and to facilitate models of PTB. ‘‘Knock out’’ and ‘‘knock down’’
predictions of treatment efficacy and outcomes. models of known and novel genes are needed.
Primates and other mammals are useful and re-
2. Genes and gene-environment interactions. A search
quire political as well as financial support.49,50
for biologic regulatory substrates and circuits inte-
gral to preterm delivery is required. The qualitative 3. Racial/ethnic disparities. PTB is among the most
and quantitative roles of host susceptibility in infec- disparate health outcomes in the United States, espe-
tious and other environmental processes, the effects cially for black Americans15-18,20,51 (Table I). Re-
of specific genes, genetic pathways, and epigenetic search on PTB should work toward attaining the
regulation on host responses and on the biology and Healthy Peoples 2010 goal of eliminating health
molecular biology of parturition are all promising disparities. Innovative research must address both
areas for research.27,34,45,46 Likely pathways of focus the underlying biology of disparate gestational length
include inflammation, which is mediated through and inequities in systems of health care delivery so
proinflammatory and anti-inflammatory pathways that responsive, culturally sensitive interventions can
involving cytokines such as tumor necrosis factor a be developed to reduce and eventually eliminate those
(TNF-a), interleukin-1b. Other pathways include inequities.15,17,18,52 Multidisciplinary approaches
those regulating pro-apoptotic and anti-apoptotic should combine biologic, sociologic, and epidemio-
pathways, tissue maintenance, remodeling and stretch, logic paradigms, including prospective family and
hemostasis and hemorrhage, cellular turnover and community cohorts. These studies should encom-
degradation.45,47,48 Promising strategies include: pass quantifiable issues involving prior reproductive
outcomes, medical conditions, infection, immune
a. High-risk phenotypes. Studies of individuals and
regulation, stress, and other biomarkers (described
families with recurrent, idiopathic very PTB
previously), poverty, depression, racism, substance
should be given priority. Such cohorts provide
and physical abuse, weight, and nutrition. Socio-
opportunities for identifying relevant genetic and
medical interventions, including those applied at the
environmental factors and predictive biomarkers,
community level should be studied. Similar ap-
including some that may be amenable to inter-
proaches have had demonstrable success in preven-
vention or individualized risk assessment.
tion and therapeutic intervention in other common
b. New analytic techniques and bio-information
complex disorders such as cardiovascular disease.
systems. New techniques for genetic, functional
Specific research priorities include:
genomic, and proteomic analysis should be de-
veloped and applied. Although candidate gene a. Risk factor analysis. Develop improved measures
approaches may be useful, microarray technolo- of risk factors to investigate the potential for higher
Green et al 631
prevalence of specific risk factors among black d. Role for the cervix. Study the mechanical and
women. immunologic role played by the uterine cervix.
b. Genetic factors. Assess differential ethnic effects on e. Immune pathways. Identify pathways that regulate
biologic responses of specific genetic factors, such maternal and fetal humoral and cell-mediated
as polymorphisms in genes regulating immune immune responses to infection and noninfectious
responses and steroid receptor polymorphisms38 inflammatory processes.
and their interactions with various environmental f. Immune modulation. Study the effects of suppres-
(including social and behavioral) factors.48 sion or augmentation of inflammation on PTB.
c. Behavioral factors. Determine the differential and Modulation may need to be general or targeted to
cumulative effects of behavioral risk factors such particular inflammatory mechanisms, possibly af-
as multiple partners, interval between pregnan- fecting associated neurodevelopmental problems
cies, sexual activity during pregnancy, use of such as cerebral palsy.
barrier contraceptive methods that may prevent
5. Stress responses and PTB. Pregnancy itself is a stress
ascending infections, douching, substance abuse
on normal physiology and endocrinology of women,
(tobacco, alcohol, and other drugs), nutrition,
with simultaneous yet competing demands to sustain
preconception weight, and pregnancy weight gain
the fetus and to adapt to foreign antigens. Some
on PTB.
pregnancy states may be more susceptible to effects
d. Health care delivery. Study how components of
from stress, such as in multifetal pregnancies, given
health care delivery systems including access, qual-
the extra physiologic demands placed on the mater-
ity of care, content, nontreatment, unequal treat-
nal, fetal, and placental systems. Maternal emotional,
ment, cost, and reimbursement impact on rates
psychosocial, nutritional, or other environmental
of PTB.
stress may increase risk of PTB via effects on several
4. Role of inflammatory responses in PTB. The micro- interacting mechanisms18,48,50,69-71: (1) neuroendocrine
bial environment and host response play pivotal roles pathways via maternal and/or fetal hypothalamic-
in both preterm labor and PPROM.18,46,53-55 Infec- pituitary-adrenal (HPA) neuroendocrine axes (eg,
tions are implicated in at least 40% of PTB,53,56,57 and ACTH, cortisol) that result in premature triggering
as much as 70% for PTB less than 32 weeks,58 as of labor and/or a greater degree of activation of the
documented by clinical signs, histologic chorioamni- placental-fetal endocrine systems, including cortico-
onitis, and microbial cultures of fetal membranes59,60 trophin releasing hormone (CRH) and estrogens; (2)
and amniotic fluid,61 consistent with an ascending immune mechanisms regulated by inflammatory re-
pathway. Inflammation may play a primary or sec- sponses, likely affecting susceptibility to maternal
ondary role in PTB.46,53,55,62 For example, treatment infection such as bacterial vaginosis51; (3) intrauter-
of infections often does not ameliorate, and may even ine/fetal inflammatory processes, thereby promoting
paradoxically increase the risk of PTB.63-65 This PTB through proinflammatory mechanisms; and (4)
unexpected effect may reflect further stimulation by maternal-placental-fetal vascular function, including
endotoxin, and/or by-products of microbial demise both uterine and umbilical blood flow.
of inflammatory pathways already triggered by Studies should be performed to delineate the role that
infection, or may reflect a shared pathway between biologic and other types of acute and chronic stressors
susceptibility to infection and an underlying exagger- play in PTB and in racial/ethnic disparities, functioning
ated inflammatory response.53,55 These responses within the context of the common complex disorder of
to infection reflect an integration of genetic and PTB. These studies should include:
environmental influences that need elucidation, with
a. Biologic assessment of stress, its timing and its
development of predictive biomarkers. Important
effects on maternal and fetal health. Focus on im-
research areas are as follows:
proved biologic definition and quantification, and
a. Micro-organisms. Document the presence and identification of functional links between the bio-
chronology of micro-organisms, including myco- logic, social, and environmental risk factors. This
plasma species, chlamydia, and viruses in the endo- should include validated quantitative biomarkers,
metrium and decidua that may contribute to PTB. such as those described previously, as well as CRH,
b. Infection. Elucidate the role of chronic endome- plasma and salivary cortisol, and other relevant
tritis, ascent of vaginal flora and of inflammation markers in the neuroendocrine stress and inflam-
at distant sites, eg, periodontal disease on preterm matory pathways; mechanisms underpinning the
labor and birth.66-68 associations between stress, the HPA axis and the
c. Effects of antibiotics. Investigate the results of autonomic nervous system, including the effects of
prenatal and preconception antibiotic treatment glucocorticoids and catecholamines; immunologic
on women with genital infection and/or prior PTB. changes such as type1/type 2 cytokine/cellular
632 Green et al
responses; vascular effects in the placental circula- risk of preterm labor and/or PPROM, as de-
tion, and related physiologic mechanisms. scribed previously.
b. Effects of stress on racial/ethnic disparities. Bio- b. Progesterone supplementation. Administered em-
logically relevant and quantified markers of stress pirically in the second trimester, progesterone
should be examined regarding their role in dis- appears to reduce the risk of recurrent PTB for a
parate outcomes among sub-populations, as de- select group of women at very high risk for
scribed in target area 4 (above). PTB.72,73 This recent clinical success should be
used to stimulate studies on the mechanisms of
6. Clinical Trials. In light of the overall paucity of prior
action, optimal form, dose and route of adminis-
success with clinical trials and interventions in reduc-
tration, identification of women who will, and just
ing the incidence of PTB, a change in expectation is
as importantly, who will not benefit from proges-
needed. The investigative approach should be broad-
terone prophylaxis use in stratified risk groups,
ened so that the research is no longer organized to
and clinical management and follow-up of off-
reduce PTB by eliminating the effect of single risk
spring.
factors. Instead it should be guided by the overriding
c. Infections. Screening and management of possible
concept of PTB as a common complex disorder, in
bacterial and nonbacterial pathogens. These stud-
which interacting environmental and heritable fac-
ies should examine effects from local, systemic
tors must be addressed, concomitantly with the
and distant infection such as bacterial vaginosis,
clinical milieu. Moreover, progress may prove to be
sexually transmitted diseases, urinary tract infec-
incremental rather than monumental. Promising
tions, or periodontal disease.66,68 In addition to
interventions need to be tested in controlled trials of
novel strategies for antibiotic therapies, treatment
sufficient sample size for stratification, enroll well-
options may include manipulation of the maternal
defined populations, and include a clinical database
and fetal cytokine milieu.
and anticipatory maternal and fetal/neonatal bio-
d. Abruption and impaired uterine-placental blood
logic sampling. Studies should be primarily random-
flow. Screening and intervention for causes and
ized trials with clearly established mechanistic
consequences of compromised placental blood
underpinnings and biologic analyses. Opportunities
flow43,44 and/or decidual hemorrhage abruption,
to clarify areas of known disparities should receive
including acquired and inherited thrombophilias
priority. Appropriate outcome measures for these
and other coagulation abnormalities,74 aberrant
studies should include perinatal morbidity and mor-
response to protease-activated receptor (PAR)-
tality, in addition to duration of prolongation of
ligand interactions.75
pregnancy and gestational age at delivery.
e. Multifetal pregnancies. Determine optimal man-
A comprehensive agenda for clinical investigation re-
agement for spontaneous and artificially conceived
quires: (a) better timing and protocols for treatments
multifetal pregnancies, including the traditional
based on improved techniques for diagnosis of preterm
approaches, such as bed rest, tocolysis, cerclage,
labor; (b) identification of high-risk women both before
and potential strategies to prevent PTB, such as
(preconception) and during early pregnancy for risk
supplemental progesterone, anticoagulation, en-
reduction and early intervention; and ultimately, (c)
hanced nutrition, and/or antioxidants.
reduction of risk among all women of reproductive
f. Assisted reproductive technologies (ART). Eluci-
age through education, risk reduction programs, and
date the relationship between ART and PTB for
innovative community-based interventions with contem-
both multiple and singleton gestations, including
porary standards. These efforts should be population-
the attributable contribution of underlying mater-
based, culturally sensitive, and appropriate for race,
nal and paternal pathology. Technical aspects
ethnicity, parity, and educational level. Successful inter-
should be studied to optimize the yield of healthy
ventions will need to address multiple risk factors simul-
births.4,76,77
taneously or sequentially, and will require tailoring for
g. Risk reduction strategies. Studies should evaluate
individuals and/or specific communities. They should
the impact on rates of PTB of comprehensive
also be optimally timed during the preconception, prena-
individual risk assessment and, where possible,
tal, and interconception periods to attain the highest
intervention with both genetic and environmen-
impact on rates of PTB and neonatal morbidity and
tal variables (Table II). Interventions include
mortality.
eliminating or minimizing environmental factors
Clinical investigations of women at increased risk of
such as smoking78 and exposure to other toxic
PTB should integrate the areas described in the major
agents and stressors, treating underlying
target areas 1 through 5 above, as well as:
medical disorders, and optimizing preconcep-
a. Biomarkers. Determine effective biologic markers tional and prenatal maternal medical and mental
with positive and negative predictive value for health.
Green et al 633
Conclusion communities to increasingly respond to this serious birth

outcome.
Productive research on PTB requires several essential
elements included in this report: promising paradigms to Acknowledgments
frame investigative approaches; innovative methodolo-
gies to address the problem; well-delineated genotypes We acknowledge the thoughtful input from the leader-
and phenotypes; sufficient access to research participants, ship of ACOG, AAP, and AWHONN, SMFM, as well
patient data, and clinical samples, facilitated by biobanks; as Dr Catherine Spong at NICHD. We also thank
clinical animal models and in vitro materials; adequate several March of Dimes staff members for their valuable
funding; and trainee support. As for other common, contributions: Drs Jennifer Howse, Michael Katz,
complex disorders, research on PTB must consider the Joann Petrini, Siobhan Dolan; and Motoko Oinuma,
contributions of genetic, psychosocial, cultural, racial/ Ann Umemoto, Lorraine Gore.
ethnic, nutritional, behavioral, and environmental fac-
tors. Research programs already aimed at prevention of References
prematurity and that are consistent with this report’s
recommendations include the March of Dimes PERI 1. Mathews TJ, Menacker F, MacDorman MF. Infant mortality
research portfolio and the newly created Prematurity statistics from the 2001 period linked birth/infant death data set.
Research Initiative (PRI), which specifically focuses on Natl Vital Stat Rep 2003;52:1-28.
2. Goldenberg RL. The management of preterm labor. Obstet
the causes of PTB,79 existing federally funded research Gynecol 2002;100:1020-37.
programs on approaches such as the use of progesterone, 3. Martin JA, Hamilton BE, Sutton PD, Ventura SJ, Menacker F,
studies on racial/ethnic disparities, community-based Munson ML. Births: final data for 2002. Natl Vital Stat Rep
programs, and a new NICHD program, ‘‘Genomic and 2003;52:1-113.
Proteomic Network for Premature Birth Research,’’80 as 4. Wright VC, Schieve LA, Reynolds MA, Jeng G, Kissin D. Assisted
reproductive technology surveillancedUnited States, 2001.
well as being incorporated in the National Children’s MMWR Surveill Summ 2004;53:1-20.
Study championed by NICHD. Sufficient funding must 5. Russell RB, Petrini JR, Damus K, Mattison DR, Schwarz RH.
be made available to support and sustain these efforts. To The changing epidemiology of multiple births in the United States.
expand the ranks of scientists focused on PTB, new Obstet Gynecol 2003;101:129-35.
investigators in PTB need to be recruited through 6. Reynolds MA, Schieve LA, Martin JA, Jeng G, Macaluso M.
Trends in multiple births conceived using assisted reproductive
sustained efforts to attract and retain culturally compe- technology, United States, 1997-2000. Pediatrics 2003;111:1159-62.
tent, junior and multidisciplinary investigators. 7. Lockwood CJ, Kuczynski E. Markers of risk for preterm delivery.
Design and testing of promising methods to prevent J Perinat Med 1999;27:5-20.
PTB are best served by enhanced understanding of basic 8. Iams JD. The epidemiology of preterm birth. Clin Perinatol
biologic mechanisms that inform well-focused, robust, 2003;30:651-64.
9. Berkowitz GS, Papiernik E. Epidemiology of preterm birth.
controlled, clinical trials and innovative community- Epidemiol Rev 1993;15:414-43.
based, participatory research. Ideally, basic, clinical, 10. Meis PJ, Goldenberg RL, Mercer BM, Iams JD, Moawad AH,
and community research programs are designed to Miodovnik M, et al. The preterm prediction study: risk factors for
reciprocally inform all responsible research on PTB. indicated preterm births. Maternal-Fetal Medicine Units Network
However, identification of effective interventions to pre- of the National Institute of Child Health and Human Develop-
ment. Am J Obstet Gynecol 1998;178:562-7.
vent PTB based on sound epidemiologic approaches may 11. Joseph KS, Kramer MS, Marcoux S, Ohlsson A, Wen SW, Allen
be possible, despite limits in the understanding of specific A, et al. Determinants of preterm birth rates in Canada from 1981
biologic mechanisms. Two recent, notable examples of through 1983 and from 1992 through 1994. N Engl J Med
successful perinatal interventions are the periconcep- 1998;339:1434-9.
tional intake of folic acid to prevent serious neural tube 12. Joseph KS, Demissie K, Kramer MS. Obstetric intervention,
stillbirth, and preterm birth. Semin Perinatol 2002;26:250-9.
defects81 and the supine positioning and safe sleep 13. Ezra Y, McParland P, Farine D. High delivery intervention rates
environment to prevent Sudden Infant Death Syn- in nulliparous women over age 35. Eur J Obstet Gynecol Reprod
drome.82,83 Promising treatments such as supplemental Biol 1995;62:203-7.
progesterone to prevent recurrent PTB also exemplify the 14. Papiernik E, Zeitlin J, Rivera L, Bucourt M, Topuz B. Preterm
need to expeditiously test the most promising interven- birth in a French population: the importance of births by medical
decision. BJOG 2003;110:430-2.
tions based on epidemiologic, laboratory, and clinical 15. Hogan VK, Njoroge T, Durant TM, Ferre CD. Eliminating
research. The urgent problem of PTB, with its profound disparities in perinatal outcomesdlessons learned. Matern Child
impact on health throughout life and on health dispar- Health J 2001;5:135-40.
ities, mandates intense, priority research attention 16. Collins JW Jr, Hammond NA. Relation of maternal race to the
through a multipronged ‘‘full court press.’’ The March risk of preterm, non-low birth weight infants: a population study.
Am J Epidemiol 1996;143:333-7.
of Dimes and its SAC are hopeful that approaching PTB 17. Rowley DL. Closing the gap, opening the process: why study social
as a common complex disorder will catalyze research, contributors to preterm delivery among black women. Matern
improvements in clinical care, and the education of Child Health J 2001;5:71-4.
634 Green et al
18. Wadhwa PD, Culhane JF, Rauh V, Barve SS, Hogan V, Sandman 39. Massett HA, Greenup M, Ryan CE, Staples DA, Green NS,
CA, et al. Stress, infection and preterm birth: a biobehavioural Maibach EW. Public perceptions about prematurity: a national
perspective. Paediatr Perinat Epidemiol 2001;15:17-29. survey. Am J Prev Med 2003;24:120-7.
19. Meis PJ, Goldenberg RL, Mercer BM, Iams JD, Moawad AH, 40. March of Dimes Prematurity Campaign. Available at: www.
Miodovnik M, et al. Preterm prediction study: is socioeconomic marchofdimes.com/prematurity/. Accessed November 10, 2004.
status a risk factor for bacterial vaginosis in black or in white 41. Baird D. The epidemiology of prematurity. J Pediatr 1964;65:
women? Am J Perinatol 2000;17:41-5. 909-24.
20. Alexander GR, Kogan MD, Himes JH, Mor JM, Goldenberg R. 42. Rasmussen SA, Moore CA, Paulozzi LJ, Rhodenhiser EP. Risk for
Racial differences in birthweight for gestational age and infant birth defects among premature infants: a population-based study.
mortality in extremely-low-risk US populations. Paediatr Perinat J Pediatr 2001;138:668-73.
Epidemiol 1999;13:205-17. 43. Ferrazzi E, Bellotti M, Galan H, Pennati G, Bozzo M, Rigano S,
21. Spong CY. No easy answers for predicting preterm labor. Obstet et al. Doppler investigation in intrauterine growth restrictiond
Gynecol 2002;100:219-20. from qualitative indices to flow measurements: a review of the
22. Iams JD, Goldenberg RL, Mercer BM, Moawad AH, Meis PJ, experience of a collaborative group. Ann N Y Acad Sci
Das AF, et al. The preterm prediction study: can low-risk women 2001;943:316-25.
destined for spontaneous preterm birth be identified? Am J Obstet 44. Rigano S, Bozzo M, Ferrazzi E, Bellotti M, Battaglia FC, Galan
Gynecol 2001;184:652-5. HL. Early and persistent reduction in umbilical vein blood flow in
23. Lockwood CJ. Predicting premature deliverydno easy task. the growth-restricted fetus: a longitudinal study. Am J Obstet
N Engl J Med 2002;346:282-4. Gynecol 2001;185:834-8.
24. Mercer BM, Goldenberg RL, Moawad AH, Meis PJ, Iams JD, 45. Hao K, Wang X, Niu T, Xu X, Li A, Chang W, et al. A candidate
Das AF, et al. The preterm prediction study: effect of gestational gene association study on preterm delivery: application of high-
age and cause of preterm birth on subsequent obstetric outcome. throughput genotyping technology and advanced statistical meth-
National Institute of Child Health and Human Development ods. Hum Mol Genet 2004;13:683-91.
Maternal-Fetal Medicine Units Network. Am J Obstet Gynecol 46. Romero R, Chaiworapongsa T, Kuivaniemi H, Tromp G. Bacte-
1999;181:1216-21. rial vaginosis, the inflammatory response and the risk of preterm
25. Iams JD, Goldenberg RL, Meis PJ, Mercer BM, Moawad A, Das birth: a role for genetic epidemiology in the prevention of preterm
A, et al. The length of the cervix and the risk of spontaneous birth. Am J Obstet Gynecol 2004;190:1509-19.
premature delivery. National Institute of Child Health and Human 47. Simhan HN, Krohn MA, Roberts JM, Zeevi A, Caritis SN.
Development Maternal Fetal Medicine Unit Network. N Engl J Interleukin-6 promoter -174 polymorphism and spontaneous pre-
Med 1996;334:567-72. term birth. Am J Obstet Gynecol 2003;189:915-8.
26. Hibbard JU, Tart M, Moawad AH. Cervical length at 16-22 48. Wadhwa PD, Culhane JF, Rauh V, Barve SS. Stress and preterm
weeks’ gestation and risk for preterm delivery. Obstet Gynecol birth: neuroendocrine, immune/inflammatory, and vascular mech-
2000;96:972-8. anisms. Matern Child Health J 2001;5:119-25.
27. Dolan S, Damus K, Green NS, Merkatz IR, Iams JD. Preterm 49. Gravett MG, Novy MJ, Rosenfeld RG, Reddy AP, Jacob T,
birth: a common complex disorder. Submitted 2004. Turner M, et al. Diagnosis of intra-amniotic infection by proteo-
28. Dolan S, Damus K, Green NS, Iams JD, Merkatz IR. Should mic profiling and identification of novel biomarkers. JAMA
preterm birth now be classified as a ‘‘common complex disorder’’? 2004;292:462-9.
Am J Obstet Gynecol 2004;191:S117. 50. Bloomfield FH, Oliver MH, Hawkins P, Campbell M, Phillips DJ,
29. Merikangas KR, Risch N. Genomic priorities and public health. Gluckman PD, et al. A periconceptional nutritional origin for
Science 2003;302:599-601. noninfectious preterm birth. Science 2003;300:606.
30. Iams J. Atherosclerosis: a model for spontaneous preterm birth. 51. Culhane JF, Rauh V, McCollum KF, Elo IT, Hogan V. Exposure
Prenat Neonat Med 1998;3:138-41. to chronic stress and ethnic differences in rates of bacterial
31. Villar J, Abalos E, Carroli G, Giordano D, Wojdyla D, Piaggio G, vaginosis among pregnant women. Am J Obstet Gynecol
et al. Heterogeneity of perinatal outcomes in the preterm delivery 2002;187:1272-6.
syndrome. Obstet Gynecol 2004;104:78-87. 52. Board on Health Sciences Policy, Institute of Medicine. Unequal
32. Porter TF, Fraser AM, Hunter CY, Ward RH, Varner MW. The treatment: confronting racial and ethnic disparities in health care.
risk of preterm birth across generations. Obstet Gynecol Washington (DC): National Academies Press; 2002.
1997;90:63-7. 53. Klein LL, Gibbs RS. Use of microbial cultures and antibiotics in
33. Treloar SA, Macones GA, Mitchell LE, Martin NG. Genetic the prevention of infection-associated preterm birth. Am J Obstet
influences on premature parturition in an Australian twin sample. Gynecol 2004;190:1493-502.
Twin Res 2000;3:80-2. 54. Ugwumadu A, Manyonda I, Reid F, Hay P. Effect of early oral
34. Dizon-Townson DS. Preterm labour and delivery: a genetic clindamycin on late miscarriage and preterm delivery in asymp-
predisposition. Paediatr Perinat Epidemiol 2001;15:57-62. tomatic women with abnormal vaginal flora and bacterial vagin-
35. Clausson B, Lichtenstein P, Cnattingius S. Genetic influence on osis: a randomised controlled trial. Lancet 2003;361:983-8.
birthweight and gestational length determined by studies in 55. Mercer BM, Goldenberg RL, Das AF, Thurnau GR, Bendon RW,
offspring of twins. BJOG 2000;107:375-81. Miodovnik M, et al. What we have learned regarding antibiotic
36. Branum AM, Schoendorf KC. Changing patterns of low birth- therapy for the reduction of infant morbidity after preterm
weight and preterm birth in the United States, 1981-98. Paediatr premature rupture of the membranes. Semin Perinatol
Perinat Epidemiol 2002;16:8-15. 2003;27:217-30.
37. Holzman C, Jetton J, Siler-Khodr T, Fisher R, Rip T. 56. Gibbs RS, Davies JK, McDuffie RS Jr, Leslie KK, Sherman MP,
Second trimester corticotropin-releasing hormone levels in Centretto CA, et al. Chronic intrauterine infection and inflamma-
relation to preterm delivery and ethnicity. Obstet Gynecol tion in the preterm rabbit, despite antibiotic therapy. Am J Obstet
2001;97:657-63. Gynecol 2002;186:234-9.
38. Wang X, Zuckerman B, Pearson C, Kaufman G, Chen C, 57. Hitti J, Tarczy-Hornoch P, Murphy J, Hillier SL, Aura J,
Wang G, et al. Maternal cigarette smoking, metabolic gene Eschenbach DA. Amniotic fluid infection, cytokines, and adverse
polymorphism, and infant birth weight. JAMA 2002;287: outcome among infants at 34 weeks’ gestation or less. Obstet
195-202. Gynecol 2001;98:1080-8.
Green et al 635
58. Creasy R, Resnik R. Maternal-fetal medicine: principles and 75. Keelan J, Helliwell R, Nijmeijer B, Berry E, Sato T, Marvin K,
practice, 5th ed. Philadelphia: Saunders; 2004. et al. 15-deoxy-delta12,14-prostaglandin J2-induced apoptosis in
59. Hitti J, Hillier SL, Agnew KJ, Krohn MA, Reisner DP, Eschen- amnion-like WISH cells. Prostaglandins Other Lipid Mediat
bach DA. Vaginal indicators of amniotic fluid infection in preterm 2001;66:265-82.
labor. Obstet Gynecol 2001;97:211-9. 76. Schieve LA, Meikle SF, Ferre C, Peterson HB, Jeng G, Wilcox LS.
60. Romero R, Gomez R, Ghezzi F, Yoon BH, Mazor M, Edwin SS, Low and very low birth weight in infants conceived with use of
et al. A fetal systemic inflammatory response is followed by the assisted reproductive technology. N Engl J Med 2002;346:731-7.
spontaneous onset of preterm parturition. Am J Obstet Gynecol 77. Jackson RA, Gibson KA, Wu YW, Croughan MS. Perinatal
1998;179:186-93. outcomes in singletons following in vitro fertilization: a meta-
61. Goldenberg RL, Hauth JC, Andrews WW. Intrauterine infection analysis. Obstet Gynecol 2004;103:551-63.
and preterm delivery. N Engl J Med 2000;342:1500-7. 78. Melvin CL, Dolan-Mullen P, Windsor RA, Whiteside HP Jr,
62. Romero R, Espinoza J, Chaiworapongsa T, Kalache K. Infection Goldenberg RL. Recommended cessation counselling for pregnant
and prematurity and the role of preventive strategies. Semin women who smoke: a review of the evidence. Tob Control
Neonatol 2002;7:259-74. 2000;9(suppl 3):III80-4.
63. Klebanoff MA, Carey JC, Hauth JC, Hillier SL, Nugent RP, 79. March of Dimes Prematurity ResearchInitiative. Available at: www.
Thom EA, et al. Failure of metronidazole to prevent preterm marchofdimes.com/aboutus/691_11117.asp. Accessed on November
delivery among pregnant women with asymptomatic Trichomonas 10, 2004.
vaginalis infection. N Engl J Med 2001;345:487-93. 80. NIHCD Genomic and Proteomic Network for PrematureBirth
64. Andrews WW, Sibai BM, Thom EA, Dudley D, Ernest JM, Research. Available at: http://grants.nih.gov/grants/guide/rfa-files/
McNellis D, et al. Randomized clinical trial of metronidazole plus RFA-HD-04-002.html. Accessed on November 10, 2004.
erythromycin to prevent spontaneous preterm delivery in fetal 81. Honein MA, Paulozzi LJ, Mathews TJ, Erickson JD, Wong LY.
fibronectin-positive women. Obstet Gynecol 2003;101:847-55. Impact of folic acid fortification of the US food supply on the
65. Andrews WW, Goldenberg RL, Hauth JC, Cliver SP. Interconcep- occurrence of neural tube defects. JAMA 2001;285:2981-6.
tional antibiotics to prevent spontaneous preterm birth (SPTB): a 82. Rasinski KA, Kuby A, Bzdusek SA, Silvestri JM, Weese-Mayer
randomized trial [abstract] Am J Obstet Gynecol 2003;189:S57. DE. Effect of a sudden infant death syndrome risk reduction
66. Jeffcoat MK, Hauth JC, Geurs NC, Reddy MS, Cliver SP, education program on risk factor compliance and information
Hodgkins PM, et al. Periodontal disease and preterm birth: results sources in primarily black urban communities. Pediatrics
of a pilot intervention study. J Periodontol 2003;74:1214-8. 2003;111:e347-54.
67. Task Force on Periodontal Treatment of Pregnant Women. 83. Changing concepts of sudden infant death syndrome: implications
American Academy of Periodontology statement regarding perio- for infant sleeping environment and sleep position. American
dontal management of the pregnant patient. J Periodontol Academy of Pediatrics. Task Force on Infant Sleep Position and
2004;75:495. Sudden Infant Death Syndrome. Pediatrics 2000;105:650-6.
68. Lopez NJ, Smith PC, Gutierrez J. Periodontal therapy may reduce
the risk of preterm low birth weight in women with periodontal
disease: a randomized controlled trial. J Periodontol 2002;73:911-24.
69. Hobel CJ, Dunkel-Schetter C, Roesch SC, Castro LC, Arora CP.
Maternal plasma corticotropin-releasing hormone associated with Appendix
stress at 20 weeks’ gestation in pregnancies ending in preterm
delivery. Am J Obstet Gynecol 1999;180:S257-63. Members of the March of Dimes Scientific Advisory
70. Hobel C, Culhane J. Role of psychosocial and nutritional stress on
poor pregnancy outcome. J Nutr 2003;133:1709S-17S. Committee on Prematurity:
71. Wadhwa PD, Garite TJ, Porto M, Glynn L, Chicz-Demet A,
Dunkel-Schetter C, et al. Placental corticotropin-releasing hor- Frederick C. Battaglia, MD, Lillian R. Blackmon, MD,
mone (CRH), spontaneous preterm birth, and fetal growth restric- Coleen Boyle, PhD, Harvey J. Cohen, MD, PhD,
tion: a prospective investigation. Am J Obstet Gynecol Margaret C. Freda, EdD, RN, Fredric Frigoletto,
2004;191:1063-9. MD, Ronald S. Gibbs, MD, Michael F. Greene, MD,
72. Meis PJ, Klebanoff M, Thom E, Dombrowski MP, Sibai B, Calvin J. Hobel, MD, Vijaya Hogan, DrPH, MPH, Jay
Moawad AH, et al. Prevention of recurrent preterm delivery by 17
alpha-hydroxyprogesterone caproate. N Engl J Med 2003;348: D. Iams, MD, Richard B. Johnston, Jr, MD, Mark A.
2379-85. Klebanoff, MD, MPH, Eve M. Lackritz, MD, Charles
73. da Fonseca EB, Bittar RE, Carvalho MH, Zugaib M. Prophylactic J. Lockwood, MD, Irwin R. Merkatz, MD, Mary Lou
administration of progesterone by vaginal suppository to reduce Moore, PhD, RNC, E. Albert Reece, MD, PhD, MBA,
the incidence of spontaneous preterm birth in women at increased
Richard H. Schwarz, MD, Joe Leigh Simpson, MD.
risk: a randomized placebo-controlled double-blind study. Am J
Obstet Gynecol 2003;188:419-24.
Research subgroup chairs: Dr Simpson–Basic science;
74. Lockwood CJ. Inherited thrombophilias in pregnant patients: Dr Iams–Clinical science; Dr Hogan–Disparities;
detection and treatment paradigm. Obstet Gynecol 2002;99:333-41. Dr Hobel–Multidisciplinary and behavioral approaches.
www.ajog.org
REVIEW ARTICLE
Headache as a side effect of combination estrogen-

progestin oral contraceptives: A systematic review
Elizabeth W. Loder, MD,a Dawn C. Buse, PhD,b Joan R. Golub, MDc
Departments of Medicine,a Psychiatry (Psychology),b and Obstetrics and Gynecology,c

Harvard Medical School, Boston, MA
Received for publication September 21, 2004; revised November 30, 2004; accepted December 28, 2004
KEY WORDS Objective: We conducted a 2-part systematic review of published studies to examine the evidence
Oral contraceptives that combination oral contraceptives can aggravate or cause headache.
Migraine Study design: We used trials with a control group to assess headache risk that was attributable to
Headache oral contraceptive use and prospective cohort trials to answer important clinical questions about
Systematic review the natural history and treatment response of headache that occurs with oral contraceptive use.
Results: Because of differences in study populations, oral contraceptive formulations, trial end
points and trial duration, it was not possible to pool data; but the evidence supports several
conclusions. There is little indication that oral contraceptives have a clinically important effect on
headache activity in most women.
Conclusion: Headache that occurs during early cycles of oral contraceptive use tends to improve
or disappear with continued use. No evidence supports the common clinical practice of switching
oral contraceptives to treat headache; however, manipulating the extent or duration of estrogen
withdrawal may provide benefit.
It is believed widely that the use of combination risk of headache as a potential or actual consequence of
estrogen-progestin oral contraceptives (OCs) can aggra- OC use. Migraine affects up to 28% of women during
vate pre-existing headache or trigger the onset of trou- their childbearing years; in 1 study in which placebo was
blesome headaches. Headache is consistently among the administered to 147 women who believed they were
most common side effects that are reported with OC use taking OCs, headache, which was the second most
and is a frequently cited reason for discontinuation.1-3 frequent complaint, was reported in 16% of cycles.4,5
Women who consult physicians because of headaches OC use and migraine are independent risk factors for
frequently are advised to discontinue or avoid OCs. ischemic stroke. Good quality evidence and guidelines
An OC effect on headache is biologically plausible, but support the belief that, for some women with headache
because the background incidence and prevalence of (principally those who have migraine with aura or
headache is high in the population of women who are additional risk factors for stroke), the risk of OC use
most likely to use OCs, it is difficult to evaluate the true is unacceptably high. The risk of stroke in women who
have migraine without aura is increased by a factor of
approximately 3; in women who have migraine with
Reprints not available from the authors. aura, the risk of stroke is increased by a factor of
doi:10.1016/j.ajog.2004.12.089
Loder, Buse, and Golub 637
approximately 6 to 8, and the stroke risk increases even journal between 1966 and June 2004; (3) be described as
further with the addition of other risk factors. For a prospective study with a contemporaneous placebo,
example, the odds ratio for stroke in women who have untreated, or active treatment control group; (4) be
migraine, who smoke, and who use OCs is approxi- involved the use of combined estrogen-progestogen OCs
mately 34.6 Guidelines from both the American College that provide active treatment for 21 days followed by 7
of Obstetrics and Gynecology and the World Health days of placebo or no treatment; (5) have elicited in-
Organization discourage the use of OCs in women who formation about changes in headache or migraine; (6)
have migraine with aura.7,8 For many women, though, reported complete and interpretable data for headache or
especially those who have migraine without aura or migraine during at least an initial cycle of OC use and
other primary headache disorders, the benefits of OC reported data on headache or migraine in the control
use may outweigh the drawbacks. Better information group. To be included in the second part of this review,
about the effect of OCs on the clinical course of articles were required to meet the following criteria: (1) be
headache would be useful for physicians and patients written in English; (2) be published fully in a peer-
who are attempting to balance the benefits and dis- reviewed medical journal between 1966 and June 2004;
advantages of OC use. This is important because OCs (3) have as a predetermined study goal the prospective
are the most effective form of contraception and provide assessment of the treatment or natural history of head-
important noncontraceptive benefits.9 ache occurring with OC use; (4) provide complete and
Previous reviews of this subject have not been interpretable data.
systematic or have focused on the safety, rather than In the first part of our review, we did not include
the tolerability, of OC use in women with headache.10-12 studies of progestin-only contraceptives or extended
Because expert opinion on this matter conflicts, we duration use of OCs because most OCs in use today
carried out a 2-part systematic review to assess the are estrogen-progestin combinations given on a cyclic
strength of the evidence for a causal association between basis.13 We did not include studies without a contempo-
combination estrogen-progestin OC use and headache raneous placebo, untreated, or active control group
and to answer important clinical questions about the because they do not allow separation of the headache
treatment and natural history of headache that occurs risk that was attributable to OC use from that that was
during OC use. In the first part of the review, we attributable to a high background incidence and prev-
evaluated controlled studies to assess the risk of alence of headache. In the second part of this review, we
headache that was attributable to OC use and to did not include studies in which headache or side effects
distinguish it from the high background level of head- were not a specific focus of a prospective study because
ache in women who use OCs. In the second part of the information about adverse events in such studies is
review, we evaluated studies that provided information collected in nonstandardized ways. In evaluating the
on the natural history of OC-associated headache or its natural history of headache that occurs with OC
response to specific treatment strategies. therapy, we did not include studies that lacked in-
formation about reasons for study dropout, because
Sources such dropouts could account for an apparent headache
improvement with continued OC use.14
We collected peer-reviewed articles that assess the asso- Data were abstracted independently from each study
ciation between migraine or headache and the use of by 2 authors and compared. Differences in interpreta-
combination estrogen-progestin OCs. Studies were iden- tion were resolved by consensus and through consulta-
tified by several methods. We searched MEDLINE 1966 tion with a third author. The results were summarized as
through June 2004, MEDLINE (R) In-Process and other narrative. For each study in which it was reported, we
non-indexed citations, the Cochrane Database of Sys- calculated the number of women with existing headache
tematic Reviews, and the Cochrane Register of Con- or migraine who reported improvement, no change, or
trolled Trials through the second quarter of 2004 using the worsening in headache or migraine during at least the
following search terms: oral contraceptives, headache, first cycle of OC use. We also calculated the number of
migraine, adverse effects, and side effects. In addition, women without existing headache or migraine who
the reference lists of articles that were identified were experienced new onset of headache or migraine. From
hand-searched. We did not attempt to contact study studies that reported only percentages, we obtained the
authors. number of women by multiplying the percentage of
patients who experienced the outcome of interest by the
Study selection total number of patients in the group and rounded to the
nearest whole number. From studies that provided only
To be included in the first part of this review, articles were graphic information, we estimated the numbers or
required to meet the following criteria: (1) be written in percentages that were involved by measuring the graph.
English; (2) be published fully in a peer-reviewed medical In crossover studies, we considered data from the first
638 Loder, Buse, and Golub
period of OC use only, to minimize carryover and attended a migraine clinic. The sample size ranged from
dropout effects. 40 to 3179 women. The Table summarizes the character-
Methods quality was assessed with the Newcastle- istics and findings of these 6 studies.
Ottawa Quality Assessment Scale.15 The studies were
also assessed against the following quality criteria: Placebo-controlled studies
whether an explicit definition of migraine or headache
was provided and used for diagnosis; whether informa- Cullberg17 reported a double-blind, placebo-controlled
tion on headache change was obtained from patient study of 301 subjects, who used 3 different OCs that all
report or was based on recorded diary information; contained 50 mg of ethinyl estradiol but had different
whether information was collected on the magnitude or types of progesterone. One hundred forty of 301 women
severity of headaches; and whether the study adjusted (47%) reported baseline headache, which is a high figure
for confounding variables or predictors of headache that may reflect the systematic and careful questioning
activity. Trials were also stratified according to whether in this study. Information about new onset headache
they studied a headache or general contraceptive-seek- was not obtained, but during the 2-month duration of
ing population and whether they used high- or low-dose the study, 16 of 105 women (15%) in the OC group who
OCs. Because the dose of estrogen that is used in OCs had pre-existing headache reported improvement in
has declined over the years, we categorized studies from headaches, compared with 7 of 35 women (20%) in
the 1960s and 1970s as using ‘‘high dose’’ OCs and the placebo group. Sixty of 105 women (57%) in the OC
studies reported in 1980 or later as using ‘‘low dose’’ group who had pre-existing headache reported no
OCs.13 important change in headache, compared with 21 of
35 women (60%) in the placebo group. Twenty-nine of
105 women (28%) in the OC group who had pre-existing
Results headache reported worsening of headache, compared
with 7 of 35 women (20%) in the placebo group. There
The literature search identified 121 articles. Most of the were no statistically significant differences between the
studies reported the results of trials that were performed OC and the placebo groups.
to evaluate the contraceptive efficacy of OCs, in which A double-dummy, crossover trial of 4 different OC
information about study dropouts or adverse events preparations in a general, contraceptive-seeking popu-
such as headache inconsistently was obtained and lation used pills that were formulated to appear identical
reported. Few studies provided information about the to avoid ‘‘profound’’ changes in reports of headache and
baseline prevalence of headache in the population that other side effects that were demonstrated with changes
was studied, and even fewer included a placebo, un- in pill appearance.18 All subjects experienced a placebo
treated, or active control group that allowed assessment cycle. The duration of observation was 4 cycles for the
of headache risk that was attributable to OC use. Details initial preparation and 2 cycles after crossover. As they
of studies excluded for these and other reasons are entered the study, subjects were assigned to coded
contained in the Appendix. treatment regimens. Headache and other symptoms
Part one: headache attributable to OC use specifically were inquired about with ‘‘yes-no’’ questions
at a baseline and subsequent monthly visits. The
Seven studies that investigated the association between presence or absence of a symptom was scored and
combination OC use and headache or migraine met our then averaged over 4 cycles. The first report of any
criteria for the first part of this review. One article complaint was also tracked, which provided a method of
reported the pooled results of 2 placebo-controlled trials estimating new onset headache. No specific definition
that were conducted with identical methods and is of headache or migraine was provided. In this study, 32
treated here as a single study.16 Three studies were of 398 women (8%) women reported headache at
prospective, placebo-controlled trials.16-18 One study baseline. Four of 76 women (5%) in the placebo group
used a crossover design in which patients served as their reported headache during the first cycle of use, compared
own controls and received placebo.19 Two studies were with 15 of 79 women (19%) in the OC group. This
prospective trials with other control groups.20,21 One difference was statistically significant. However, after the
study was prospective with an untreated control first cycle of use, headache complaints in the OC group
group.20 No study met all methods criteria. All but 1 declined and were not significantly different from those in
study16 were conducted and published in the 1960s or the placebo group. Because no dropouts for headache
1970s and studied OCs with higher estrogen contents were recorded, it is unlikely that this is the reason for the
than OCs that are now in common use. Six studies decline in headache complaints over time. Of note,
examined the effect of OC use on headache or migraine complaints of headache did spike slightly in a group of
in women who attended contraceptive clinics, and 1 women who were switched from a progesterone-only to
study examined the effect of OC use in women who an estrogen-containing OC. The authors concluded from
Table Summary of trials included in Part I

Characteristics and Newscastle-Ottawa
Study Quality Score (NOS)* Results
17
Cullberg General contraceptive-seeking population; No significant difference between OC and placebo
double blind, placebo-controlled (n = 301); group on measures of worsening, improvement,
3 OCs, all with 50 mg ethinyl estradiol or no change in headache over 2 months; no
but different progestins. NOS = 5 significant difference among preparations with
different progestins.
Goldzieher et al18 General contraceptive-seeking population; Complaints of headache significantly higher in OC
double dummy (vaginal foam), placebo- group only for first cycle of use; no significant
controlled, cross over (n = 398); 4 OCs, 3 difference when results averaged over duration
with R50 mg ethinyl estradiol, 1 progestin of the study; 4/76 women (5%) in 1st cycle of
only. NOS = 8 placebo had new-onset headache compared with
31/238 women in 1st cycle of OC (13%).
Ryan19 Migraine population; open label, placebo- Migraine worse in 70%, improved in 30% over
controlled, crossover (n = 40); single OC 2-month duration of study.
with 50 mg ethinyl estradiol, 500 mg
norgestrel. NOS = 4
Coney et al16 General contraceptive-seeking population; No significant difference between OC and placebo
single-blind, placebo-controlled (n = 684); group on measure of headache reported as adverse
single OC with 20 mg ethinyl estradiol, event over 6-month duration of study.
100 mg levonorgestrel. NOS = 8
Herzberg and Coppen20 General contraceptive-seeking population; No significant difference in headache complaints
open label (n = 152); 6 OCs, all with R50 between OC users and control group over 11-month
mg ethinyl estradiol, and different progestins; study; 4% of OC group reported new onset or
barrier method control group. NOS = 7 worsened headache, compared with none of
control group.
Herzberg et al21 General contraceptive-seeking population; open Slightly more women in OC group reported moderate
label (n = 272); 3 OCs, all with R50 mg to severe headaches than in the IUD group at 3 of 4
ethinyl estradiol or equivalent and different follow-up visits; statistical comparisons not
progestins; IUD control group. NOS = 6 performed; 25/86 women (30%) who discontinued
OCs did so because of headache, compared with 0 in
the IUD group; depression did not predict
discontinuation because of headache.
Diddle et al22 General contraceptive-seeking population; open No significant difference in headache complaints
label (n = 3179, with 7710 women in untreated between OC users and untreated control subjects.
control group); multiple OCs, all R50 mg
ethinyl estradiol. NOS = 5
* Score on Newcastle-Ottawa scale for assessing the quality of cohort studies: Studies are awarded up to 9 stars for meeting measures of quality in
study selection procedures, comparability of groups, and outcome measures; the higher the number of stars, the higher the quality of a study.15
their averaged scores over all cycles that ‘‘headaches Ryan19 conducted a prospective, open-label, cross-
occur spontaneously in about 6-12% of cycles in this over study of 40 women who attended a headache clinic.
group of subjects, down from a pretreatment level of 20- Patients with contraindications to OCs were excluded.
30%. A significant increase in nausea and vomiting, All patients had moderate or severe migraine at baseline,
headache, and nervousness could be demonstrated sta- according to commonly accepted criteria for that
tistically only in the first treatment cycle with high- condition. Twenty women were assigned randomly to
estrogen agents; the frequency of all other symptoms in receive a combination OC therapy that contained 0.5 mg
all other cycles fell within the placebo range.It is norgestrel and 0.05 mg ethinyl estradiol for 2 months;
concluded that the customary uncontrolled clinical trials the other 20 women received placebo. After 2 months,
of OCs attribute to these agents a substantial incidence of patients crossed over to the other study treatment. There
side effects which, in fact, would be present without the were no dropouts. Using a proprietary headache index,
medication. Side effects such as nausea, headache.can the author calculated that headaches worsened in 28 of
be produced by high-estrogen agents, but the true 40 women (70%) and improved in 12 of 40 women
incidence is far less than generally supposed.’’ The (30%).
number of patients whose headaches improved or who Another study pooled data from 2 placebo-controlled
had no change in headaches was not reported. randomized trials in a general contraceptive-seeking
population.16 The trials were aimed at assessing weight libido before starting contraception and at intervals
change and other adverse events that were attributable during the first year of use. Side effects caused 25% of
to an OC that contained 20 mg of ethinyl estradiol and OC users and 13% of IUD users to discontinue.
100 mg of levonorgestrel. Six hundred eighty-four Depression, headaches, and loss of libido were the
women were assigned randomly in blocks of 4 to most common reasons for stopping OCs. None of the
placebo or OC. The 2 groups were similar with respect IUD users gave headache as a reason for stopping
to baseline characteristics, including headache fre- contraception; 25 of 86 women (30%) who discontinued
quency. More than one half of the women were former OCs gave headache as the reason. At the baseline visit,
OC users. Subjects kept daily diary cards and recorded 17% of women in the OC group reported headache,
any symptoms that occurred; in addition, at each compared with 30% at the second visit, 25% at the third
monthly visit subjects were asked, ‘‘How have you visit, 27% at the fourth visit, and 20% at the final visit.
been feeling since your last visit?’’ Over the course of At the baseline visit, 17% of women in the IUD group
this 6-month study, 31% of the women in the OC group reported headache, compared with 13% at the second
and 32% of the women in the placebo group reported visit, 26% at the third visit, 14% at the fourth visit, and
headache as an adverse event, a difference that was not 13% at the final visit. The authors did not report
statistically significant (P O .05). The authors concluded whether these differences were statistically significant.
that possible estrogen-related side effects such as head- This study also examined the percentage of women with
ache occurred with a similar incidence between groups, moderate or severe headaches within the group that
and low comparable numbers of participants discon- continued OCs and the group that stopped or changed
tinued because of these adverse events. OCs. No differences were detected, which led the
authors to conclude that some women were prepared
to tolerate the side effects. They also noted that those
Active control studies women who changed because of headaches did not have
a baseline depression score that was significantly differ-
A prospective cohort study followed 152 women who ent form those who changed for other reasons. This
started OCs for the first time or restarted after at least 12 study did not distinguish between new onset headache
months off.20 Six different types of OCs were used, all of and headache that improved or worsened.
which contained R50 mg of estrogen, predominantly
ethinyl estradiol. A control group of 40 women who Untreated control studies
were beginning a barrier method of contraception was
also studied. Questionnaires that obtained specific in- Diddle et al22 studied 3179 women who used several
formation on headache (including information on different OCs, all of which contained R50 mg estrogen.
whether the headache was moderate or severe, which The duration of observation is unclear from the study
is a good proxy for migraine) were administered at description, and it is not clear how questions were asked
baseline. On follow-up side effects were noted when about headache or what the interval of visits was.
symptoms appeared for the first time or worsened Patients who attended a family planning clinic were
during the course of the survey. One hundred thirty- enrolled consecutively in the study, and the definition of
six women (89%) in the OC group responded to the migraine that was used was a good approximation of
baseline and first-month questionnaire, compared with current International Headache Society criteria (head-
27 women (68%) in the control group. Thirty-five aches defined as migraine if they were described as
women (23%) in the OC group reported moderate or throbbing and had 2 of 5 of the following symptoms:
severe headache at baseline, compared with 14 women unilateral location, associated with nausea, preceded by
(35%) of the control group. At the first follow-up aura, family history migraine, history of cyclic vomiting.
questionnaire, which was administered at 5 weeks, 4% The incidence and prevalence of headaches in these
of the OC group and none of the control group reported women was compared with 7710 untreated control
new onset or worsened headache. Thirteen of 31 women subjects. Demographic characteristics of the 2 groups
(42%) who discontinued OC use during the 11-month were not reported. At baseline, 33 of 3179 of OC
period identified headache as the reason. Of those who subjects(1%) had migraine, and 44 of 3179 women
continued, moderate or severe headache was reported by (1%) had ‘‘tension’’ headache, a remarkably low prev-
7% of the OC group at 5 months and by 2% at 11 alence that suggests that subjects with headache were
months. excluded systematically from or chose to avoid study
Another study evaluated 272 women who were using participation. One hundred three women in the OC
an intrauterine device (IUD) or 1 of 3 OCs, 2 of which group (3%) complained of headache at some point
contained ethinyl estradiol 50 mg and 1 that contained during OC use, which included all patients who had pre-
50 mg of mestranol.21 Women completed a baseline existing headache. Twenty-six of 103 women had new
questionnaire regarding depression, headaches, and onset headache, for an incidence of !1%. Eight percent
of the women in the untreated control group reported and ‘‘1h’’ (defined as women with ‘‘headache.’’ Thus, the
headache, although no breakdown is given of those who 1m group contained rather well-defined cases of mi-
had migraine compared with tension headache. No graine only.
significant differences in headache were demonstrated Before treatment there were 226 women (14%) in the
among the many different OC preparations that were 1m group and 88 women (5%) in the 1h group, a total of
used in the study or between OC users and those in the 314 of 1676 women (19%) with headache. Of the 1676
control group. Twenty-three of the 103 women who had women who were studied, 1362 women (81%) had no
headache on the OC therapy discontinued because of pre-existing headache. Sixty-six of 362 women (18%)
that, but the other 82 women voiced less degree of with pre-existing migraine (group 1m) experienced
headache when reassured of the relative safety of taking a worsened condition while undergoing OC therapy;
the medication. The authors commented that ‘‘.if the authors commented that there was a trend for
headache occurred in relation to the pill the discomfort women to take sequential rather than combined pills
generally appeared shortly after the medication was to improve the condition but that the numbers were too
withheld. A third of the complaining treated patients low for definite conclusions to be drawn. Fifteen of 199
had migraine.There was a small number of women patients (8%) in the 1h group reported worsening of
where the contraceptive medication either accentuated headache; 42 of 362 women (12%) with pre-existing
the discomfort or produced it for the first time.’’ migraine (1m group) had no headaches during treat-
ment, and 87 of 362 women (24%) had fewer headaches.
In the 1h group, 24 of 119 women (20%) had fewer
Part two: clinically relevant evidence about headaches. The differences between the 2 groups were
the clinical course and treatment response not statistically significant.
of headache that occurs with OC This study also provided information on new onset
headache with OC use. Of the 1352 women with no
Nineteen studies were identified that met criteria for headache before treatment, 140 women (10%) had
inclusion in this part of the review and provided headache during treatment. The authors had calculated
clinically relevant evidence about the natural history an age-weighted ‘‘expected frequency’’ of 1.14%, and
or treatment of headache that occurred with OCs. determined that the difference was ‘‘highly significant’’
Results are grouped according to clinical topic. with a probability value of !.001. They also examined
whether there was a family history of migraine in the
Influence of expectations and belief women who had headache on OCs and found a family
history in 22.7% versus 12.5% in the 2 groups, a statisti-
Wimberly et al23 assessed 218 women to determine
cally significant difference.
whether expectations about side effects were associated
Ernst et al25 studied 3679 subjects, all of whom were
with experiencing those side effects. Twenty-five subjects
new users of a 20-mg ethinyl estradiol/150- mg desogestrel
(15%) anticipated having more headaches before taking
pill for a mean of 3.4 cycles. Of these women, 1327 (36%)
OCs; 32 women (19%) reported more headaches at 3
had headache at baseline; 191 women (14%) reported
months, a correlation no greater than that expected by
worsening headache during OC use. New onset of
chance alone. Most participants previously had used
headache was reported in 0.5%. Seven hundred fifty-
OCs.
nine women (57%) who received OCs reported improve-
The natural history of OC-associated headache ment in headache, although 377 women (28%) reported
no change in headache. No dropouts because of headache
Larsson-Cohn and Lundburg24 followed 1676 women were reported.
from a general contraceptive-seeking population for an Brill et al26 studied 3267 women for up to 18 cycles who
average of 17 cycles. This study involved 16 different received an OC that contained 30 mg ethinyl estradiol and
OCs, all of which were taken cyclically. Baseline and 0.075 mg gestodene; 513 of 3226 women (16%) reported
periodic information on headache was obtained, and the pre-existing headaches. Improvement after 3 cycles of
study used explicit criteria for migraine. Subjects were treatment was reported by 63% of subjects, although no
divided into a group of women who had headache definition of ‘‘improvement’’ is provided. No information
continuously, periodically, or recurrently and a group is provided on the percentages of women with pre-existing
who had headaches only occasionally or not at all. The headache who had no change or worsening in headache.
authors commented that they used the results from the Headache emerged as a new complaint in 8.8% of the
first group ‘‘to see what happened to women with pre- women between cycles 1 and 3, in 3.9% of women in cycles
existing headaches’’ and the results from the second 4 through 6, in 3.9% of women in cycles 10 through 12,
group ‘‘to see if OCs induced headache.’’ The first group and in 2.0% of women between cycles 16 and 18.
was further subdivided into ‘‘1m’’ (defined as women Headache was not among the reasons for study dropout.
with migraine on the basis of a detailed questionnaire) The authors commented that ‘‘headaches, including
migraine-like symptoms.lessened in severity and had The authors noted that the prevalence of headaches
a very low incidence after the initial cycles.’’ decreased significantly over the time of the entire study
Berger et al27 performed a study that examined the and that their statistical calculations did not show any
persistence of side effects that were reported during OC significant difference between baseline headache and
use. One hundred sixty women were assigned randomly to headache during OC treatment. They noted no signifi-
receive 1 of 3 OCs, all of which contained R50 mg of cant differences in headache complaints among the
ethinyl estradiol. Subjects were specifically questioned various preparations that were evaluated. No dropouts
about the presence or absence of each symptom at 2-week because of headache were reported.
intervals. The percentage of women who reported any In another trial that was undertaken to assess the side
side effect during the first 3 cycles of OC use was effects (including headache) of 5 OCs, the authors noted
approximately 50% in all groups and did not differ that ‘‘There was a significant reduction (P ! .10) in the
significantly depending on OC type. This study examined proportions of.users reporting most side effects be-
the probability of experiencing a side effect in the second tween cycles 1 and 3 and cycles 1 and 6.’’32 Generally,
cycle of use if it occurred in the first. The probability of the reduction was larger between cycles 1 and 3 than
experiencing headache in the second cycle if it was between cycles 3 and 6.
experienced in the first was .33 (1/3 women), and the Privrel and Daubenfeld33 evaluated an OC with 30 mg
probability of experiencing headache in the third cycle if it ethinyl estradiol and 75 mg gestodene in 246 women for
was experienced in the first was .10 (1/10 women). up to 6 months. Only 4 subjects discontinued because of
Rekers28 conducted a study of 1613 women and headache. The baseline prevalence of headache was
divided them into ‘‘starters’’ who had never used OCs 7.5%; this decreased to 4.4% at cycle 3 and to 4.8%
before and ‘‘switchers’’ who had previously used OCs. at cycle 6.
He noted that ‘‘.especially for nausea and headache
there were distinct differences between the two groups. Effects of age
Whereas the OC starters had higher incidences in the
Headache prevalence in women peaks during later
first treatment cycle compared with pretreatment, the
childbearing years.4 Susceptibility to OC-associated
incidences decreased already in the first cycle of use in
headache also appears to increase with age. In the pre-
the ‘switcher’ group.’’ He concluded that those women
viously reported Larsson-Cohn and Lundburg study, 24
who had not used OCs previously needed some time to
the highest incidence of new onset headache was in
adjust to the exogenous hormones, as indicated by the
women aged 36 to 40 years. Guillebaud34 reported the
increased incidence of nausea and headache in the first
incidence of headache from a large pharmaceutical
treatment cycle.
database of subjects who used a 30-mg ethinyl estradiol
Skouby29 reported the results of a multicenter clinical
OC. When stratified by age, the percentage of cycles in
trial with 1921 subjects that compared 2 low-dose OCs.
which headache was reported was lowest in subjects
This is one of the few trials for which dropouts because
aged 16 to 24 years, intermediate in women aged 25 to
of headache and other side effects were carefully
34 years, and highest in women who were R35 years.
accounted. The 24-month dropout rate because of
headache was 2.2%, and the prevalence of headache Effects of different progesterone types
dropped from 6.7% of subjects after the first cycle of OC and doses
use to 3.1% after 24 months of use. The author
concluded that this decline could not be explained by The dose and type of progesterone in OCs does not appear
the dropouts alone, which indicated that the occurrence to influence headache. A trial by Koetsawang et al35
of such side effects during hormonal treatment is time- compared 2 OCs that contained 30 mg ethinyl estradiol
dependent and decreases during OC treatment. but different progesterone (desogestrel or gestodene).
Schramm and Steffens30 performed a multicenter trial Over the 6-cycle duration of the study, headache com-
of an OC that contained 20 mg of ethinyl estradiol in plaints in the 2 groups did not differ significantly. Dunson
2620 women. This study specifically assessed not only et al36 assigned 892 subjects randomly to 1 of 2 OC
worsening headache but also the possibility of improve- preparations. Both preparations contained 30 mg of
ment in women who noted ‘‘migraine/headache’’ at ethinyl estradiol, but one preparation contained norges-
baseline. In this study, 61.6% of women with previous trel, and the other contained norethindrone. Subjects
headache said it no longer existed as a problem after 12 were specifically questioned about headache at each
cycles; 20% reported worsening; 3% reported improve- follow-up visit. No significant differences in headache
ment, and 4% reported that the condition was un- activity were reported between the 2 groups. The pre-
changed. The remainder did not answer the question. viously discussed placebo-controlled study by Cullberg17
Nilsson et al31 performed a randomized, double- also examined whether women whose OC contained
blind, crossover study of 4 high-dose OCs. Headaches a higher dose of progesterone were more likely to report
were classified as ‘‘slight,’’ ‘‘marked,’’ or ‘‘disabling.’’ headache, which was not the case.
Effects of treatment strategies that strategies that affect the duration or magnitude of
manipulate the magnitude or duration estrogen withdrawal appear to minimize subsequent
headache or migraine during OC use. One small trial
of estrogen exposure or withdrawal
evaluated estrogen supplementation during the pill-free
Headache related to OC use is widely believed to be an week of OCs and concluded that this had a beneficial
estrogen withdrawal symptom. A number of studies effect on migraine.42
examined the effect of treatment strategies that alter the Ziaei et al43 studied 143 women who took an OC that
magnitude or duration of estrogen exposure or with- contained 30 mg ethinyl estradiol and 150 mg levonorges-
drawal. One such study, whose principal intent was to trel for 3 months. After a 1-month washout period,
compare symptoms that were experienced during the 21 subjects were given the same preparation vaginally for 3
days on active pills compared with the 7 days off, months. Subjects were questioned about symptoms at
separated headache complaints into ‘‘any headache’’ baseline and then during the last month of use of each
and ‘‘headaches rated R5’’ (on a 0-10 point pain method. The side effects of vertigo and headache were
scale).37 Two hundred sixty-two subjects returned pro- reported together. Eighty-three subjects (58%) reported
spectively recorded daily diary information about symp- headache when they used OCs, compared with only 11
toms, including headache, that were attributed women (8%) with vaginal use, a difference that was highly
commonly to estrogen withdrawal. The authors sepa- statistically significant. The authors ascribed these differ-
rated subjects into ‘‘new starts,’’ which included women ences to the fact that vaginal steroids are ‘‘absorbed
who had never used OCs or who had used OCs gradually into the systemic circulation and can reach the
previously but had been off at least 3 months before target organs in the hypothalamic-pituitary–ovarian axis
study entry, and ‘‘current users,’’ who were already without first undergoing passage through the liver.’’
using other OCs but agreed to be switched to a study
preparation that contained %35 mg ethinyl estradiol and Nonhormonal treatment strategies for
a progestin. In most evaluated cycles, severe headache OC-related headache
was statistically significantly more likely to occur during
the placebo pill week and the few days leading up to it, Two studies that met our criteria evaluated nonhor-
compared with the 3 weeks on active OC. The authors monal treatment strategies to prevent OC-related head-
concluded that headache activity in the last few days of ache. In the first, a multivitamin supplement that was
active pills could be explained by the fact that combi- given to 500 women in conjunction with OCs had no
nation low-dose OCs with !35 mg of ethinyl estradiol effect on a number of adverse events, including head-
do not reliably produce complete ovarian suppression ache. This study excluded subjects with ‘‘frequent or
and that estrogen levels decline during the last week of severe migraine headaches.’’44 Diddle et al22 found that
active pills before the hormone-free interval.38,39 the use of diuretics for those women with headache on
The observation of Sulak et al37 about incomplete OCs did not produce relief.
ovarian suppression with OCs that contained very low
estrogen doses may explain the reason that switching to Clinical significance of headache that occurred
such OCs does not appear to reduce headache com- with OC use
plaints. Gerais and Rushwan40 studied 165 women who
used an OC that contained 50 mg of mestranol for 3 There is little information on the clinical significance or
months, after which 81 subjects continued that OC for 3 consequences of headache that occurred in association
months and 84 subjects were switched to an OC that with OC use. One large study of O16,000 women in
contained 30 mg of ethinyl estradiol. The switch did not a large health maintenance organization was designed to
reduce complaints of headache or other symptoms. evaluate major diseases that would require hospitaliza-
In another study, there was a statistically significant tion or that might lead to death. The authors concluded
increase in headache in patients who were switched from that results that were based on the few cases of migraine or
an OC that contained 50 mg ethinyl estradiol to an OC that tension headache that were reported on the hospital
contained 35 mg. (P ! .1).32 It has been suggested that discharge records showed no evidence of a higher risk
ovarian suppression is more complete with a 5-day among users that could not be accounted for by chance.45
hormone-free interval and that ‘‘shortening the hor-
mone-free interval might increase the contraceptive safety Comment
margin and decrease prevalence of symptoms such as
breast tenderness and headaches.’’36 Continuous or Good evidence for this issue is scarce. Because of
extended duration OC use to prevent OC-related head- differences in study populations, OC formulations, trial
ache was first suggested in the 1960s.41 This method of OC end points, and trial durations, it is not possible to pool
use is becoming more popular, although no study to date data for quantitative analysis. There is considerable
has carefully evaluated its effect on headache. Other uncertainty about optimal methods of assessment of
harmful treatment effects in systematic reviews, which a 1 in 10 chance of experiencing headache in the third
we considered in developing review questions and in month.
judging study aims and quality.46 Despite these difficul- Our review has several limitations. The search strat-
ties, we believe our results support a number of clinically egy may have missed some European studies that were
relevant conclusions: not captured through a MEDLINE search. The decision
(1) Controlled trials provide the strongest evidence of to exclude non-English articles may also have affected
a cause and effect relationship. Those trials that exist do our results. We also did not contact study authors to
not suggest a strong, durable relationship between OC inquire about non-published studies. Among the studies
use and headache for most women. In controlled trials, that we identified, most studies reported simply on
increases in headache activity occur in the early cycles of ‘‘headache,’’ and few used accepted diagnostic criteria
OC use, but few or no persistent statistical differences to classify headache. None adequately adjusted for
can be demonstrated in headache activity among groups important predictors of headache activity (such as age
of women who receive OCs and those in control groups. or the subject population enrolled in study), and this is
Furthermore, this improvement is not due to subject a possible explanation for some of the findings. In most
dropout. This evidence from controlled trials is consis- studies, no information was available regarding the
tent with evidence and investigator impressions in severity or functional consequences of the headaches
numerous less rigorous trials. that were reported. Statistics on the number of women
(2) Some women do appear to have a higher risk of who discontinue trial participation because of headache
headache exacerbation or new-onset headache attribut- or migraine would be a useful way of assessing the
able to OC use. This higher risk is most apparent in clinical relevance of such symptoms, but these statistics
women with a strong personal or family history of were reported infrequently.
troublesome headaches, particularly migraine. The in- The studies included in this review used different OCs
cidence also increases with age. Even within these higher with varying types and amounts of estrogen. In most,
risk groups, some women note improvement in head- the dose was higher than that in currently used
ache with OC use; most women report no change in combination OCs, which generally contain estrogen
overall headache activity, and headache complaints doses of !50 mg of ethinyl estradiol or occasionally
decrease with continued use. mestranol.9 Because a higher dose of estrogen may be
(3) The dose or type of progestin does not appear to more likely to lead to subsequent headache on with-
influence headache. The relationship between headache drawal, the results of this review probably overestimate
and the estrogen dose of OCs is more complex. the possibility of headache from OC use. If anything,
Headache that is related to OC use generally is pre- this limitation would strengthen, rather than weaken,
cipitated by estrogen withdrawal during the pill-free or the main conclusions of this review.
placebo pill week of treatment. It is not surprising then Some studies that were included in our review may
that estrogen supplementation during the pill-free week also have overestimated headache risk by the method in
or vaginal administration of contraceptives may de- which the information about headache was elicited. A
crease headache. However, switching to OCs that prospective, randomized investigation of the effect of
contain a very low dose of estrogen (eg, 20 or 25 mg of study design on reported rates of symptoms with OCs
ethinyl estradiol), which minimizes the magnitude of showed that asking about specific symptoms increased
estrogen withdrawal, does not improve headache. This the reporting of those symptoms compared with symp-
paradox may be due to the fact that OCs with very low toms elicited by general inquiry.48 At the same time, it is
estrogen doses do not suppress ovarian function com- also possible that some of the observed improvements in
pletely. This lack of benefit on headache activity is in headache over time are due to placebo or other non-
addition to lower rates of patient satisfaction with and specific effects that were related to participation in
adherence to very low-dose OCs because of break- a study.
through bleeding.47 Thus, an OC that provides 30 or Despite these limitations, it seems reasonable to
35 mg of ethinyl estradiol appears to be the most conclude that concerns about OC effects on headache
reasonable recommendation for women with headache should not weigh heavily in the decision about whether
who choose to use OCs. No nonhormonal treatment to use them. Rather, the contraceptive and noncontra-
strategies have been demonstrated to benefit OC-asso- ceptive benefits of OCs must be weighed against other
ciated headache. tolerability and safety concerns. For women with
(4) Regardless of cause, when headache begins or migraine, that discussion should focus on the risk of
worsens in conjunction with OC use, it tends to improve stroke that is associated with the interaction of mi-
or disappear despite continued use. Women who expe- graine, OC use, and their individual risk factors for
rience headache in the first cycle of OC use can be stroke. Further study of this issue is important because
counseled that they have only a 1 in 3 chance of there have been no large methodologically sound studies
experiencing headache the following month and only that used the OC formulations that are currently most
prescribed. Future studies should use precise definitions 20. Herzberg B, Coppen A. Changes in psychological symptoms in
of headache and migraine and capture prospective, women taking oral contraceptives. Br J Psychiatr 1970;116:161-4.
21. Herzberg BN, Drape KC, Johnson AL, Nicol GC. Oral
systematic information about the magnitude and func- contraceptives, depression, and libido. Br Med J Clin Res 1971;
tional effects of any headache changes in carefully 3:495-8.
matched OC and control groups. 22. Diddle AW, Gardner WH, Williamson PJ. Oral contraceptive
medications and headache. Am J Obstet Gynecol 1969;105:507-11.
23. Wimberly YH, Cotton S, Wanchick AM, Succop PA, Rosenthal
SL. Attitudes and experiences with levonorgestrel 100 mg/ethinyl
References estradiol 20 mg among women during a 3-month trial. Contracep-
tion 2002;65:403-6.
1. Rosenberg MJ, Waugh MS. Oral contraceptive discontinuation: 24. Larsson-Cohn U, Lundberg PO. Headache and treatment with
a prospective evaluation of frequency and reasons. Am J Obstet oral contraceptives. Acta Neurol Scand 1970;46:267-8.
Gynecol 1998;179:577-82. 25. Ernst U, Baumgartner L, Bauer U, Janssen G. Improvement of
2. Khan MA. Side effects and oral contraceptive discontinuation in quality of life in women using a low-dose desogestrel-containing
rural Bangladesh. Contraception 2001;64:161-7. contraceptive: results of an observational clinical evaluation. Eur J
3. Colli E, Tong D, Penhallegon R, Parazzini F. Reasons for Contracept Reprod Health Care 2002;7:238-42.
contraceptive discontinuation in women 20–39 years old in New 26. Brill K, Schnitker J, Albring M. Long-term experience with a low-
Zealand. Contraception 1999;59:227-31. dose oral contraceptive. Gynecol Endocrinol 1990;4:277-86.
4. Lipton RB, Diamond S, Reed M, Diamond M, Stewart WF. 27. Berger GS, Edelman DA, Talwar PP. The probability of side
Migraine diagnosis and treatment: results from the American effects with Ovral, Norinyl 1/150 and Norlestrin 1. Contraception
migraine study II. Headache 2001;41:638-45. 1979;20:447-53.
5. Aznar-Ramos R, Giner-Velazquez J, Lara-Ricalde R, Martinez- 28. Rekers H. Multicenter trial of a monophasic oral contraceptive
Manautou J. Incidence of side effects with contraceptive placebo. containing ethinyl estradiol and desogestrel. Acta Obstet Gynecol
Am J Obstet Gynecol 1969;105:1144-9. Scand 1988;67:171-4.
6. Becker WJ. Use of oral contraceptives in patients with migraine. 29. Skouby SO. Oral contraception with a triphasic combination of
Neurology 1999;41:786-93. gestodene and ethinyl estradiol: results of a multicenter clinical
7. The use of hormonal contraception in women with coexisting study. Int J Fertil (Suppl) 1987;32:45-8.
medical conditions: ACOG practice bulletin no. 18. Int J Gynecol 30. Schramm G, Steffens D. A 12-month evaluation of the CMA-
Obstet 2001;75:93-106. containing oral contraceptive Belara: efficacy, tolerability and anti-
8. World Health Organization. Improving access to quality care in androgenic properties. Contraception 2003;67:305-12.
family planning: medical eligibility criteria for contraceptive use, 31. Nilsson A, Jacobson L, Ingemanson CA. Side effects of an oral
2nd ed. Geneva: World Health Organization; 2000. contraceptive with particular attention to mental symptoms and
9. Petitti DB. Combination estrogen-progestin oral contraceptives. sexual adaptation. Acta Obstet Gynecol Scand 1967;46:537-56.
N Engl J Med 2003;349:1443-50. 32. Edelman DA, Kothenbeutel R, Levinski MJ, Kelly SE. Compar-
10. Curtis KM, Chrisman CE, Peterson HB. Contraception for ative trials of low-dose combined oral contraceptives. J Reprod
women in selected circumstances. Obstet Gynecol 2002;99:1100-12. Med 1983;28:195-200.
11. Benson MD, Rebar RW. Relationship of migraine headache 33. Privrel T, Daubenfeld O. clinical experience in Switzerland with
and stroke to oral contraceptive use. J Reprod Med 1986;31: the new monophasic oral contraceptive Minulet. Br J Clin Pract
1082-8. 1988;42:292-8.
12. Massiou H, MacGregor EA. Evolution and treatment of migraine 34. Guillebaud J. The 150/30 formulation: experience in the United
with oral contraceptives. Cephalalgia 2000;20:170-4. Kingdom. J Reprod Med 1983;28(suppl):66-70.
13. Petitti DB. Combination estrogen-progestin oral contraceptives. 35. Koetsawang S, Charoenvisal C, Banharnsupawat L, Singhakovin
N Engl J Med 2003;349:1443-50. S, Kaewsuk O, Punnahitanont S. Multicenter trial of two mono-
14. Bassol S, Alvarado G, Arreola RG, Celis-Gonzalez C, Pena EP, phasic oral contraceptives containing 30 mg ethinylestradiol and
Flores JG, et al. A 13-month multicenter clinical experience of either desogestrel or gestodene in Thai women. Contraception
a low-dose monophasic oral contraceptive containing 20 mg ethinyl 1995;51:225-9.
estradiol and 75 mg gestodene in Latin American women. 36. Dunson TR, McLaurin VL, Isrankura B, Leelapattana B,
Contraception 2003;67:367-72. Mukherjee R, Perez-Palacios G, et al. A comparative study of
15. Wells G, Shea B, O’Connell D, Peterson J, Welch V, Losos M, two low-dose combination oral contraceptives. Contraception
et al. Newcastle-Ottawa scale (NOS) for assessing the quality of 1993;48:109-19.
nonrandomized studies in meta-analysis. Last accessed September 37. Sulak PJ, Scow RD, Preece C, Riggs MW, Kuehl TJ. Hormone
16, 2004. Available at http://www.ohri.ca/programs/clinical_ withdrawal symptoms in oral contraceptive users. Obstet Gynecol
epidemiology/oxford.htm. 2000;95:261-6.
16. Coney P, Washenik K, Langley RGB, DiGiovanna JJ, Harrison 38. Fitzgerald C, Feichtinger W, Spona J, Elstein M, Ludicke F,
DD. Weight change and adverse event incidence with a low-dose Muller U, et al. A comparison of the effects of two monophasic
oral contraceptive: two randomized, placebo-controlled trials. low dose oral contraceptives on the inhibition of ovulation. Adv
Contraception 2001;63:297-302. Contraception 1994;10:5-18.
17. Cullberg J. Mood changes and menstrual symptoms with different 39. Killick SR, Fitzgerald C, Davis A. Ovarian activity in women
gestagen/estrogen combinations: a double-blind comparison with taking an oral contraceptive containing 20 mg ethinyl estradiol and
a placebo. Acta Psychiatr Scand 1972;236(suppl):259-76. 140 mg desogestrel: effects of low estrogen doses during the
18. Goldzieher JW, Moses LE, Averkin E, Scheel C, Taber BZ. hormone-free interval. Am J Obstet Gynecol 1998;179(suppl):
A placebo-controlled double-blind crossover investigation of the S18-24.
side effects attributed to oral contraceptives. Fertil Steril 40. Gerais AS, Rushwan H. A crossover pill study among Sudanese
1971;22:609-23. women. Int J Gynaecol Obstet 1985;23:229-33.
19. Ryan RE. A controlled study of the effect of oral contraceptives on 41. Whitty CWM, Hockaday JM, Whitty MM. The effect of oral
migraine. Headache 1978;17:250-2. contraceptives on migraine. Lancet 1966;1:856-9.
42. MacGregor EA, Hackshaw A. Prevention of migraine in the pill- of oral contraceptives: volume III, an interim report: a comparison
free week of combined oral contraceptives using natural oestrogen of disease occurrence leading to hospitalization or death
supplements. J Fam Plann Reprod Health Care 2002;28:27-31. in users and nonusers of oral contraceptives. J Reprod Med
43. Ziaei S, Rajaei L, Faghihzadeh S, Lamyian M. Comparative study and 1980;(6 Suppl):345-72.
evaluation of side effects of low-dose contraceptive pills administered 46. McIntosh HM, Woolacott NF, Bagnall AM. Assessing harmful
by the oral and vaginal route. Contraception 2002;65:329-31. effects in systematic reviews. BMC Med Res Methodol 2004;4:19.
44. Basnayake S, Higgins JE, Miller PC, Rogers SM, Kelly SE. Early 47. Williams JK. Rationale for new oral contraceptive dosing. Int J
symptoms and discontinuation among users of oral contraceptives Fertil Womens Med 2004;49:30-5.
in Sri Lanka. Stud Fam Plann 1984;15:285-90. 48. Talwar PP, Berger GS. A prospective, randomized study of oral
45. Ramcharan S, Pellegrin FA, Ray RM, Hsu JP. The Walnut Creek contraceptives: the effect of study design on reported rates of
Contraceptive Drug Study. A prospective study of the side effects symptoms. Contraception 1979;20:329-37.
Appendix: Excluded studies

Citation Reason for exclusion*
Affinito P, Monterubbianesi M, Primizia M, et al. Efficacy, cycle control and side-effects of two monophasic 1,2,5,10
combination oral contraceptives: gestodene/ethinylestradiol and norgestimate/ethinylestradiol.
Gynecol Endocrinol 1993;7:259-66.
Akerlund M. Comparative profiles of reliability, cycle control and side effects of two oral contraceptive 1,5,9,10
formulations containing 140 mg desogestrel and either 30 mg or 20 mg ethinyl oestradiol. BJOG
1993;100:832-8.
Apelo RA, Suplido A. Clinical assessment of a new triphasic oral contraceptive. Clin Ther1985;8:61-70. 1,2,9
Archer DF, Maheux R, DelConte A, et al. A new low-dose monophasic combination oral contraceptive 1,2, 5,10
(Alesse) with levonorgestrel 100 mg and ethinyl estradiol 20 mg. Contraception 1997;55:139-44.
Archer DF, Maheux R, DelConte A, et al. Efficacy and safety of a low-dose monophasic combination 1,2,4,5,10
oral contraceptive containing 100 mg levonorgestrel and 20 mg estradiol. Am J Obstet Gynecol
1999;181(suppl):S39-S44.
Bannemerschult R, Hanker JP, Wunsch C, et al. A multicenter, uncontrolled clinical investigation of the 1,2,5,9
contraceptive efficacy, cycle control, and safety of a new low dose oral contraceptive containing twenty
mg ethinyl estradiol and 100 mg levonorgestrel over six treatment cycles. Contraception 1997;56:
285-90.
Bassol S, Alvarado A, Celis C, et al. Latin American experience with two low-dose oral contraceptives 1,2,5,9,10
containing 30 mg ethinylestradiol/75 mg gestodene and 20 mg ethinylestradiol/150 mg desogestrel.
Contraception 2000;62:131-5.
Bassol S, Alvarado G, Arreola RG, et al. A 13-month multicenter clinical experience of a low-dose 1,2,5,9,10
monophasic oral contraceptive containing 20 mg ethinylestradiol and 75 mg gestodene in Latin
American women. Contraception 2003;67:367-72.
Becker WJ. Migraine and oral contraceptives. Can J Neurol Sci 1997;24:16-21. 8
Becker WJ. Use of oral contraceptives in patients with migraine. Neurology 1999;53:519-25. 8
Benagiano G. Comparison of two monophasic oral contraceptives: gestodene/ethinyl estradiol versus 1,2,5,10
desogestrel/ethinyl estradiol. Int J Fertil 1989;32(suppl):29-44.
Carr BR, DelConte A Using a low-dose contraceptive in women 35 years of age and over: 20 mg estrodial/ 1,2,3,4,5,9,10
100 mg levonorgestrel. Contraception 2002;65:397-402.
Carroll JD. Migraine and oral contraception. Proceedings of the International Headache Symposium, 1,11,13
Ellsinore, Denmark, 16-18 May 1971.
Counell ED. A comparative study on oral contraceptives. J Reprod Med 1970;5:38-41. 1,2,5,10
Couturier EGM, Bomhof MAM, Knuistingh N, van Duijn NP. Menstrual migraine in a representative Dutch 1,7,8,11
population sample: prevalence, disability, and treatment. Cephalalgia 2003;23:302-8.
Cullberg, J, Gelli MG, Jonsson CO. Mental and sexual adjustment before and after six months’ use of an oral 1,2,3,5
contraceptive. Acta Psychiatr Scand 1969;45:259-76.
Cupini LM, Matteis M, Truisi E, Calabresi P, Bernardi G, Silvestrini M. Sex-hormone-related events in 1,2,11
migrainous females: a clinical comparative study between migraine with aura and migraine without
aura. Cephalalgia 1995;15:140-4.
Dalton K. Migraine and oral contraceptives. Headache 1976;15:247-51. 1,2,7,8,11
de Andrade RP. A multicenter clinical evaluation of a new monophasic combination: minulet 1,2,3,5,10
(gestodene and ethinyl estradiol). Int J Fertil. 1989;34(suppl):22-30.
Dieben TCM. A comparative study of the clinical efficacy of Marvelon and a triphasic combination. 1,2,5,10
Organorama 1984;21:16-9.
Appendix (Continued)
Dionne P, Vickerson F. A double-blind comparison of two oral contraceptives containing 50 and 30 mg. 1,2,5,10
ethinyl estradiol. Curr Ther Res Clin Exp. 1974;16:281-8.
Endrikat J, Jaques A, Mayerhofer M, Pelissier C, Müller U, Düsterberg B. A twelve-month comparative 1,2,3,5,10
clinical investigation of two low-dose oral contraceptives containing 20 mg ethinylestradiol/75 mg
gestodene and 20 mg ethinylestradiol/150 mg desogestrel, with respect to efficacy, cycle control and
tolerance. Contraception 1994;52:229-35.
Endrikat J, Cronin M, Gerlinger C, Ruebig A, Schmidt W, Dusterberg B. Open, multicenter comparison of 1,2,3,5,6,10
efficacy, cycle control, and tolerability of a 23-day oral contraceptive regimen with 20 mg ethinyl
estradiol and 75 mg gestodene and a 21-day regimen with 20 mg ethinyl estradiol and 150 mg
desogestrel. Contraception 2001;64:201-7.
Endrikat J, Cronin M, Gerlinger C, Ruebig A, Schmidt W, Dusterberg B. Double-blind, multicenter 1,2,4,5,6,10
comparison of efficacy, cycle control, and tolerability of a 23-day versus a 21-day low-dose oral
contraceptive regimen containing 20 mg ethinyl estradiol and 75 mg gestodene. Contraception
2001;64:99-105.
Endrikat J, Hite R, Bannemerschult R, Gerlinger C, Schmidt,W. Multicenter, comparative study of cycle 1,2,5,10
control, efficacy and tolerability of two low-dose oral contraceptives containing 20 mg ethinylestradiol/
100 mg levonorgestrel and 20 mg ethinylestradiol/500 mg northisterone. Contraception 2001;64;3-10.
Foidart JM. The contraceptive profile of a new oral contraceptive with antimineralocorticoid and 1,2,3,5,10
antiandrogenic effects. Eur J Contracept Reprod Health Care 2000;5(suppl):25-33.
Foidart JM, Wuttke W, Bouw GM, et al. A comparative investigation of contraceptive reliability, cycle 1,2,3,5,10
control and tolerance of two monophasic oral contraceptives containing either drospirenone or
desogestrel. Eur J Contracept Reprod Health Care 2000;5:124-34.
Fotherby K. Clinical experience and pharmacological effects of an oral contraceptive containing 20 mg 1,2,3,4,5,10
oestrogen. Contraception 1992;46:477-88.
Fotherby K. Twelve years of clinical experience with an oral contraceptive containing 30 ug 1,2,3,4,5,10
ethinyloestradiol and 150 mg desogestrel. Contraception 1995;51:3-12.
Gestodene Study Group 322. The safety and contraceptive efficacy of a 24-day low-dose oral contraceptive 1,2,3,5,6,9,10
regimen containing gestodene 60 mg and ethinylestradiol 15 mg. Eur J Contracept Reprod Health Care
1999;4(suppl):9-15.
Golbs S, Domhardt R, Presl J, et al. Clinical findings with the oral contraceptive combination 1,2,5,9,10
ethinylestradiol/dienogest in the Czech Republic. Methods Find Exp Clin Pharmacol 2002;24:689-96.
Granella F, Sances G, Zanferrari C, Costa A, Martignoni E, Manzoni G. Migraine without aura and 1,2,7,11
reproductive life events: a clinical epidemiological study in 1300 women. Headache 1993;33:385-9.
Granella F, Sances G, Pucci RE, Ghiotto N, Nappi G. Migraine with aura and reproductive life events: a case 2,7, 11
control study. Cephalalgia 200;20:701-7.
Grant ECG. Relation of arterioles in the endometrium to headache from oral contraceptives. Lancet 1,2,3,7,11,12
1965;1:1143-4.
Grant ECG. Relation between headaches from oral contraceptives and development of endometrial 1,2,7,11
arterioles. BMJ 1968;3:402-5.
Hampton RM, Short M, Bieber E, et al. Comparison of a novel norgestimate/ethinyl estradiol oral 1,2,3,5,10
contraceptive (Ortho Tri-Cyclen Lo) with the oral contraceptive Loestrin Fe 1/20. Contraception
2001;63:289-95.
Huber J. Clinical experience with a new norgestimate-containing oral contraceptive. Int J Fer- 1,2,3,5,10
til1992;37(suppl):47-53.
Huber J, Foidart JM, Wuttke W, et al. Efficacy and tolerability of a monophasic oral contraceptive 1,2,3,5,10
containing ethinylestradiol and drospirenone. Eur J Contracept Reprod Health Care 2000;5:25-34.
Ismail MT. A prospective study of a monophasic oral contraceptive containing 30 mg ethinyl oestradiol and 1,5,10
150 mg desogestrel (Marvelon). Malays J Reprod Health 1994;12:43-8.
Kamal I, Hefnawi F, Younis N. Lynestrenol 1 mg-mestranol 0.1 mg: a new low dose oral contraceptive: 1,2,3,5,6,10
a clinical study. Egypt Popul Fam Plann Rev 1970;3:11-5.
Kelsch Kappius RE, Goolkasian P. Group and menstrual phase effect in reported headaches among college 1,2,3,5,10,11
students. Headache 1987;27:491-4.
Kaunitz AM. Efficacy, cycle control, and safety of two triphasic oral contraceptives: Cyclessa (Desogestrel/ 1,2,3,5,10
Ethinyl Estrodiol) and Ortho-Novym 7/7/7 (Norethindrone/Ethinyl Estradiol): a randomized clinical
trial. Contraception 2000;61:295-302.
Kirkman RJE, Pedersen JH, Fioretti P, Roberts HE. Clinical comparison of two low-dose oral contraceptives, 1,2,3,5,10
Minulet and Mercilon, in women over 30 years of age. Contraception 1994;49:33-46.
Kudrow L. The relationship of headache frequency to hormone use in migraine. Headache 1975;15:36-49. 1,11
Kuhl H, Gahn G, Romberg G, et al. A randomized cross-over comparison of two low-dose oral contraceptives 2,3,5,10
upon hormonal and metabolic parameters. Contraception 1985;31:583-93.
Lachnit-Fixson U, Aydinlik S, Lehnert J. Clinical comparison between a monophasic preparation and 13
a triphasic preparation. Advances in fertility control and treatment of sterility. Lancaster,
UK: MTP Press; 1984. p. 71-9.
LaGuardia KD, Shangold G, Fisher A, Friedman A, Kafrissen M, and the Norgestimate Study group. Efficacy, 1,2,3,5,10
safety and cycle control of five oral contraceptives regimens containing norgestimate and ethinyl
estradiol. Contraception 2003;67:431-7.
Lammers P, op ten Berg M. Phase III clinical trial with a new oral contraceptive containing 150 mg 6
desogestrel and 20 mg ethinylestradiol. Acta Obstet Gynecol Scand 1991;70:497-500.
Lammers P, Blumenthal PD, Huggins GR. Developments in contraception: A comprehensive review of 8
Desogen (Desogestrel and Ethinyl Estradiol). Contraception 1998;57:1S-27S.
Maitra N, Gulmezoglu AM, Meirik O. Comparison of acceptability of low-dose oral contraceptives 8
containing norethisterone, levonorgestrel, gestodene, desogestrel and norgestimate. In: The Cochrane
Library 2000;4.
MacGregor EA. Hormonal contraception and migraine. J Fam Plann Reprod Health Care 2001;27:49-52. 8
Massiou H, MacGregor EA. Evolution and treatment of migraine with oral contraceptives. Cephalalgia 8
2000;20:170-4.
Merchiers E, Thiery M, Van Kets H, et al. Gestodene: clinical experience with an innovative progestogen 1,2,5,10
used in a combination oral contraceptive preparation. Acta Therapeutica 1990;16:129-37.
Moggia AV. Contraception with levonorgestrel, 0.15 mg, and ethinyl estradiol 0.03 mg. Clinical studies in 1,2,3,4,5,10
Latin America. J Reprod Med 1983;28(suppl):77-80.
Morigi EM, Pasquale SA. Clinical experience with a low dose oral contraceptive containing norethisterone 1,2,3,5,10
and ethinyl oestradiol. Curr Med Res Opin 1978;5:655-62.
Nilsson A, Jacobston L, Ingemanson CA. Side effects of an oral contraceptive with particular attention to 1,2,7,8,11
mental symptoms and sexual adaptation. Acta Obstet Gynecol Scand 1967;46:537-56.
O’Connell EB. A comparative study on oral contraceptives. J Reprod Med 1970;5:71-4. 1,2,3,5,10
Oelkers W, Foidart JM, Dombrovicz N, et al. Effects of a new oral contraceptive containing an 1,2,3,5,10
antimineralocorticoid progestogen, drospirenone, on the rennin-aldosterone system, body weight,
blood pressure, glucose tolerance, and lipid metabolism. J Clin Endocrinol Metab 1995;80:1816-21.
Percival-Smith RKL, Yuzpe AA, Desrosiers JAJ, et al. Cycle control on low-dose oral contraceptives: 1,2,3,10
a comparative trial. Contraception 1990;42:253-62.
Phillips BM. Oral contraceptive drugs and migraine [letter]. BMJ 1968;2:99. 1,2,3,5,7,11
Ramos R, Apelo R, Osteria T, Vilar E. A comparative analysis of three different dose combinations of oral 1,9
contraceptives. Contraception 1989;39:165-79.
Rosenberg MJ, Meyers A, Roy V. Efficacy, cycle control, and side effects of low- and lower-dose oral 1,2,5,10
contraceptives: a randomized trial of 20 mg and 35 mg estrogen preparations. Contraception
2000;60:321-9.
Sartoretto JN, Ortega-Recio JC, Moraes R, Filho FN. Clinical studies with a low dose estrogen-progestogen 1,2,3,9
combination. Contraception. 1977;15:563-70.
Serfaty D, Vree ML. A comparison of the cycle control and tolerability of two ultra low-dose oral 1,3,5,9
contraceptives containing 20 mg ethinylestradiol and either 150 mg desogestrel or 75 mg gestodene.
Eur J Contracept Reprod Health Care 1998;3:179-89.
Sheth A, Jain U, Sharma S, et al. A randomized, double-blind study of two combined and two progestogen- 1,2,3,5,10
only oral contraceptives. Contraception 1982;25:243-52.
Silbergeld S, Brast N, Noble EP. The menstrual cycle: a double-blind study of symptoms, mood and 1,2,5,10
behavior, and biochemical variables using Enovid and placebo. Psychosom Med 1971;33:411-28.
Singh M, Thomas D, Singh R, Saxena BB, Ledger WJ. A triphasic oral contraceptive pill, CTR-05: clinical 1,2,3,5,10
efficacy and safety. Eur J Contracept Reprod Health Care 1996;1:285-92.
Sluglett J, Lawson J. Side-effects of oral contraceptives. Lancet 1967;1:612. 1,8,9
Suthipongse W, Taneepanichskul S. An open-label randomized comparative study of oral contraceptives 1,2,3,5,10,11
between medications containing 3 mg drospirenone/ 30 mg ethinylestradiol and 150 mg levonogestrel/
30 mg ethinylestradiol in Thai women. Contraception 2004;69:23-6.
Tantbirojn P, Taneepanichskul S. Clinical comparative study of oral contraceptives containing 30 mg 1,2,3,5,10
ethinylestradiol/150 mg levonorgestrel, and 35 mg ethinylestradiol/250 mg norgestimate in Thai
women. Contraception 2002;66:401-5.
Taneepanichskul S, Kriengsinyot R, Jaisamrarn U. A comparison of cycle control, efficacy, and side effects 1,2,3,5,10
among healthy Thai women between two low-dose oral contraceptives containing 20 mg ethinylestra-
diol/75 mg gestodene (Meliane) and 30 mg ethinylestradiol/75 mg gestodene (Gynera). Contraception
2002;66:407-9.
Utian WH. Oestrogen, headache and oral contraceptives. South Afr Med J 1974;48:2105-8. 2,3,8,10
van de Walle J, Weijers MJ. Clinical evaluation of a new oral contraceptive, ’Pregnon’. Curr Med Res Opin 1,2,3,6,9,10
1975;3:151-6.
van Trappen Y, Duvivier P, Thiers M. Clinical experience with an ethinylestradiol-desogestrel OC. 1,2,5,10
Arzneimittel Forschung 1989;39:717-9.
Walling M. A multicenter efficacy and safety study of an oral contraceptive. Contraception 1992;46: 1,2,5,10
313-26.
Weber-Diehl F, Lehnert J, Lachnit U. Comparison of two triphasic oral contraceptives containing either 1,2,3,9
gestodene or norethindrone. Contraception 1993;48:291-301.
Weijers MJ. Clinical trial of an oral contraceptive containing desogestrel and ethinyl estradiol. Clin Ther 1,2,3,5,10
1982;4:359-66.
Whitty CWM, Hockaday UM. Migraine: a follow-up study of 92 patients. Br Med J Clin Res 1968;1:735-6. 1,11
Wiseman A, Bowie J, Cogwell E, et al. Marvelon: clinical experience in the UK. Br J Fam Plann 1984;10: 8,13
38-42.
Woutersz TB. A new ultra-low-dose combination oral contraceptive. J Reprod Med 1983;28(suppl):81-4. 1,2,3,5,10
* Reasons for exclusion.
1. No control group.
2. No baseline headache data.
3. Cannot extract data.
4. Duplicate publication.
5. Did not inquire systematically about or capture information on headache.
6. Nonstandard OC regimen.
7. Not prospective.
8. Not a clinical trial.
9. Inadequate information on patient dropout or flow throughout trial.
10. Headache change not a prespecified study outcome.
11. Not an inception cohort.
12. Not in English or not within specified time period.
13. Abstract, book chapter, or other non-journal publication.
www.ajog.org
GENERAL OBSTETRICS AND GYNECOLOGY: GYNECOLOGY
Colposcopic and histopathologic evaluation of women

participating in population-based screening for human
papillomavirus deoxyribonucleic acid persistence
Kristina Elfgren, MD, PhD,a,* Eva Rylander, MD, PhD,b Thomas Rådberg, MD, PhD,c
Björn Strander, MD,d Anders Strand, MD, PhD,e Kirsti Paajanen, MD,f
Inga Sjöberg, MD, PhD,g Walter Ryd, MD, PhD,h Ilvars Silins, MD, PhD,i
Joakim Dillner, MD, PhDi for the Swedescreen Study Group
From the Department of Obstetrics and Gynecology, Karolinska Institutet, Karolinska University Hospital, Huddinge,
Stockholm, Swedena; Department of Obstetrics and Gynecology, Karolinska Institutet, Danderyds University Hospital,
Stockholm, Swedenb; Departments of Gynecology,c and Pathology,h and Onkologiskt Centrum,d Sahlgrenska
University Hospital, Gothenburg, Sweden; Department of Dermato-Venereology, Akademiska Sjukhuset, Uppsala,
Swedene; Department of Obstetrics and Gynecology, MAS University Hospital, Malmö, Swedenf; Department of
Obstetrics and Gynecology, Umeå University Hospital, Umeå, Swedeng; and Department of Medical Microbiology,
Lund University, University Hospital at Malmö, Malmö, Swedeni
Received for publication November 2, 2004; revised January 3, 2005; accepted January 25, 2005
KEY WORDS Objective: Evaluation of colposcopic and histopathological findings in women screened for
Human cervical human papillomavirus deoxyribonucleic acid persistence.
papillomavirus Study design: A total of 12 527 women, aged 32 to 38 years old, attending the population-based
Deoxyribonucleic acid cervical cancer screening program in Sweden were randomized 1:1 to mock testing or human
Colposcopy papillomavirus deoxyribonucleic acid testing by general primer 5C/6C polymerase chain
Cervical cancer reaction and subsequent typing. Human papillomavirus deoxyribonucleic acid–positive women
screening with a normal Papanicolaou smear (n = 341) and an equal number from the control group were
human papillomavirus tested on average 19 months later. One hundred nineteen women with
type-specific human papillomavirus persistence and 111 controls were referred to colposcopy, and
84.8% attended.
Results: Histopathology from colposcopically directed biopsies confirmed cervical intraepithelial
neoplasia grade 2 or 3 in 28 of 100 of the women with human papillomavirus deoxyribonucleic
acid persistence and in 2 of 95 among controls.
Supported by grants from the Swedish Cancer Society, Europe Against Cancer, and the EU Biomed 5 project HPV-Based Cervical Cancer
Screening.
* Reprint requests: Dr. Kristina Elfgren, Department of Obstetrics and Gynecology, K 57, Karolinska University Hospital, Huddinge, S-141 86
Stockholm, Sweden.
E-mail: kristina.elfgren@karolinska.se
doi:10.1016/j.ajog.2005.01.056
Elfgren et al 651
Conclusion: Among women with normal Papanicolaou smear attending population-based

screening, the positive predictive value of human papillomavirus deoxyribonucleic acid
persistence for detection of biopsy-confirmed cervical intraepithelial neoplasia 2 or 3 was 29%.
Carcinoma of the uterine cervix is a major health findings. To address the issue of verification bias as well
issue for women, being the leading cause of death in as the issue of specificity, a control group of women
cancer among women in the developing world.1,2 The referred at random from the same cohort was subjected
most successful preventive strategy has been population- to the same colposcopic assessment in a double-blinded
based Papanicolaou smear screening.3 Management of fashion. To ensure generalizability, the study was per-
women with abnormal Papanicolaou smears includes formed within a randomized intervention trial that is
colposcopy and colposcopy-directed biopsies and surgi- entirely nested within the real-life population-based,
cal removal of histologically verified precancerous invitational screening program in Sweden. We report
lesions (cervical intraepithelial neoplasia [CIN]). the results of the histopathological and colposcopical
However, there are limitations of the Papanicolaou outcome of screening for HPV persistence, compared
smear. The poor specificity for high-grade CIN results in with random referral of women from the general pop-
overtreatment and diagnostic labeling of healthy ulation.
women, whereas the limited sensitivity necessitates re-
peated screening.
Infection with human papillomavirus (HPV) is con-
Material and methods
sidered necessary for the development of the disease.4,5 Study group
The HPV genome from a genital high-risk type is
detected in up to 96.6% of the cervical carcinomas.6 In the organized screening program, women aged 23 to
HPV infections are common, with the highest preva- 50 years are invited by letter for screening at 3-year
lence at 20 to 24 years of age, with change of male sexual intervals. The files of the population registry are first
partners as the main risk factor for acquiring an HPV checked against cytology registries and women who
infection.7 Most HPV infections will clear spontane- have had a Papanicolaou smear taken within the
ously within 5 years.8-10 At the age of 35 years, the previous 18 months are not invited. In the present
prevalence is less than 10% in most populations.11-13 study, the study base was defined as the entire popula-
Persistence of the viral infection is found mainly among tion aged 32 to 38 years resident in 5 different regions in
women infected with high-risk types like HPV 16.14 Sweden (Stockholm, Gothenburg, Malmö, Umeå, and
Development of CIN and eventually cervical cancer will Uppsala), with the following additional inclusion crite-
occur among women with persistent oncogenic HPV ria: not having been sorted out from the organized
infection.15 screening program because of recent opportunistic
Several studies have shown that HPV deoxyribonu- Papanicolaou smear; having responded to the invita-
cleic acid (DNA) testing has a higher sensitivity than the tional letter, and having provided informed consent to
Papanicolaou smear for the identification women with participate in the HPV screening trial. The study was
high-grade CIN.16-19 Testing of HPV DNA has there- approved by the Institutional Review Board of the
fore been suggested as an additional, or possibly stand- Karolinska Institute (decision number 96/305).
alone, improved screening method.20 The specificity of Altogether 12 527 women were enrolled. A complete
HPV DNA testing may be improved by restricting the account of the quality assessment of the enrollment will
screening to women O30 years of age and requiring be the subject of a separate publication. Examples of
repeated HPV DNA positivity (ie ‘‘persistence’’) before incorrect enrollments that were detected and excluded
referral to colposcopy.21 from the study include an illegible or incorrect personal
A major question when considering HPV screening is identification number of the woman, women showing up
whether the clinical management of women referred from for HPV screening without having been invited, and
HPV screening programs should differ from the conven- women moving between cities who were invited twice.
tional management being applied to women referred At enrollment, both a Papanicolaou smear and
from Papanicolaou smear screening programs. However, a brush sample for HPV DNA testing were collected.
few studies have specifically studied the colposcopic The Stockholm, Gothenburg, Uppsala, and Malmö
assessment of women referred from population-based organized screening programs use an endocervical brush
HPV screening programs. To find out whether colpo- sampling for cytology. After obtaining the Papanicolaou
scopic assessment is sensitive in this group of women, we smear, the endocervical brush with cells from both the
took ‘‘blind’’ biopsies in case of normal colposcopic endo- and ectocervix was placed in 1 ml NaCl, frozen at
652 Elfgren et al
ÿ20(C at the enrollment center, shipped frozen, and had an adequate (b-globin PCR positive) second HPV
stored frozen at ÿ80(C at the laboratory. The screening test done. Fourteen women were HPV positive only for
program in Umeå uses a broom type of brush (Cervex other HPV type(s) and were classified as having cleared
Brush, Rovers Medical Devices B.V. Oss, The Nether- the original infection. The 119 women who were positive
lands). A pilot study determined that the most appro- for the same type of HPV DNA in both tests as well as
priate HPV DNA sampling method in that setting was 111 women, selected at random from the 337 control
to take a second sample with the endocervical brush women, were referred to colposcopy (Figure 1). Finally,
after the Cervex sample. Testing for inhomogeneity in 195 of the 230 invited women (84.8%) attended the
HPV DNA results (prevalence and sensitivity) have not colposcopy at 1 of the 5 colposcopy centres. Thirty-three
revealed any differences that could be attributable to women declined participation (18 women with HPV
this difference in sampling strategy. persistence and 15 control women). One woman (case
Women were randomized 1:1 to the intervention arm subject) had been referred to routine colposcopy because
(HPV testing) (N = 6257) or the control arm (sample of a CIN diagnosed between the 2 HPV tests. One
stored frozen without analysis) (N = 6270). The ran- woman (control group) had never had sexual inter-
domization file was made at the Stockholm Regional course. The study flow is summarized in the Figure.
Cancer Registry using computer-generated random num- The cervical samples from the 95 control women,
bers. Randomization results were not released until the taken both at enrollment and at the second sampling
HPV DNA samples had arrived at the virus laboratory. visit, were retrieved and HPV tested and typed by the
The samples from the intervention arm were screened same methods as described above. The HPV tests that
by a general HPV general primer 5C/6C-mediated were performed in the control group had no influence on
polymerase chain reaction (PCR)–enzyme immunoas- the selection of patients for the study.
say.22,23 Samples from Stockholm, Malmö, and Umeå The 195 colposcopies were performed according to
were screened at the regional virus laboratories, whereas a standardized protocol by 8 different gynecologists with
the samples from Uppsala and Gothenburg were sent to specialist training in colposcopy. Both women and
the Malmö laboratory for screening. All positive sam- colposcopists were blinded for the HPV test results.
ples were sent to Malmö for repeat analysis and HPV The protocol included judgment of the transformation
typing by reverse-dot blot hybridization.15 Samples were zone (TZ) for maturation, visibility, and border to the
scored as positive only if confirmed by successful typing columnar epithelium, acetowhitening, and iodine stain-
by reverse-dot blot hybridization.15 ing. First, 5% acetic acid was applied to assist in
Among the 6089 HPV DNA tests that fulfilled the identifying undifferentiated epithelia or inflammation
sample adequacy test (b-globin PCR), 433 were HPV as well as true CIN. Thereafter, 5% potassium iodine
DNA positive (Figure). A control group of 409 women was applied to distinguish the dimension and borderline
was selected at random from the control arm. Ninety- of an abnormal area. Biopsies were taken from all
two of the 433 HPV DNA-positive women and 11 of 409 acetowhite and iodine-negative lesions. If colposcopy
control women were referred to colposcopy because of was considered normal (ie, no acetowhite or iodine-
concomitant abnormal cytological findings. The remain- negative lesions were identified), 2 biopsies were taken
ing 739 women were invited for a second Papanicolaou from an area at 12 o’clock and 6 o’clock on the
smear and a new HPV test at least 12 months later (on ectocervix close to the squamocolumnar junction. All
average 19 months). Repeated contact attempts were acetowhite areas, including metaplasia (undifferentiated
made to nonresponding women for the entire time the epithelium), inflammation, and neoplasia were consid-
study was open. The longest time interval between ered as abnormal colposcopy.
enrollment test and follow-up test was 54 months. The An endocervical cell sample for cytology was taken
randomization code was not revealed to the women or using a cytobrush. Endocervical cytobrush cytology in
the personnel managing the visits and the sampling, combination with ectocervical biopsies has been shown
neither at the second smear nor at the invitation to to have an equal or higher sensitivity than endocervical
colposcopy. Two hundred seventy of 341 (79.2%) of curettage in combination with ectocervical biopsies.24
HPV DNA-positive women and 337 of 398 (84.7%) of The Papanicolaou smears taken at the time of the
the control women attended the second Papanicolaou second HPV test of the 195 women who attended
smear and HPV test. colposcopy were evaluated only after the colposcopies
Reasons for nonparticipation were active refusal (68 had been performed.
women), relocation to city not covered by trial (18 All histopathological specimens underwent routine
women), no response at all to letters and also not histopathological evaluation, which formed the basis of
possible to reach by phone (40 women), pregnancy (4 clinical management of the patients. All histopatholog-
women), other (1 woman). ical specimens were sent for review by a single expert
Type-specific HPV persistence was detected in 119 of reviewer (W. R.) who was blinded to the HPV status
268 (44.4%) of the women in the intervention arm that of the women. When the routine diagnoses and expert
Elfgren et al 653
Figure Selection of patients for the study.

654 Elfgren et al
Table I Colposcopic and histopathologic findings in women participating in population-based screening for HPV DNA persistence
(intervention) or in random referral (control)
Normal colposcopy, N = 100 Abnormal colposcopy,* N = 95
Ectocervical Intervention Control Total Intervention Control Total
histopathology group group (%) group group (%)
Benign 30 53 83 (83.0) 20 26 46 (48.4)
Koilocytosis 1 2 3 (3.0) 6 3 9 (9.5)
Atypia 1 1 2 (2.0) 0 0 0
CIN 1 3 1 4 (4.0) 9 5 14 (14.7)
CIN 2 4 0 4 (4.0) 8 1 9 (9.5)
CIN 3 1 0 1 (1.0) 15 1 16 (16.8)
Missing diagnosis 1 2 3 (3.0) 1 0 1 (1.1)
Total number 41 59 100 59 36 95
* Includes all acetowhite lesions.
colposcopy, blind biopsies revealed 1 case of CIN 3, 4

Table II Ectocervical histopathology after referral because
of HPV persistence, in comparison with random referral cases of CIN 2, 4 cases of CIN 1, and 2 cases of
unspecified atypia (Table I).
Intervention Control group
Of 100 women in the intervention group, biopsies
Ectocervical group: HPV HPV were taken in 98 women. Specimens from 28 of 98 (29%)
histopathology persistence, N (%) N (%) positive contained CIN 2 or 3, whereas 57 of 98 (58%) had
Benign 50 (50) 79 (83.2 %) 5* histopathological diagnoses within normal limits (nor-
Koilocytosis 7 (7.0) 5 (5.3 %) 0 mal, inflammatory, or koilocytosis without atypia)
Atypia 1 (1.0) 1 (1.1 %) 0 (Table II). In the control group, the biopsies showed
CIN 1 12 (12.0) 6 (6.3 %) 0 CIN 3 in 1 woman, CIN 2 in 1, and CIN 1 in 6 women
CIN 2 12 (12.0) 1 (1.1 %) 0 (6.3%). Analysis of the stored consecutive HPV tests in
CIN 3 16 (16) 1 (1.1 %) 1*
the control group revealed that the woman with CIN 3,
Missing 2 (2.0) 2 (2.1 %) 0
diagnosis but none of the other control women with CIN, had had
Total 100 95 a persistent HPV infection (Table II).
The blinded expert review resulted in some changes in
* The woman with CIN 3 in the control group and 1 of 5 HPV-
positive women with benign histopathology in the control group were histopathological diagnoses. According to routine his-
found to have type-specific HPV persistence in both tests; the other 4 topathology, only 22 women in the intervention group
HPV-positive women in the control arm were HPV negative in their and 1 woman in the control group had CIN 2 or 3.
baseline cervical sample. Furthermore, specimens reported to have koilocytosis
was reduced from 42 to 12 after review. Koilocytosis
re-review diagnoses were compared, 15 diagnoses were had no significant association with HPV status, neither
found to be discrepant (differed in grading by more than before nor after expert review (Table II).
1 step). These specimens were re-reviewed by a third Twenty-eight of 90 endocervical cytologies in the
pathologist. intervention group and 1 of 81 in the control group were
abnormal (Table IIIa). One case of CIN 3 and 2 cases of
CIN 2 were detected only by the endocervical cytology
Results and not by the histopathological examination of the
ectocervical biopsies (Tables II, IIIa, and IIIb).
Altogether 95 of 195 (48.7%) of the women who The population-based positive predictive value of
underwent colposcopy had an abnormal colposcopy referral because of HPV persistence for detection of
(including all acetowhite lesions), 59 of 100 (59%) in histopathologically confirmed CIN 2 or 3 among women
the intervention group with persistent HPV infection with a prior negative smear was 29% (28/98 referred
and 36 of 95 (37.9%) in the population-based control women), an 18-fold increased risk, compared with
group (Table I). These lesions predicted CIN 2 or 3 in random referral from the population (relative risk
colposcopy-directed biopsy verified by expert histopath- 18.2; 95% CI 4.0-114.5). The risk of detection of CIN
ological re-review in 23 of 58 (39.7%) of the women in 2 or 3 in HPV-positive women, compared with referral
the intervention group but in only 2 of 36 women in the of HPV-negative women was 34.4 (95% confidence
control group (Table I). Among the women with normal interval 4.8-696.3).
Elfgren et al 655
Table IIIa Endocervical cytology in relation to referral because of HPV persistence or random referral
Population-based controls, N = 95
Endocervical cytology HPV persistence HPV positive
(cytobrush) conclusion N = 100 (%) All (%) at second test
Benign 61 (61.0) 80 (84.2) 3
HPV infection 1 (1.0) 0 0
Atypia 10 (10.0) 1 (1.1) 0
CIN 1 12 (12.0) 0 0
CIN 2 4 (4.0) 0 0
CIN 3 1 (1.0) 0 0
Adenomatous lesions 1 (1.0) 0 0
Missing cytology 10 (10.0) 14 (14.7) 3*
* Two of these 3 women had a persistently positive HPV test.
Table IIIb Combination of endocervical cytology and ectocervical histopathology in relation to referral because of HPV persistence or
random referral
Population based controls, N = 95
Endocervical cytology HPV persistence HPV positive
and/or histology N = 100 (%) All (%) at second test
Benign 42 (42.0) 79 (83.2) 5*
HPV infection 4 (4.0) 5 (5.3) 0
Atypia 4 (4.0) 2 (2.1) 0
CIN 1 17 (17.0) 6 (6.3) 0
CIN 2 14 (14.0) 1 (1.1) 0
CIN 3 17 (17.0) 1 (1.1) 1*
Adenomatous lesions 1 (1.0) 0 0
Missing cytology and histology 1 (1.0) 1 (1.1) 0
* The woman with CIN 3 in the control group and 1 of 5 HPV-positive women with benign histopathology in the control group were found to have
type-specific HPV persistence in both tests; the other 4 HPV-positive women in the control arm were HPV negative in their baseline cervical sample. In
case of different diagnoses in cytology and histopathology, the worst diagnosis has been entered.
The positive predictive value of abnormal colposcopy second sampling occasion were found among the women
for presence of CIN 2 or 3 was 39.7% (23/58 biopsied in the HPV screening group (Table IV).
women) in women referred because of HPV persistence
but only 5% (2/36) among women with abnormal
colposcopy in the control group. Comment
‘‘Blind’’ biopsies among women with normal colpo-
scopy revealed 1 CIN 3 and 4 CIN 2 in the 40 women We found that population-based screening for HPV
referred because of HPV persistence but in none of 57 persistence and subsequent referral to colposcopy has
women referred at random (Table I). a high predictive value (29%) for the presence of CIN 2
In 156 of 195 women who attended colposcopy, the or 3. Even among women with 2 consecutive normal
second Papanicolaou smear taken concomitantly with cytologies, the positive predictive value was 10%,
the second HPV test was normal. Among these, 16 supporting the concept that additional or alternative
(10%) had CIN 2 or 3 in the biopsies (Table IV). All of screening tools may be required. By design, all women
these women belonged to the HPV screening group. with atypical Papanicolaou smears at enrollment into
Therefore, the positive predictive value of HPV DNA the study (whether HPV positive or negative) were taken
persistence for detection of histopathologically con- care of according to routine clinical practice. Our study
firmed high-grade CIN 2C among women who both is therefore performed in the setting of screening for
had a normal smear at baseline and also are cytologi- HPV persistence as an additional screening test on top
cally normal before the colposcopy is 16 of 73 (22.5%). of cytology.
Nine of the 17 women (53%) with cytological signs of Our studies of the clinical management of women
atypia or CIN 1 in cytology had histopathological CIN with screening-detected HPV persistence found that
2 or 3 (Table IV). Almost all the abnormal smears at the ‘‘blind’’ biopsies in women with HPV persistence, but
656 Elfgren et al
Table IV Cytological diagnoses at the Papanicolaou smear taken concomitantly with the second HPV test that preceded the
colposcopy, in comparison with histopathological diagnoses of subsequently taken biopsies
Preceding Papanicolaou HPV screening group Population-based control group

smear diagnosis N (Pap) Histopathological diagnosis N (HP) N (Pap) Histopathological diagnosis N (HP)
Inadequate 4 Benign 4 Benign 4
koilocytosis 1 koilocytosis
CIN 3 3 CIN 3
Benign 71 Benign 39 85 Benign 73
Koilocytosis 4 Koilocytosis 5
Atypi UNS 1 Atypi UNS
CIN 1 9 CIN 1 6
CIN 2 9 CIN 2
CIN 3 7 CIN 3
Missing diagnosis 2 Missing diagnosis 1
Koilocytosis 1 Benign 1
Atypi 9 Benign 4 1 Benign
CIN 1 1 CIN 1
CIN 2 2 CIN 2
CIN 3 2 CIN 3 1
Glandular atypia 1 Benign 1
CIN 1 7 Koilocytosis 2
CIN 1 1
CIN 2 2
2 CIN 3 2
CIN 2 1 CIN 3 1
Adenocarcinoma? 1 Benign 1
N (Pap), number of women with the different cytological diagnoses at the follow-up Papanicolaou smear. The cytological results from 10 of 95 women are
missing. N (HP), number of women with the different subsequent histopathological diagnoses, in relation to the cytological diagnoses of the preceding
Papanicolaou smear. Atypi UNS, atypia.
with a normal colposcopy, resulted in detection of CIN and the nonexistent specificity of the diagnosis found in
3 in 1 of 40, CIN 2 in 4 of 40, and CIN 1 in 3 of 40 the present study indicates that this is appropriate.
women. It is debatable whether this suggests that blind According to the results in our population-based
biopsies may have a place in the clinical management of study of more than 12 000 women, repeat HPV testing in
women with screen-detected HPV persistence. However, primary screening for cervical cancer seems promising
the fact that there were some CIN lesions detected, in because a substantial number of high-grade CIN lesions
conjunction with the known risk for future development that had been missed by cytology were detected.
of high-grade CIN in women with HPV persistence,25 Although screening based on etiology-oriented princi-
emphasizes that women with HPV persistence, but no ples is considered preferable, the fact that at most a few
detectable lesion, need to be closely followed up. percent of initially HPV-infected individuals actually
The TZ was only partially visible in 31% of the develop cervical cancer has resulted in doubts about
colposcopies. With age the TZ extends into the endo- whether HPV-based screening has sufficient specificity.
cervix and hence more often will be out of reach for The current study has found a high positive predictive
colposcopic judgment and biopsy. The fact that 28 of 90 value (29%) of HPV-based screening, even among
women with HPV persistence (31%) had abnormal women with normal Papanicolaou smears believed to
endocervical cytology, only partially overlapping with be a low-risk group for CIN 2 or 3 development.
the results of the ectocervical biopsies, confirms that Several features of the study design that tend to
endocervical cytology should be a part of the clinical improve the positive predictive value should be noted.
management of women referred because of screen- The study was restricted to an age group (32 to 38 years)
detected HPV persistence. with rather low population-based HPV prevalence.13 An
The proportion of women with koilocytosis accord- HPV test with documented high sensitivity and speci-
ing to histopathology was similar among those with ficity24 was used and subjected to extensive interlabor-
HPV persistence and among those randomly referred. atory comparisons of reliability before use.25 Repeatedly
Histopathologic diagnosis of koilocytosis is currently positive HPV tests taken at rather long intervals
classified as a finding within normal limits in Sweden, (average 19 months) were required before referral,
Elfgren et al 657
and, finally, the testing included HPV typing and 11. Cuzick J, Beverley E, Ho L, Terry G, Sapper H, Mielzynska I,
required that repeatedly HPV-positive tests should et al. HPV testing in primary screening of older women. Br J
Cancer 1998;81:554-8.
contain the same HPV type before referral. 12. Jacobs MV, Walboomers JMM, Snijders PJF, Voorhorst FJ,
In conclusion, our results imply that the proposed Verheijen RHM, Fransen-Daalmeijer N, et al. Distribution of 37
HPV screening algorithm has an acceptable specificity mucosotropic HPV types in women with cytologically normal
and positive predictive value and results in detection and cervical smears: the age-related patterns for high-risk and low-risk
treatment of a significant number of women with the types. Int J Cancer 2000;87:221-7.
13. Forslund O, Antonsson A, Edlund K, van den Brule AJC,
established precursor for cervical cancer (CIN 2 or 3), Hansson BG, Meijer CJLM, et al. Population-based type-specific
even among women with a normal Papanicolaou smear. prevalence of high-risk human papillomavirus infection in middle-
aged Swedish women. J Med Virol 2002;66:535-41.
14. Elfgren K, Kalantari M, Moberger B, Hagmar B, Dillner J. A
population-based five-year follow-up study of cervical human
Acknowledgments papillomavirus infection. Am J Obstet Gynecol 2000;183:561-7.
15. Nobbenhuis MAE, Walboomers JMM, Helmerhorst TJM, Ro-
The participants of the Swedescreen study group were: zendaal L, Remmink ans J, Rise EKJ, et al. Relation of human
Ann Kristin Andersson, Ola Forslund, Bengt-Göran papillomavirus status to cervical lesions and consequences for
Hansson, Anna Palmstierna-Bengtsson, Björn Hagmar, cervical-cancer screening. Lancet 1999;354:20-5.
Anders Hjerpe, Bo Johansson, Hilde Larsson, Sven 16. Cuzick J, Sasieni P, Davies P, Adams J, Normand C, Frater A,
Törnberg, Charlotte Wistrand, Karin Edlund, Göran et al. A systematic review of the role of human papillomavirus
(HPV) testing within a cervical screening programme: summary
Wadell, and Margareta Larsson. and conclusions. Br J Cancer 2000;83:561-5.
17. Schiffman M, Herrero R, Hildesheim A, Sherman ME, Bratti M,
Wacholder S, et al. HPV DNA testing in cervical cancer screening.
References JAMA 2000;283:87-93.
18. Clavel C, Bory J-P, Putaud I, Mangeonjean C, Lorenzato M,
1. Parkin M, Pisani P, Ferlay J. Estimates of the world incidence of Nazeyrollas P, et al. Human papillomavirus testing in primary
25 major cancers in 1990. Int J Cancer 1999;80:827-41. screening for the detection of high grade cervical lesions: a study of
2. Pisani P, Parkin M, Bray F, Ferlay J. Estimates of the worldwide 7932 women. Br J Cancer 2001;89:1616-23.
mortality from 25 cancers in 1990. Int J Cancer 1999;83:18-29. 19. Kulasingam SL, Hughes JP, Kiviat NB, Mao C, Weiss NS,
3. Läärä E, Day NE, Hakama M. Trends in mortality from cervical Kuypers JM, et al. Evaluation of human papillomavirus testing
cancer in the Nordic countries: association with organized screen- in primary screening for cervical abnormalities. JAMA
ing programmes. Lancet 1987;30:1247-9. 2002;288:1749-57.
4. Zur Hausen H. Papillomaviruses and cancer: from basic studies to 20. Van Ballegooijen M, van den Akker-van Marle E, Warmerdam
clinical application. Natl Rev Cancer 2002;5:342-50. PG, Meijer CJLM, Walboomers JM, Habbema JD. Present
5. Schiffman MH, Bauer HM, Hoover RN, Glass AG, Cadell DM, evidence on the value of HPV testing for cervical cancer screening:
Rush B, et al. Epidemiological evidence that human papillomavi- a model-based exploration of the (cost-) effectiveness. Br J Cancer
rus causes most cervical intraepithelial neoplasia. J Natl Cancer 1997;76:651-7.
Inst 1993;85:958-63. 21. Dillner J. Can cervical cancer screening programs be improved by
6. Munoz N, Bosch X, de Sanjose S, Herrero R, Castellsague X, Shah incorporating screening for human papillomavirus infection?
K, et al, for the International Agency for Research on Cancer Cancer J 1998;11:272-6.
Multicenter Cervical Cancer Study Group. Epidemiologic classifi- 22. Jacobs MV, Snijders PJ, van den Brule AJ, Helmerhorst TJ, Meijer
cation of human papillomavirus types associated with cervical CJLM, Walboomers JMM. A general primer GP5C/6C mediated
cancer. N Engl J Med 2003;348:518-27. PCR-enzyme immunoassay method for rapid detection of 14 high-
7. Karlsson R, Jonsson M, Edlund K, Evander M, Gustavsson A, risk and 6 low-risk human papillomavirus genotypes in cervical
Boden E, et al. Lifetime number of partners as the only scrapings. J Clin Microbiol 1997;35:791-5.
independent risk factor for human papillomavirus infection: 23. Jacobs MV, Snijders PJ, Voorhorst FJ, Dillner J, Forslund O,
a population-based study. Sex Transm Dis 1995;22:119-27. Johansson B, et al. Reliable high-risk HPV DNA testing by
8. Hildesheim A, Schiffman MH, Gravitt PE, Glass AG, Greer CE, polymerase chain reaction: an intermethod and intramethod
Zhang T, et al. Persistence of type-specific human papillomavirus comparison. J Clin Pathol 1999;52:498-503.
infection among cytologically normal women. J Infect Dis 24. Mogensen ST, Bak M, Dueholm M, Frost L, Knoblauch NO,
1994;169:235-40. Praest J, et al. Cytobrush and endocervical curettage in the
9. Evander M, Edlund K, Gustafsson A, Jonsson M, Karlsson R, diagnosis of dysplasia and malignancy of the uterine cervix. Acta
Rylander E, et al. Human papillomavirus infection is transient in Obstet Gynecol Scand 1997;76:69-73.
young women: a population-based cohort study. J Infect Dis 25. Rozendaal L, Walboomers JM, van der Linden JC, Voohorst FJ,
1995;171:1026-30. Ke P, Helmerhorst TJ, et al. PCR-based high-risk HPV test in
10. Ho GY, Bierman R, Beardsley L, Chang CJ, Burk RD. Natural cervical cancer screening gives objective risk assessment of women
history of cervicovaginal papillomavirus infection in young with cytomorphologically normal cervical smears. Int J Cancer
women. N Engl J Med 1998;338:423-8. 1996;68:766-9.
www.ajog.org
The effect of bright light therapy on depression associated

with premenstrual dysphoric disorder
Catherine Krasnik, PhD,a Victor M. Montori, MD,b,c Gordon H. Guyatt, MD,b
Diane Heels-Ansdell, MSc,b Jason W. Busse, DC,b,* for the Medically
Unexplained Syndromes Study Group
Department of Medical Science, Psychiatry and Behavioural Neuroscience,a Department of Clinical Epidemiology and
Biostatistics,b McMaster University, Hamilton, Ontario, Canada; Department of Medicine, Mayo Clinic College of
Medicine, Rochester, MNc
Received for publication October 17, 2004; revised December 30, 2004; accepted January 25, 2005
KEY WORDS Objective: This systematic review summarizes the evidence from randomized clinical trials of
Premenstrual bright light therapy for treatment of premenstrual dysphoric disorder.
syndrome Study design: The authors performed a systematic review and meta-analysis of randomized
Premenstrual clinical trials. They searched MEDLINE, AMED, CINAHL, Digital Dissertations, EMBASE,
dysphoric disorder and the Cochrane Central Register of Controlled Trials. The main outcome measure was the
Bright light therapy change in depressive symptom scores as measured by the Hamilton Depression Rating Scale and
Meta-analysis the Beck Depression Inventory.
Results: Four crossover trials studying a total of 55 participants met inclusion criteria. Three
trials showed similar results; one fully unblinded trial showed a much larger effect. The pooled
effect size from the random-effects model of the 3 higher quality trials was ÿ0.20 (95% CI ÿ0.48
to 0.07).
Conclusion: The small size of trials and correspondingly wide confidence limits, and
methodologic limitations of the trials, leaves the impact of bright light therapy for relief of
premenstrual depressive symptoms uncertain. The current evidence justifies neither enthusiastic
dissemination nor confident rejection of this therapeutic modality.
Ms Catherine Krasnik is funded by a Canadian Institute of Health Premenstrual syndrome comprises periodic behav-
Research Studentship Award. Dr Victor Montori is a Mayo Founda- ioral and somatic symptoms that include tension,
tion Scholar. Dr Jason Busse is funded by a Canadian Institute of irritability, hyperphagia, hypersomnia, carbohydrate
Health Research Fellowship Award. craving, and dysphoria.1,2 Seventy-five percent of men-
* Reprint requests: Jason W. Busse, CLARITY Research Group, struating women experience mental or somatic symp-
Department of Clinical Epidemiology & Biostatistics, Hamilton
General Hospital, 7 North, Room 727, 237 Barton Street East,
toms during the luteal phase of the menstrual cycle.1-4
Hamilton, Ontario, L8L 2X2, Canada. Six to 8% of North American women (particularly
E-mail: j.busse@utoronto.ca young women5) have symptoms severe enough that they
doi:10.1016/j.ajog.2005.01.055
Krasnik et al 659
impair social or work-related function, resulting in missing information. Two reviewers (CK and JWB)
premenstrual dysphoric disorder (PMDD), according independently assessed study quality by appraising the
to DSM-IV. PMDD can lower quality of life to an method of randomization, concealment of allocation,
extent similar to that of major depression, and may be blinding, and handling of withdrawals.13
associated with an increase in suicidal ideation and
suicide attempts.3,6,7
The current first-line treatment for PMDD is selective Statistical analysis
serotonin reuptake inhibitors.1 These antidepressants
may be intolerable or result in serious adverse effects, For each trial, we computed an effect size for the change in
particularly in adolescents.8-10 Treatments such as bright depressive symptoms using Hedges’ adjusted g method,
light therapy, which may also act to increase brain a computation of Cohen’s d (effect size) that adjusts for
serotonin levels and production, may represent a safe small-sample-size bias.13 This method expresses the
and attractive alternative.11 Although widely promoted difference between treatment arms in terms of the
as an effective intervention in the lay literature, ran- standard deviations for the outcome measure. Following
domized controlled trials of bright light therapy per- Cohen, we considered an effect size of %0.2 as small, of
formed to date have failed to resolve the controversy 0.3 to 0.5 as moderate, and of O0.5 as large.13 For 2
regarding effectiveness.12 To further inform this issue, studies for which we could not obtain the relevant
we conducted a systematic review and meta-analysis to standard deviations,14,15 we imputed these using the
determine the extent to which bright light therapy correlation coefficient estimated from a study for which
reduces depressive symptoms in patients with PMDD. we had complete individual patient data.16 The estimation
of these studies’ effect sizes did not change the interpre-
tation of the pooled estimates across the entire range of
Material and methods values for the correlation coefficient (0.1–0.9). In one
instance, authors reported the treatment effect using the
Search strategy and eligibility criteria odds ratio for a 50% reduction in Hamilton scores.15 We
converted this odds ratio to an effect size.17 We did not
We searched the following electronic databases, from have access to period-specific data, and treated crossover
their inception to September 11, 2004: MEDLINE, studies as if they were parallel design trials, which is
AMED, CINAHL, Digital Dissertations, EMBASE, a conservative assumption.
and the Cochrane Central Register of Controlled Trials. We used the Review Manager software13 to conduct
We used the terms phototherapy, light, light therapy, random-effects meta-analyses. Heterogeneity between
light exposure, premenstrual syndrome, late luteal-phase studies was assessed by using the Q statistic (c2),18 and
dysphoric disorder, and premenstrual dysphoric disor- we used the I2 statistic to quantify inconsistency, the
der. We also searched the reference sections of included proportion of between-study variability that was not due
trials, and contacted experts in the field requesting to chance.19 The a priori potential explanations for
information about unpublished or ongoing trials. We heterogeneity in this meta-analysis that provided the basis
posted a query in the electronic list for members of the for subgroup analyses were: lack of blinding, severity of
Society for Light Treatment and Biological Rhythms depression at baseline, and the bright light therapy dosing
(www.sltbr.org). Eligible studies were placebo-controlled regimen, ie, intensity, duration, and whether participants
randomized trials investigating the effect of bright received bright light therapy during the morning or the
light therapy on depressive symptoms, measured using afternoon. We evaluated the effect of the intervention in
validated instruments, in patients with PMDD. Two these subgroups of trials using a test for interaction.20
reviewers (CK and JWB) independently assessed study Because both the Hamilton Depression Rating Scale
eligibility and achieved perfect agreement. (HDRS) and the Beck Depression Inventory (BDI)
measure the same underlying domain (depression), we
Data extraction and quality assessment pooled these data within trials that reported both
outcomes16,21,22 to improve the precision of each trial’s
Two reviewers (CK and JWB) extracted the relevant weighted average. We calculated a measure of variance
data using a standardized protocol and data collection of these weighted averages by assigning a correlation of
form. Data extracted included the method of random 0.61 between the HDRS and BDI. Correlations between
treatment allocation, concealment of allocation, blind- the BDI and the HDRS have been reported in the
ing of study participants, intensity, duration and timing literature to range from 0.61 to 0.86,23,24 and we elected
of light therapy used, outcome measures, and adverse to use the most conservative estimate. Further, we
effects. Reviewers resolved disagreements through dis- elected to perform sensitivity analyses by assuming
cussion. Two trials lacked sufficient data for analysis, correlations of 1 and 0 in order to investigate the impact
and the authors were contacted directly to acquire the of this approach on our findings.
660 Krasnik et al
Figure The impact of bright light therapy versus control on depression associated with PMDD, assuming a correlation of 0.61
between the HDRS and the BDI.
Results study mentioned adverse effects, which were increased

agitation and eye strain.15 One participant reported
Our search identified 102 articles; 8 published stud- agitation severe enough to cause her to withdraw from
ies14,15,21,22,25-28 and 2 unpublished studies16,29 were re- treatment, and another reduced her light treatment time
trieved for determination of eligibility. A participant in in half. All 4 trials reported that bright light therapy was
The Society for Light Treatment and Biological Rhythms effective in reducing depressive symptoms15,16,21,22; how-
listserve30 provided one unpublished study16; the other ever, these conclusions were based on the differences in
was located through a dissertation database.26 Of the 8 baseline and post-treatment scores within the treatment
studies, 2 were observational studies and did not examine phase rather than the appropriate comparison, end of
depressive symptoms,27,28 and 1 did not make use of treatment vs end of control phase results.
a validated instrument to measure depression.29 This left Assuming a correlation of 0.61 between the HDRS
7 eligible trials, each of which measured depressive and the BDI, the pooled standardized mean difference
symptoms as an outcome, by use of either the HDRS or in depression scores from 4 trials15,16,21,22 was ÿ0.39
the BDI. Four of these studies shared the same partic- (95% CI ÿ0.85 to 0.06); however, heterogeneity was
ipants.14,15,25,26 The most recent trial that reported on all substantial (I2 = 45.3%; c2 = 5.49, P = .14). After
previous participants15 provided data for analysis. excluding the fully unblinded trial,16 the I2 was reduced
All eligible trials lacked an adequate description of to 0% (c2 = 0.34, P =.84), and the estimate of effect
randomization procedures. Allocation concealment was size was ÿ0.20 (95% CI ÿ0.48 to 0.07) (Figure). Our
either not reported15,21,22 or reported as not done.16 sensitivity analyses, assuming extremes in correlations
Only after contacting the authors were we able to clarify between the HDRS and the BDI, did not substantially
blinding and allocation concealment status: only 1 alter these findings.
trial22 blinded participants as well as clinical and
research personnel. Two trials15,21 blinded research Comment
personnel but not participants, and in 1 trial16 neither
the participants nor the research or clinical personnel Evidence from small trials with limited safeguards
were blinded. Only 1 study concealed allocation.22 against bias suggests a small effect of bright light
Three trials16,21,22 specified DSM diagnostic criteria therapy in the treatment of depressive symptoms in
for establishing PMDD among participants. Only 1 women with PMDD, but the imprecise results remain
Krasnik et al 661
consistent with no effect and with a moderate effect. 7. Wittchen HU, Becker E, Lieb R, Krause P. Prevalence, incidence
Clinical trials of PMDD have shown a substantial and stability of premenstrual dysphoric disorder in the community.
Psychol Med 2002;32:119-32.
placebo effect,31 and including only randomized trials 8. Healy D. Lines of evidence on the risks of suicide with selective
strengthens our findings. Our analysis revealed that serotonin reuptake inhibitors. Psychother Psychosom 2003;72:71-9.
claims of superiority in the included trials were based 9. Dimmock PW, Wyatt KM, Jones PW, O’Brien PM. Efficacy of
on changes from baseline during the treatment phase selective serotonin-reuptake inhibitors in premenstrual syndrome:
and not on comparisons of treatment against the control a systematic review. Lancet 2000;356:1131-6.
10. Jureidini JN, Doecke CJ, Mansfield PR, Haby MM, Menkes DB,
phase. This misinterpretation of the data may have Tonkin AL. Efficacy and safety of antidepressants for children and
contributed to the numerous claims in the lay literature adolescents. BMJ 2004;328:879-83.
for the efficacy of bright light therapy in management of 11. Lambert GW, Reid C, Kaye DM, Jennings GL, Esler MD. Effect
PMDD. A recent Internet search using the terms ‘‘bright of sunlight and season on serotonin turnover in the brain. Lancet
light therapy’’ and ‘‘premenstrual syndrome’’ (Google, 2002;360:1840-2.
12. Born L, Steiner M. Current management of premenstrual syn-
Dec 27, 2004) yielded 5310 hits, with the vast majority drome and premenstrual dysphoric disorder. Curr Psychiatry Rep
promoting this therapy, and a number citing the trials 2001;3:463-9.
we evaluated as proof of efficacy (eg, http://www. 13. Cochrane collaboration: Cochrane reviewers’ handbook 4.2.3,
outsidein.co.uk/res_pms.htm). The available evidence, Updated March 2004. Oxford, UK: Cochrane Collaboration; 2004.
while suggesting the possibility of small effects of bright 14. Cerda GM, Parry BL. The effects of bright light therapy on
symptoms of depression, anxiety, and hibernation in patients with
light therapy, is limited by methodologic quality and premenstrual syndrome. J Women Health 1993;3:5-15.
sample size. Larger trials are needed to define what role, 15. Parry BL, Mahan AM, Mostofi N, Klauber MR, Lew GS, Gillin
if any, bright light therapy has in treatment of de- JC. Light therapy of late luteal phase dysphoric disorder: an
pression associated with PMDD. extended study. Am J Psychiatry 1993;150:1417-9.
16. Flory R, Iott-Gluth C. The effects of bright light on the physical
and psychological correlates of premenstrual dysphoric disorder.
Hollins University, Virginia (unpublished).
17. Chinn S. A simple method for converting odds ratios to effect size
Acknowledgments for use in meta-analysis. Stat Med 2000;19:3127-31.
18. DerSimonian R, Laird N. Meta-analysis in clinical trials. Control
The Medically Unexplained Syndromes Study Group is Clin Trials 1986;7:177-88.
composed of the following individuals: Dr Maziar Badii, 19. Higgins JPT, Thompson SG. Quantifying heterogeneity in a meta-
Dr Arthur Barsky, Dr Jason W. Busse, Dr John Dufton, analysis. Stat Med 2002;21:1539-58.
Dr Nelson Greidanus, Dr Gordon H. Guyatt, Ms 20. Altman DG, Bland JM. Interaction revisited: the difference
between two estimates. BMJ 2003;326:219.
Catherine Krasnik, Dr Victor M. Montori, Mr Edward
21. Oprendek ST, Parry BL, Brown S. Differential reduction in
Mills, Dr Roohi Qureshi, Dr Steven Reid, and Dr Ping symptoms of late luteal phase dysphoric disorder as a function
Wu. Catherine Krasnik led the writing of the manuscript of light therapy. J Women Health 1994;3:115-24.
and, with Jason Busse, developed the concept for the 22. Lam RW, Carter D, Misri S, Kuan AJ, Yatham L, Zis AP. A
study. Victor Montori, Jason Busse, and Gordon controlled study of light therapy in women with late luteal phase
dysphoric disorder. Psychiatr Res 1999;86:185-92.
Guyatt contributed to the study design, data analyses,
23. Beck AT, Steer RA, Garbin MG. Psychometric properties of the
and manuscript revisions. All authors approved the final Beck Depression Inventory: twenty-five years of evaluation. Clin
manuscript. Psychol Rev 1988;8:77-100.
24. Groth-Marnat G, editor. The Beck Depression Inventory. The
handbook of psychological assessment. 3rd ed. New York: John
Wiley and Sons; 1997. p. 123-6.
References 25. Parry BL, Berga SL, Kripke DF, Gillin JC. Melatonin and
phototherapy in premenstrual depression. Prog Clin Biol Res
1. Grady-Weliky TA. Premenstrual dysphoric disorder. N Engl J 1990;341B:35-43.
Med 2003;348:433-8. 26. Parry BL, Berga SL, Mostofi N, Sependa PA, Kripke DF, Gillin
2. Freeman EW. Premenstrual syndrome and premenstrual dysphoric JC. Morning versus evening bright light treatment of late luteal
disorder: definitions and diagnosis. Psychoneuroendocrinology phase dysphoric disorder. Am J Psych 1989;146:1215-7.
2003;28:25-37. 27. Parry BL, Hauger R, Lin E, Le Veau B, Mostofi N, Clopton PL,
3. Ross LE, Steiner M. A biopsychosocial approach to premenstrual et al. Neuroendocrine effects of light therapy in late luteal phase
dysphoric disorder. Psych Clin N Am 2003;26:529-46. dysphoric disorder. Biol Psychiatry 1994;36:356-64.
4. Halbreich U. The etiology, biology, and evolving pathology of 28. Parry BL, Udel C, Elliott JA, Berga SL, Klauber MR, Mostofi N,
premenstrual syndromes. Psychoneuroendocrinology 2003; 28: et al. Blunted phase-shift responses to morning bright light in
55-99. premenstrual dysphoric disorder. J Biol Rhythms 1997;12:443-56.
5. Freeman EW, Rickels K, Schweizer E, Ting T. Relationships 29. Schwartz MF. Premenstrual syndrome: clinical application of
between age and symptom severity among women seeking medical phototherapy. Doctoral thesis; 1993.
treatment for premenstrual symptoms. Psychol Med 1995; 25: 30. Society for Bright Light Treatment and Biological Rhythms.
309-15. Available at: www.sltbr.org. Accessed December 30, 2004.
6. Landen M, Eriksson E. How does premenstrual dysphoric 31. Freeman EW, Rickels K. Characteristics of placebo responses in
disorder relate to depression and anxiety disorders? Depress medical treatment of premenstrual syndrome. Am J Psych 1999;
Anxiety 2003;17:122-9. 156:1403-8.
www.ajog.org
Tumor-specific p53 sequences in blood and peritoneal

fluid of women with epithelial ovarian cancer
Elizabeth M. Swisher, MD,a,b,* Melissa Wollan, BS,a,b Sarita M. Mahtani, BS,a
Julia B. Willner, MD,a Rochelle Garcia, MD,d Barbara A. Goff, MD,b
Mary-Claire King, PhDa,c,d
Division of Medical Genetics,a Department of Medicine, Division of Gynecologic Oncology,b Department of

Obstetrics and Gynecology, Department of Genome Sciences,c and Department of Pathology,d University of
Washington, Seattle, WA
Received for publication September 28, 2004; revised December 21, 2004; accepted January 25, 2005
KEY WORDS Objective: Free tumor DNA in body fluids may be an important biomarker. We tested whether
Ovarian cancer tumor-specific mutated p53 DNA can be detected in blood and peritoneal fluid from women with
Tumor DNA epithelial ovarian cancer.
p53 mutations Study design: Sequencing of tumor DNA identified somatic p53 mutations. Free DNA from
Prognosis matched blood or peritoneal fluid was evaluated for the tumor-specific p53 mutation using
Peritoneal fluid a ligase detection reaction.
Results: Sixty-nine of 137 tumors (50%) had p53 mutations. Plasma or serum from 21 (30%) of
the 69 informative cases contained the tumor-specific p53 mutation. Circulating tumor was an
independent predictor of decreased survival in multivariate analysis (P = .02). We detected
tumor DNA in peritoneal fluid in 28 of 30 (93%) cases, including all 6 cases with negative
cytology.
Conclusion: One third of women with ovarian cancer have circulating tumor DNA and an
associated reduced survival. Free tumor DNA can be detected in the majority of peritoneal fluid
samples.
Patients with cancer have higher levels of free DNA

in plasma or serum than do cancer-free individuals, and
levels of plasma DNA appear to be associated with
Supported by a Kimmel Scholar Award from the Sidney Kimmel clinical status.1,2 In 1996, Nawroz et al identified free
Foundation for Cancer Research, and NIH K08CA096610 (to EMS) tumor DNA using tumor-specific microsatellite alter-
and NIH R01CA27632 (to MCK). MCK is an American Cancer ations in one third of patients with head and neck
Society Professor.
cancer.3 Since that time many studies have used a variety
* Reprint requests: Elizabeth M. Swisher, MD, University of
Washington, Department of Obstetrics and Gynecology, 1959 NE
of molecular alterations to identify circulating tumor
Pacific St, Seattle, WA 98195-6460. DNA in various solid tumors.4,5 The relative proportion
E-mail: swishere@u.washington.edu of cancer patients that have free tumor DNA varies
doi:10.1016/j.ajog.2005.01.054
Swisher et al 663
Figure 1 (A) DNA sequencing of p53 exon 8 in tumor DNA from case 155 revealing a G to T missense mutation at nucleotide 797
resulting in a glycine to valine substitution. (B) Ligase detection reaction using oligomers specific for the p53 797 GOT missense
mutation in exon 8. For case 155, normal DNA (N) from lymphocytes and a second negative control sequence (C) does not contain
the target sequence. Paired tumor (T) and plasma (P) DNA both demonstrate this tumor-specific p53 mutation, consistent with the
presence of tumor DNA in plasma.
widely between studies, likely secondary to differences in approved by the Human Subjects Committee of the
tumor types, stage distribution of patients studied, Institutional Review Board. Platinum resistance was
target alterations evaluated, and assays used. The defined as less than a complete response to chemother-
relative proportion of ovarian cancer patients with apy or relapse within 6 months of completing chemo-
tumor-specific circulating DNA is not certain. therapy. Tumors were surgically staged according to the
The identification of tumor DNA in blood or other International Federation of Obstetrics and Gynecology
body fluids is a promising biomarker for a variety of (FIGO) criteria.14
cancers. Identifying tumor-specific molecular alterations DNA was extracted from lymphocytes and 137
in urine, saliva, sputum, and stool can be a noninvasive tumors that were microdissected, if necessary, to attain
diagnostic test for cancer.6-9 Most ovarian cancers a neoplastic cellularity R70%. Plasma, serum, and cell-
spread through local dissemination in the peritoneal free peritoneal fluid DNA was extracted using the
cavity or through local lymphatic channels, and ovarian QIAamp DNA Mini kit (Qiagen, Valencia, Calif) as
cancers differ from other solid tumors in the rarity of previously described.2 Some samples were concentrated
hematogenous metastasis. Conceivably, the differences in a speed vac 2-4!. Ten to 15 mL of plasma DNA was
in spread patterns of ovarian cancer might lessen the used in amplification reactions. Plasma DNA was
likelihood of finding associated circulating tumor DNA quantified after extraction using the PicoGreen dsDNA
compared with other malignancies. Quantification Kit (Molecular Probes, Inc, Eugene, Ore)
p53 mutations are the most common single somatic according to the manufacturer’s instructions. Total
alteration in ovarian cancer, and occur in early as well DNA in plasma was calculated based on 100% extrac-
as advanced staged disease.10,11 Mutations in p53 may tion efficiency.
be a sensitive indicator of the presence of circulating
tumor DNA.12,13 We tested whether tumor-specific p53
mutations could identify circulating tumor DNA in DNA sequencing
women with epithelial ovarian cancer. We evaluated the DNA was polymerase chain reaction (PCR) amplified for
association of tumor DNA in plasma or serum with all p53 coding exons (2-11) and flanking regulatory
patient prognosis. We likewise tested paired peritoneal regions. Primer sequences and PCR conditions are avail-
fluid samples to identify tumor DNA in peritoneal able from the authors. PCR products were purified and
fluids. sequenced using Big Dye Terminator chemistry (Perkin-
Elmer, Boston, Mass), and run on an ABI 3100 DNA
sequencer (Applied Biosystems, Foster City, Calif).
Sequencing data were analyzed using Sequencher soft-
Tissues: DNA extraction and quantitation ware (Gene Codes Corporation, Ann Arbor, Mich). All
chromatograms from tumor and plasma sequences were
Blood and tumor tissue was collected by the University reviewed and compared with those from corresponding
of Washington Gynecologic Oncology Tissue Bank as normal DNA.
664 Swisher et al
Ligase detection reaction Table I Ovarian tumor characteristics and the presence of
For each p53 exon harboring a mutation, nested PCR circulating tumor DNA
products were used in the ligase detection reaction plasma No tumor Tumor
or serum, tumor, peritoneal fluid (if available), and DNA DNA All cases
normal DNA. For each p53 mutation, 2 10-15 base Grade
oligomers were designed containing the mutant sequence 1-2 3 (6%) 2 (10%) 14 (10%)
at the 3 prime end of the 5 prime oligomer. The second 3 45 (94%) 19 (90%) 123 (90%)
oligomer was specific to the sequence just adjacent (3 FIGO stage
prime) to the mutant sequence. The mutation-specific I 2 (4%) 1 (5%) 13 (10%)
II 2 (4%) 0 6 (4%)
oligomer was end labeled with g32P using phosphonu-
III 39 (81%) 16 (76%) 97 (71%)
cleotide kinase (Roche, Indianapolis, Ind). Ligase reac-
IV 5 (11%) 4 (19%) 21 (15%)
tions included 4 mg Salmon sperm DNA (Invitrogen, Surgical cytoreduction*
Carlsbad, Calif), 5 nmol 32P- labeled mutation-specific Optimal (!1 cm) 27 (61%) 12 (57%) 81 (62%)
oligomer, 5 nmol unlabeled adjacent oligomer, 0.25 mmol Suboptimal (R1cm) 17 (39%) 9 (43%) 49 (38%)
dithiothreitol, 12.5 U thermo-stable ligase (Ampligase, Total 48 21 137
Epicentre, Madison, Wis), Ampligase buffer (Epicentre), * Data for volume of residual disease after initial surgery were not
5 mL purified PCR product and water in a 25 mL reaction. available for 7 cases. Adequacy of surgical cytoreduction is assessed by
Ligase reactions were cycled 35 times at 94( for 30 the largest residual tumor diameter at completion of surgery.
seconds and 37( for 10 minutes. Ligase was inactivated
at 99( for 10 minutes. Specific oligomer sequences are or serum samples (30%, Figure 1). Detection of free
available from the authors. Radiolabeled ligated products tumor DNA was similar in plasma (18/61, 30%) and
were separated on a 6% polyacrylamide gel, and detected serum samples (3/8, 38%). Free DNA concentration in
with autoradiography. plasma of women with ovarian cancer ranged from 1 to
Every plasma sample was evaluated by ligase detection 99 ng/mL (median 35 ng/mL), and was similar between
reaction in tandem with a positive control (tumor DNA cases with and without tumor DNA in plasma.
known to contain the mutation) and a negative control All women received primary therapy containing a tax-
(normal DNA from case known not to contain the ane and platinum agent. Table II indicates the clinical
mutation). To test the sensitivity of the assay we diluted characteristics of the tumors relative to the absence or
tumor DNA into normal DNA. We reliably detected the presence of tumor DNA in plasma. There was no
target p53 sequence at 1:100 dilution, making this assay 10 difference between the distribution of stage, grade, or
to 20 times more sensitive than DNA sequencing. adequacy of surgical cytoreduction. One of 3 informative
Statistics FIGO stage I cases demonstrated circulating tumor
DNA. Preoperative serum CA-125 levels were similar
Contingency tables were evaluated with the Fisher exact between cases with and without tumor DNA in plasma.
tests or with chi-square using InStat (Graphpad Soft-
ware, San Diego, Calif). Survival curves were generated Tumor DNA in plasma and serum
according to the Kaplan-Meier method using Prism 4 predicts poor survival
(Graphpad Software). Differences between survival
curves were tested with the log-rank method. Multivar- Overall survival was significantly reduced in cases with
iate analysis was performed using Cox proportional tumor DNA in plasma (log rank test: P = .01, Figure 2,
hazards modeling with Statview (SAS, Cary, NC). median survival 28 vs 56 months). Among the 64 FIGO
stage III and IV cases, surgical cytoreduction
(P = .0004) and stage (P = .06) were additional pre-
Results
dictors of survival. Cox multiple logistic regression of
Tumor-specific variants in p53 sequences stage III and IV cases revealed circulating tumor DNA
(P = .02) and surgical cytoreduction (P = .003), but
Somatic p53 mutations were detected in 69 of 137 not stage (P = .2), as significant in predicting overall
tumors (50%) (Table I). Forty-eight (70%) mutations survival. Circulating tumor DNA was not associated
were missense, occurring exclusively in exons 5-8. with likelihood of complete response to initial chemo-
Twenty-one (30%) mutations were null mutations, con- therapy (77% vs 73%) or platinum resistance (29% vs
sisting of 10 nonsense (14%), 9 deletion (13%), and 47%, P = .23). Two patients with, and no patients
2 splice site (3%) mutations. Twelve (17%) mutations without tumor DNA in plasma developed brain metas-
occurred in exons 4 (n = 7), 9 (n = 2), or 10 (n = 3). tases (P = .09, Fisher exact test, 95%CI 1.2-30.4),
Of the 69 cases with somatic p53 mutations, the exceeding the predicted !1% incidence of brain metas-
tumor-specific p53 sequences were detected in 21 plasma tases in ovarian cancer.15
Swisher et al 665
Table II Detection of tumor DNA in peritoneal fluid in those cases with negative cytology
Tumor ID Fluid source Surgery Grade FIGO stage Histology p53 alteration nucleotide
58 Ascites 1( surgery 3 IIIC Papillary serous 535 COT
131 Wash 1( surgery 2 IA Endometrioid 818 GOA
178 Wash 1( surgery 3 IA Carcinoma (NOS) 976 del31
187 Wash 1( surgery 1 IIIA Borderline serous 742 COT
124 Wash Recurrence 3 IIIC Carsinosarcoma 817 COT
97 Wash Second look 3 IIIC Endometrioid 833 COT
NOS, Not otherwise specified.
Tumor DNA in peritoneal fluid

In 30 cases with somatic p53 mutations, we evaluated
free DNA in peritoneal washings or ascites for the
tumor-specific sequence using the ligase detection re-
action. We identified the tumor-specific p53 sequence in
the peritoneal fluid of 28 of 30 (93%) women with
intraperitoneal ovarian cancer, including all 6 cases with
no malignant cells identified on cytopathology (Table
II), and 22 of 24 cases with malignant cytology. We
identified tumor DNA in the peritoneal fluid of 2 early
stage cases that were limited to a single ovary without
surface involvement and had negative cytology. Overall,
cytology was positive in 24/30 cases (80%), while tumor
DNA was identified in 28/30 (93%). Cytology identified
malignant cells in 1 of 3 FIGO stage I cases, while tumor Figure 2 Kaplan-Meier analysis of overall survival relative
DNA was identified in all 3 stage I cases. All 30 cases to the presence of circulating tumor DNA. Tumor DNA in
were correctly identified by either cytology or the plasma or serum was a significant predictor of decreased
presence of tumor DNA. survival among all 69 informative cases (P = .01, log rank
test), with a median survival of 28 months vs 54 months in
women without tumor DNA in plasma.
Comment
Tumor-derived DNA has been identified in the body cytologic detection of malignant cells. However, the use
fluids of patients with a variety of cancers, including of tumor DNA in peritoneal fluid as a diagnostic tool is
colorectal, head and neck, lung, bladder, kidney, and limited at present by the lack of a p53 mutation in half
prostate.6-8,16 Because the ovary is in direct contact with of tumors and a dearth of other specific somatic DNA
the peritoneal cavity, we hypothesized that the detection sequence alterations in ovarian cancer. More than 80%
of tumor-specific alterations in peritoneal fluid would be of ovarian cancers associated with germline mutations in
a sensitive indicator of disease. Tumor DNA has been BRCA1 or BRCA2 contain p53 mutations.19,20 Thus,
previously identified in cytologically malignant perito- p53 mutations are likely to be more sensitive for
neal fluid samples from women with ovarian cancer, but identification of tumor DNA in women at elevated
rarely in cytologically negative or early stage sam- genetic risk of ovarian cancer.
ples.17,18 Using the p53 ligase detection reaction, we Free tumor DNA in plasma or serum was present in
identified tumor DNA in 28 of 30 peritoneal fluids, one third of women with advanced ovarian cancer, and
including 6 cases that were negative for malignant cells was a strong independent predictor of decreased survival.
as determined by the cytopathologist (Table II). Presence of circulating tumor DNA has also been
Tumor DNA was identified in peritoneal fluid from associated with worse prognosis for patients with mela-
all 3 stage I cases, while cytology identified malignant noma, colorectal, breast, and esophageal cancer.21-24 The
cells in only 1 case. Combining cytology and tumor quantity of total DNA among women with ovarian
DNA analyses correctly identified all 30 cases. Our cancer did not predict the presence of tumor-derived
increased detection rate in peritoneal fluid compared to DNA sequences in plasma. Thus, simply quantifying
previous studies may be secondary to our use of cell-free DNA in plasma does not predict survival, nor does it
instead of cellular peritoneal DNA.18 The detection of substitute for specific assays that identify tumor-derived
free tumor DNA in peritoneal fluid could augment sequences.
666 Swisher et al
We anticipated that the sensitive ligase detection 2. Sozzi G, Conte D, Mariani L, Lo Vullo S, Roz L, Lombardo C,
reaction would detect tumor-specific p53 mutations in et al. Analysis of circulating tumor DNA in plasma at diagnosis
and during follow-up of lung cancer patients. Cancer Res 2001;61:
a greater proportion of cases than DNA sequencing. 4675-8.
However, an identical proportion of plasma samples 3. Nawroz H, Koch W, Anker P, Stroun M, Sidransky D. Micro-
were positive for tumor alterations using each method. satellite alterations in serum DNA of head and neck cancer
The consistency of the proportion of patients with patients. Nat Med 1996;2:1035-7.
circulating tumor DNA, regardless of the assay, and 4. Esteller M, Sanchez-Cespedes M, Rosell R, Sidransky D, Baylin
SB, Herman JG. Detection of aberrant promoter hypermethyla-
the strong association with clinical outcome suggests an tion of tumor suppressor genes in serum DNA from non-small cell
underlying biological difference in ovarian cancers lung cancer patients. Cancer Res 1999;59:67-70.
associated with circulating tumor DNA. 5. Mulcahy HE, Lyautey J, Lederrey C, Qi Chen X, Anker P, Alstead
A recent study identified tumor-specific methylation EM, et al. A prospective study of K-ras mutations in the plasma of
changes in the serum of most women with ovarian pancreatic cancer patients. Clin Cancer Res 1998;4:271-5.
6. Sidransky D, Tokino T, Hamilton SR, Kinzler KW, Levin B,
cancer.18 Methylation-sensitive PCR (MSP) has been Frost P, et al. Identification of ras oncogene mutations in the stool
shown to be highly sensitive, and the differences in of patients with curable colorectal tumors. Science 1992;256:102-5.
assays and target sequences may explain the lower 7. Fliss MS, Usadel H, Caballero OL, Wu L, Buta MR, Eleff SM,
detection rate in this study. However, MSP results et al. Facile detection of mitochondrial DNA mutations in tumors
may be difficult to replicate between laboratories, are and bodily fluids. Science 2000;287:2017-9.
8. Sidransky D, Von Eschenbach A, Tsai YC, Jones P, Summerhayes
more easily contaminated, and may lead to false-positive I, Marshall F, et al. Identification of p53 gene mutations in bladder
results secondary to incomplete conversion of DNA cancers and urine samples. Science 1991;252:706-9.
during bisulfite treatment.25 Further studies are needed 9. Rosas SL, Koch W, da Costa Carvalho MG, Wu L, Califano J,
to compare the detection of tumor-specific methylated Westra W, et al. Promoter hypermethylation patterns of p16,
promoter sequences and somatically mutated sequences O6-methylguanine-DNA-methyltransferase, and death-associated
protein kinase in tumors and saliva of head and neck cancer
with various techniques to optimize identification of patients. Cancer Res 2001;61:939-42.
circulating tumor DNA in women with ovarian cancer. 10. Okamoto A, Sameshima Y, Yokoyama S, Terashima Y, Sugimura
A higher proportion of patients with pancreatic, T, Terada M, et al. Frequent allelic losses and mutations of the p53
colon, and lung cancers have circulating tumor DNA gene in human ovarian cancer. Cancer Res 1991;51:5171-6.
than we found in women with ovarian cancer using 11. Kohler MF, Kerns BJ, Humphrey PA, Marks JR, Bast RC,
Berchuck A. Mutation and overexpression of p53 in early-stage
comparably sensitive assays.4,5,24 Pancreatic, lung, and epithelial ovarian cancer. Obstet Gynecol 1993;81:643-50.
colon cancers have much higher rates of hematogenous 12. Silva JM, Dominguez G, Garcia JM, Gonzalez R, Villanueva MJ,
spread than is seen with ovarian cancer. The lower Navarro F, et al. Presence of tumor DNA in plasma of breast
proportion of women with ovarian cancer with circu- cancer patients: clinicopathological correlations. Cancer Res 1999;
lating tumor DNA in this study may relate to the low 59:3251-6.
13. Hibi K, Robinson CR, Booker S, Wu L, Hamilton SR, Sidransky
rate of hematogenous dissemination seen in ovarian D, et al. Molecular detection of genetic alterations in the serum of
cancer patients. If this is true, understanding the differ- colorectal cancer patients. Cancer Res 1998;58:1405-7.
ences between those cases with and without circulating 14. Pecorelli S, Benedet JL, Creasman WT, Shepherd JH. FIGO
ovarian tumor DNA may provide insight into the staging of gynecologic cancer. 1994-1997 FIGO Committee on
unusual spread patterns of epithelial ovarian cancer. Gynecologic Oncology. International Federation of Gynecology
and Obstetrics. Int J Gynaecol Obstet 1999;65:243-9.
The relatively high incidence of brain metastases in our 15. Cohen ZR, Suki D, Weinberg JS, Marmor E, Lang FF,
small set of patients with circulating tumor DNA Gershenson DM, et al. Brain metastases in patients with ovarian
supports the hypothesis that circulating tumor DNA is carcinoma: prognostic factors and outcome. J Neurooncol 2004;66:
associated with a more aggressive tumor biology and an 313-25.
increased risk of hematogeneous spread. 16. Cairns P, Esteller M, Herman JG, Schoenberg M, Jeronimo C,
Sanchez-Cespedes M, et al. Molecular detection of prostate cancer
Tumor DNA in plasma is a significant, independent in urine by GSTP1 hypermethylation. Clin Cancer Res 2001; 7:
predictor of poor prognosis, and is associated with more 2727-30.
aggressive tumor behavior. The identification of free 17. Parrella P, Zangen R, Sidransky D, Nicol T. Molecular analysis of
tumor DNA in peritoneal fluid compares favorably to peritoneal fluid in ovarian cancer patients. Mod Pathol 2003;16:
cytopathologic evaluation, and warrants further inves- 636-40.
18. Ibanez de Caceres I, Battagli C, Esteller M, Herman JG, Dulaimi
tigation as a diagnostic marker. Free tumor DNA in E, Edelson MI, et al. Tumor cell-specific BRCA1 and RASSF1A
blood or peritoneal fluid may represent an important hypermethylation in serum, plasma, and peritoneal fluid from
new biomarker in ovarian cancer. ovarian cancer patients. Cancer Res 2004;64:6476-81.
19. Buller RE, Lallas TA, Shahin MS, Sood AK, Hatterman-Zogg M,
Anderson B, et al. The p53 mutational spectrum associated with
References BRCA1 mutant ovarian cancer. Clin Cancer Res 2001;7:831-8.
20. Rhei E, Bogomolniy F, Federici MG, Maresco DL, Offit K,
1. Leon SA, Shapiro B, Sklaroff DM, Yaros MJ. Free DNA in the Robson ME, et al. Molecular genetic characterization of BRCA1-
serum of cancer patients and the effect of therapy. Cancer Res and BRCA2-linked hereditary ovarian cancers. Cancer Res 1998;
1977;37:646-50. 58:3193-6.
Swisher et al 667
21. Muller HM, Widschwendter A, Fiegl H, Ivarsson L, Goebel G, markers in the plasma of melanoma patients. Cancer Res 2001;
Perkmann E, et al. DNA methylation in serum of breast cancer 61:5723-6.
patients: an independent prognostic marker. Cancer Res 2003;63: 24. Lecomte T, Berger A, Zinzindohoue F, Micard S, Landi B, Blons
7641-5. H, et al. Detection of free-circulating tumor-associated DNA in
22. Kawakami K, Brabender J, Lord RV, Groshen S, Greenwald BD, plasma of colorectal cancer patients and its association with
Krasna MJ, et al. Hypermethylated APC DNA in plasma and prognosis. Int J Cancer 2002;100:542-8.
prognosis of patients with esophageal adenocarcinoma. J Natl 25. Sasaki M, Anast J, Bassett W, Kawakami T, Sakuragi N, Dahiya
Cancer Inst 2000;92:1805-11. R. Bisulfite conversion-specific and methylation-specific PCR:
23. Taback B, Fujiwara Y, Wang HJ, Foshag LJ, Morton DL, a sensitive technique for accurate evaluation of CpG methylation.
Hoon DS. Prognostic significance of circulating microsatellite Biochem Biophys Res Commun 2003;309:305-9.
www.ajog.org
Melanoma, thyroid, cervical, and colon cancer

risk after use of fertility drugs
Michelle D. Althuis, PhD,a Bert Scoccia, MD,b Emmet J. Lamb, MD,c
Kamran S. Moghissi, MD,d Carolyn L. Westhoff, MD,e Jerome E. Mabie, BS,f
Louise A. Brinton, PhDa,*
Division of Cancer Epidemiology and Genetics, National Cancer Institute,a Bethesda, MD; University of Illinois,b
Chicago, IL; Stanford University,c Stanford, CA; Wayne State University,d Detroit, MI, Columbia University,e
New York, NY; Information Management Services, Inc,f Rockville, MD
Received for publication December 9, 2004; accepted January 28, 2005
KEY WORDS Objective: This study was undertaken to evaluate melanoma, thyroid, colon, and cervical cancer
Ovulation induction risks after clomiphene or gonadotropins.
Clomiphene Study design: Retrospective cohort of 8422 women (155,527 women-years) evaluated for
Gonadotropins infertility (1965-1988). Through 1999, cancers were ascertained by questionnaire, cancer and
Cancer death registries. Poisson regression estimated adjusted rate ratios (RRs).
Infertility Results: Clomiphene use did not significantly increase risk of melanoma (RR = 1.66; 95% CI,
0.9-3.1), thyroid (RR = 1.42; 95% CI, 0.5-3.7), cervical (RR = 1.61; 95% CI, 0.5-4.7), or colon
cancer (RR = 0.83; 95% CI, 0.4-1.9). We found no relationship between clomiphene dose or
cycles of use and cancer risk at any site. Clomiphene use may impart stronger effects on risks of
melanoma (RR = 2.00; 95% CI, 0.9-4.6) and thyroid cancer among women who remained
nulliparous (RR = 4.23; 95% CI, 1.0-17.1). Gonadotropins did not increase cancer risk for these
sites.
Conclusion: Fertility drugs do not appear to have strong effects on these cancers. Nonetheless,
follow-up should be pursued to assess long-term risks and to monitor effects among women who
remain nulliparous.
Cumulative exposure to reproductive hormones has reported that clomiphene citrate increases uterine cancer
been shown to increase risk of hormone-sensitive carci- risk in a dose-response and time-dependent manner,3
nomas, particularly those of the endometrium, breast, and that fertility drug use may also be associated with
and ovary that have distinct age-specific rate patterns slight increases in breast and ovarian tumors after long
that are attenuated after menopause.1,2 We recently latency periods.4,5
Links between hormonal factors and melanoma,6
colon cancer,7 thyroid cancer,8 and cervical cancer,9
* Reprint requests: Louise Brinton, National Institutes of Health,
National Cancer Institute, 6120 Executive Boulevard, EPS 7068,
although considerably weaker, have also been suggested
Bethesda, MD 20892-7234. based primarily on purported associations with parity or
E-mail: brintonl@exchange.nih.gov exogenous hormone usage. Few reports have assessed
doi:10.1016/j.ajog.2005.01.091
Althuis et al 669
the risk of these cancers associated with fertility treat- hormones; and anthropometric factors. For patients for
ments among infertile women, with small numbers of whom we were unable to obtain questionnaire data, we
cancers and inconsistent results.10-13 This report sum- had accurate location information on cancer status
marizes melanoma, thyroid, cervical, and colon cancer through clinic records (216), or cancer registries (2347)
risk among a large cohort of women from different if the women last resided in California, Florida, Illinois,
clinical sites in the United States. Strengths of the study Massachusetts, Minnesota, New Jersey, New York, and
included nearly 20 years of average follow-up and Texas (eg, the states in which the majority of patients
information on other predictors of cancer risk, including were last known to reside). We attempted to medically
specific causes of infertility and reproductive status verify cancers reported in the questionnaires by obtain-
through follow-up. ing discharge summaries, operative reports, and pathol-
ogy reports from the institutions where the diseases
had been diagnosed and/or treated. Eleven of the self-
Materials and methods reported melanomas and 1 colon cancer were found on
medical record review to be benign and were excluded
Brinton et al5 previously described this retrospective from analytic consideration.
cohort study, which was conducted at 5 large reproduc-
tive endocrinology practices in the following metropol-
itan areas: Boston, Mass; New York City, NY; Chicago, Statistical methods
Ill; Detroit, Mich; and the San Francisco Bay Area,
Calif. The institutional review boards at the collaborat- Person-years accrual began 1 year after clinic registra-
ing centers as well as at the National Cancer Institute tion and continued through the earliest date of cancer
(NCI) approved the study protocol. Briefly, eligible diagnosis, death, or date last known alive and free of
patients were evaluated for infertility between 1965 and cancer (as indicated by the last clinic visit, questionnaire
1988 at participating centers where they were seen more completion, or linkage against cancer registry data).
than once, or if seen only once, were referred by another Patients having cancer registry searches had variable
physician who provided substantive medical informa- study ending dates, depending on the completeness of
tion. Patients with either primary or secondary infertility registration, which ranged from the end of 1997 to 1999.
(nulligravid and gravid, respectively) were eligible for the Otherwise, December 31, 1999, defined the end of the
study, whereas those who were evaluated for reversal of study period. Patients lost to follow-up after their initial
a tubal ligation were not. Medical records for 12,193 clinic visit, those who denied access to their records, and
eligible women were abstracted for information pertain- 10 woman who had cancer diagnosed within 1 year of
ing to all procedures and tests (allowing a determination their registration clinic visit were excluded, leaving 8,422
of different causes of infertility), medications prescribed analytic study subjects and 155,527 person-years
(including clomiphene citrate and a variety of human (mean = 18.8 years) of follow-up.
gonadotropins, namely, Pergonal, Humegon, or Metro- We used 2 analytic approaches to assess cancer risk
din), menstrual and reproductive histories, and other among the cohort members. We first calculated stan-
factors that might affect health status. dardized incidence ratios (SIRs) and 95% CI comparing
A total of 9751 (80.0%) of the patients were success- cancer rates of infertile women with those of US women.
fully traced with the use of several sources, including SIRs were computed as the number of observed cancer
clinic records, telephone directories, credit bureaus, events divided by the expected number of events that were
postmasters, motor vehicle administration records, and based on age, race, and calendar year-specific incidence
the National Death Index. A total of 1319 of the eligible disease rates for women from cancer registry rates avail-
women (10.8%) indicated on contact that they did not able through the Surveillance Epidemiology and End
want to participate in the study. For these women, we Results (SEER) Program of the NCI. The SEER pro-
retained only information on calendar year and age at gram has population-based catchment areas and is widely
study entry, and race. used to estimate cancer burden in the United States.
For the patients traced as alive (we identified 272 The second analytic approach involved analyses
patients as deceased), clinic records, completed ques- within the cohort of infertile women, which allowed
tionnaires, and cancer registries provided information on multivariate adjustment for potential confounding fac-
the development of cancers. We mailed questionnaires to tors. Rate ratios (RRs) and their 95% CIs for developing
patients beginning in 1998, with telephone follow-up cancer associated with administration of ovulation-
attempted for nonrespondents. A total of 5597 of the stimulating drugs (ever use, total dosage, cycles prescribed,
patients completed the questionnaire that ascertained interval since first use) as compared with nonusers were
information on sociodemographic factors, updated estimated by Poisson regression with the use of standard
health status, and lifestyle factors, including menstrual, methods.14 For all analyses, the RRs were adjusted for
pregnancy, and breastfeeding history; use of exogenous study site, age at follow-up (!40, 40-49, 50C), and
670 Althuis et al
as well as colon cancer (SIR = 1.76) and melanoma

Table I Standardized incidence ratios comparing cancer risk
among infertile women to the general population (SIR = 1.57). Thyroid (SIR = 0.99) and cervical cancer
risk (SIR = 0.61) were the only malignancies postulated
Observed Expected SIR 95% CI
to have a hormonal cause that were not more frequently
All sites, excluding 581 473.8 1.23 1.1-1.3 diagnosed among infertile women in the study cohort
nonmelanotic skin compared with the general population. Infertile women
Hormone-sensitive did not have a higher risk of cancers at other sites, when
cancers
assessed individually or in aggregate (data not shown).15
Uterine corpus 39 24.9 1.57 1.1-2.1
Breast 292 226.5 1.29 1.2-1.4
Clomiphene-exposed women were at higher risk of
Ovary 45 22.7 1.98 1.4-2.7 tumors at only 2 sites: uterus (SIR = 2.14; 95% CI, 1.3-
Postulated hormone- 3.3 discussed in detail in Althuis et al3) and melanoma
sensitive cancers (SIR = 2.00; 95% CI, 1.3-3.1). Risk among women not
Colon 28 15.9 1.76 1.2-2.6 exposed to clomiphene was similar to the general pop-
Melanoma 42 26.7 1.57 1.1-2.1 ulation for both of these cancer sites, with SIRs of 1.25
Thyroid 18 18.1 0.99 0.6-1.6 (0.8-1.9) and 1.28 (0.8-2.0), respectively. We found no
Cervical 14 23.0 0.61 0.3-1.0 difference in risk among ‘‘ever used’’ compared with
SIRs were computed as the number of observed cancer events divided ‘‘never used’’ clomiphene users for the remaining cancer
by the expected number of events based on age, race, and calendar sites. In addition, cancer risk did not vary by gonado-
year-specific incidence disease rates for women from cancer registry tropin use when comparing cohort members were com-
rates available through the SEER Program.
pared with the US population (data not shown).
calendar year of follow-up (before 1980, 1980-1989, Internal analyses of cancer risk
1990, or later), and gravidity at entry. Factors that
were available from the medical records, such as cause of To assess the influence of ovulation-stimulating drugs
infertility, smoking history, and body mass at entry, on cancer risk while adjusting for other predictors of
were included in the regression models, as necessary, to cancer risk (including gravidity), subsequent analyses
evaluate their roles as potential confounding or modi- were within the population of infertile women, compar-
fying factors. For the subjects who completed the ing drug users with nonusers. We found little evidence
questionnaire, we evaluated additional predictors of to suggest that melanoma, thyroid, cervical, or colon
risk, such as parity and gravidity at follow-up and cancer risk was increased by ovulation-stimulating drugs
hormone replacement use. within the cohort of infertile women (Table II). Al-
though point estimates for melanoma (RR = 1.66; 95%
CI, 0.9-3.1), thyroid cancer (RR = 1.42; 95% CI, 0.5-
Results 3.7), and cervical cancer risk (RR = 1.61; 95% CI, 0.5-
Description of subjects included in analysis 4.7) were modestly elevated among clomiphene users,
these findings were not statistically significant and there
The median year of first evaluation was 1978 and the was no dose-response with more detailed parameters of
median age of the study subjects at first evaluation was 30 drug usage (dosage and cycles) for any site. Clomiphene
years. Nearly 80% of the subjects were known to be white use appeared to impart stronger effects on risks of
and 43% were evaluated for primary infertility. A total of melanoma, thyroid, and cervical cancer among women
3276 (39%) of the study subjects were prescribed clomi- who were followed for 15 or more years, with RRs
phene to treat their infertility, whereas 865 (10%) received relative to never users of 2.08 (0.9-4.9), 1.54 (0.3-8.0),
gonadotropins. Subjects included in the analyses and and 2.67 (0.6-11.9), respectively. However, these findings
those excluded were not significantly different according were based on small numbers and did not reach statis-
to calendar year and age at first evaluation; however, a tical significance. Risk associated with ever compared
larger proportion of the subjects excluded from analysis with never clomiphene use was 0.83 for colon cancer,
had missing information on race (30% vs 11%).5 with no increase with dose, cycles of use, or latency. The
models for these data were stable, with little difference in
SIRs analysis of cancer risk estimates after adjustment for potential confound-
ing factors or when we restricted our outcomes to
Infertile study subjects developed cancers at higher rates medically validated cases (that is, 11 melanoma [high
than women in the general SEER population (SIR = 1.23; proportion were based on self-report], 13 thyroid, 6
95% CI, 1.1-1.3) (Table I). Elevation in risk was evident cervical, and 24 colon cancers).
for tumors that are most well-recognized as having Although limited by small numbers of events, we
hormonal causes, namely, cancers of the uterine corpus assessed if the relationship between clomiphene use and
(SIR = 1.57), breast (SIR = 1.29), and ovary (SIR = 1.98), cancers varied according to other predictors of cancer
Althuis et al 671
Table II Ovulation stimulating drug use and cancer risk among infertile women
Melanoma (n = 42) Thyroid (n = 18) Cervical (n = 14) Colon* (n = 28)
Infertility treatment n RR 95% CI n RR 95% CI n RR 95% CI n RR 95% CI
Clomiphene
Never 21 1.00 10 1.00 7 1.00 20 1.00
Ever 21 1.66 0.9-3.1 8 1.42 0.5-3.7 7 1.61 0.5-4.7 8 0.83 0.4-1.9
Dosage (mg)
1-900 8 1.78 0.8-4.1 5 2.48 0.8-7.4 3 1.85 0.5-7.4 2 0.56 0.1-2.4
901-2250 9 2.23 10.-5.0 1 0.55 0.1-4.4 2 1.42 0.3-7.0 4 1.22 0.4-3.6
O2250 4 0.95 0.3-2.9 2 1.07 0.2-5.1 2 1.48 0.3-7.6 2 0.73 0.2-3.2
Cycles
!6 16 1.88 0.9-3.7 8 2.14 0.8-5.6 5 1.68 0.5-5.4 5 0.74 0.3-2.0
6C 5 1.20 0.4-3.3 0 d 2 1.43 0.3-7.3 3 1.04 3.0-3.6
Year since first use
!15 11 1.48 0.7-3.2 6 1.55 0.5-4.4 4 1.37 0.4-5.0 4 0.82 0.3-2.5
15C 9 2.08 0.9-4.9 2 1.54 0.3-8.0 3 2.67 0.6-11.9 4 0.98 0.3-3.0
Missing 1
Gonadotropins
Never 38 1.00 16 1.00 12 1.00 28
Ever 4 0.90 0.3-2.6 2 1.10 0.2-4.9 2 1.39 0.3-6.4 0
All models adjusted for attained age, calendar time, study sites, and gravidity at entry. Estimates of risk associated with clomiphene were also adjusted
for anovulatory disorders, a primary indication for use. Additional adjustment for other causes of infertility did not appreciably change risk estimates.
* Additionally adjusted for body mass index at study entry (quartiles, kg/m2).
risk (Table III). Melanoma (RR = 1.72), thyroid cancer cancers diagnosed among infertile women are attribut-
(RR = 1.83), and cervical cancer risk (RR = 2.89) as- able to underlying causes of infertility, to the women’s
sociated with clomiphene usage was highest among low parity, or to fertility drugs has been unclear. We
women who were nulligravid at entry when compared previously reported the possible association of fertility
with unexposed gravid women. For melanoma treatment with hormone sensitive cancers of the breast,
(RR = 2.00) and thyroid cancers (RR = 4.23), risk ovary, and uterus in this cohort.3-5 This report examines
was also elevated among women who remained nullip- risks related to melanoma, thyroid, cervical, and colon
arous at follow-up, with no elevation among parous cancers, which have been suggested as having possible
women. Cervical cancer risk associated with clomiphene hormonal causes. Consistent with previous investiga-
use was not modified by parity at follow-up. Interpre- tions,11-13,16-23 infertile women in our study were at
tation of these findings is complicated by small numbers elevated risk of melanoma (SIR = 1.57) and colon
of cancer and missing information on parity at follow- cancer (SIR = 1.76), but not thyroid (SIR = 0.99) or
up that was available primarily from women who cervical cancer (SIR = 0.61).
completed the questionnaire. Colon cancer risk associ- In addition to comparisons with the general popula-
ated with clomiphene usage was not modified by either tion, we were able to make comparisons within the
gravidity at entry or parity at follow-up. In addition, the population of infertile patients, allowing us to attempt to
relationship between clomiphene use and cancer risk at disentangle the effects of fertility drugs with other
these 4 sites was not significantly modified by specific predictors of cancer risk, such as parity and specific
causes of infertility (such as anovulatory disorders or causes of infertility. It is reassuring that we found little
endometriosis) or attained age (data not shown). evidence to suggest an association between ovulation-
Analysis of the relationship between gonadotropins stimulating drug use and cancer risk, and that our
and cancer risk was limited by the small number of findings changed little after adjustment for potential
exposed women (Table II). Fewer than 5 cases at each confounding factors such as reproductive and smoking
site were exposed to gonadotropins, with no apparent history, cause of infertility, or body mass. There was no
elevation in cancer risk. significant association overall or with increasing drug
use for melanoma (RR = 1.60, based on 21 exposed
Comment cases), thyroid (RR = 1.42, based on 8 exposed cases),
cervical (RR = 1.61, based on 7 exposed cases), or colon
Infertile women, as shown in this study and other cohort cancer (RR = 0.83, based on 8 exposed cancers). We do,
investigations,16-19 are at higher risk of cancer than however, present some evidence suggesting that mela-
women from the general population. Whether excess noma, thyroid, and cervical cancer risk may increase
672 Althuis et al
Table III Modification of cancer risk associated with ever compared with never clomiphene use by reproductive status
Melanoma (n = 42) Thyroid cancer (n = 18) Cervical cancer (n = 14) Colon cancer (n = 28)
Cancers RR 95% CI Cancers RR 95% CI Cancers RR 95% CI Cancers RR 95% CI
Gravidity at entry
Gravid
No clomiphene 13 1.00 5 1.00 3 1.00 12 1.00
Clomiphene 12 1.49 0.7-3.3 4 1.20 0.3-4.5 3 1.42 0.3-7.1 4 0.55 0.2-1.7
Nulligravid
No clomiphene 8 0.78 0.3-1.9 5 1.26 0.4-4.4 4 1.63 0.4-7.3 8 0.84 0.3-2.1
Clomiphene 9 1.72 0.7-4.1 4 1.83 0.5-6.9 4 2.89 0.6-13.1 4 0.83 0.3-2.6
Parity at follow-up
Parous
No clomiphene 18 1.00 4 1.00 4 1.00 7 1.00
Clomiphene 11 0.96 0.5-2.1 2 0.73 0.1-4.0 4 1.43 0.4-5.8 4 0.94 0.3-3.2
Nulliparous
No clomiphene 3 0.45 0.1-1.5 2 1.35 0.2-7.4 1 0.42 0.1-4.2 2 0.76 0.2-3.6
Clomiphene 8 2.00 0.9-4.6 4 4.23 1.0-17.1 2 1.31 0.2-8.2 3 1.95 0.5-7.6
Parity missing 2 6 3 12
Models estimated cancer risk associated with ever compared with never clomiphene use adjusted for attained age, calendar time, and study site.
with time because clomiphene usage with RRs of 2.08 among women who had used clomiphene for 12 or more
(0.9-4.9), 1.54 (0.3-8.0), and 2.67 (0.6-11.9), respectively, menstrual cycles (RR = 2.2; 95% CI, 0.5-10.2, based on
among women followed for 15 years or more. These 4 exposed cases, 3 of whom had ovulatory abnormali-
hints of a possible latency effect may suggest that ties).23 Consistent with our results, 2 other prior studies
clomiphene is an initiator of carcinogenesis and is that evaluated the relationship between ovulation-stim-
consistent with the fact that carcinogenesis is a long ulating agents and melanoma found no increase in
process, which takes many years. Accrual of more cancer cancer risk,23,27 although only 1 (a case-cohort design)
events via continued follow-up of this and other infertile was able to examine the effect of specific fertility drugs.27
cohorts is necessary for clarification. Thyroid cancer has been postulated to have a hor-
We also found that clomiphene use appeared to monal cause because it is 3 times more frequent in
impart stronger effects on the risks of melanoma women than men and because risk has been linked to
(RR = 2.00) and thyroid cancer (RR = 4.23) among oral contraceptive use10 and reproductive factors, par-
women whom remained nulliparous through follow-up, ticularly among those diagnosed at young ages.8 To
a finding that is based on few cancers and that also date, little evidence of an association between ovulation-
requires confirmation. Clomiphene-associated cancer stimulating agents and thyroid cancer risk has emerged.
risk among nulliparous women persisted after adjust- A pooled-analysis of case-control studies reported a
ment for underlying causes of infertility such as anov- nonsignificant increased risk of thyroid cancer after
ulatory disorders, which have been associated with infertility drug use (odd ratio [OR] = 1.6; 95% CI,
melanoma,12,15 and endometriosis, which has been as- 0.9-2.9),10 although 1 of the included studies reported a
sociated with melanoma, thyroid, and colon cancers.15 significant 4-fold excess risk.28 Unlike the case-control
We cannot entirely exclude the possibility that women design that relies on patient reports of complex infertil-
who remained nulliparous and who used clomiphene ity treatment, the cohort design enables obtaining
had more severe underlying disease for which we were information directly from medical records. This allows
unable to account. for more detailed assessment of specific agents and
Early studies linking parity, oral contraceptives, and doses. Cohort studies before ours have not evaluated
postmenopausal hormone therapy to melanoma postu- thyroid cancer risk after use of ovulation-stimulating
lated a hormonal cause for this cancer.24 A relationship agents because of the few cancers diagnosed.
between ovulation-stimulating drugs and melanoma was Although cervical cancer is not generally viewed as a
first suggested by case reports25,26 and by elevations in hormonally-related tumor, relationships of the disease
melanoma risk seen among cohorts of infertile with increasing parity and long duration of oral contra-
women.11,12,16,17,23 For the latter, too few cancers have ceptive use9 have raised concerns regarding effects of
been diagnosed to disentangle the effects of infertility other hormonal agents. The most informative data
medications from their indications for usage. Of the derive from a retrospective cohort study by Rossing
larger investigations, a cohort study in Seattle found no et al conducted in Seattle in which 36 in situ and invasive
elevation in melanoma risk overall, but risk was elevated cervical cancers were detected.13 In the current study,
Althuis et al 673
which is in line with other studies that have shown that melanoma, thyroid, cervical, or colon cancers. None-
parity is a risk factor for this cancer,9 infertile women theless, because clomiphene is one of the most widely
were at a decreased risk of having cervical cancer used drugs in the management of infertility,30 extended
develop compared with the general population. Con- follow-up of infertile patients is necessary to accrue
trary to the study by Rossing et al, which reported a more cancer events and to clarify the relationship
reduction in cervical cancer risk associated with clomi- between clomiphene citrate and cancer risk, particu-
phene (RR = 0.4), we found the risk among women who larly among infertile women who remain nulliparous.
had taken clomiphene was elevated relative to nonusers
(RR = 1.61; 95% CI, 0.5-4.7). Neither our study nor the
study by Rossing et al reported any apparent relation
according to dose or duration of use. It remains unclear References
as to whether clomiphene may act predominately as an
estrogen agonist or antagonist on the cervix. 1. Pike MC, Spicer DV, Dahmoush L, Press MF. Estrogens, proges-
togens, normal breast cell proliferation, and breast cancer risk.
Hormonal factors have been inconsistently associated Epidemiol Rev 1993;15:17-35.
with colon cancer risk in women, with some studies 2. Ries LAG, Eisner MP, Kosary CL, Hankey BF, Miller BA, Clegg
showing an inverse relationship with parity and hor- L, et al, editors. SEER cancer statistics review, 1975-2001.
mone replacement therapy.7,29 Colon cancer risk asso- Bethesda, MD: National Cancer Institute. Available at: http://
ciated with infertility and its treatment has received little seer.cancer.gov/csr/1975_2001/, 2004.
3. Althuis MD, Moghissi KS, Westhoff CL, Scoccia B, Lamb EJ,
attention with only 1 other published study, which Lubin JH, et al. Uterine cancer after use of clomiphene citrate to
reported results similar to ours. Specifically, in an induce ovulation. Am J Epidemiol 2005;161:607-15.
Australian cohort of infertile women, 1 colon cancer was 4. Brinton LA, Scoccia B, Moghissi KS, Westhoff CL, Althuis MD,
diagnosed among women exposed to in vitro fertilization Mabie JE, et al. Breast cancer risk associated with ovulation-
compared with 3 cases among unexposed women.23 stimulating drugs. Hum Reprod 2004;19:2005-13.
5. Brinton LA, Lamb EJ, Moghissi KS, Scoccia B, Althuis MD,
Whether fertility drugs other than clomiphene increase Mabie JE, et al. Ovarian cancer risk after the use of ovulation-
cancer risk requires further investigation. We previously stimulating drugs. Obstet Gynecol 2004;103:1194-203.
have reported a non-significant elevation in breast cancer 6. Franceschi S, Baron AE, La Vecchia C. The influence of female
risk among women who used gonadotropins for 6 or more hormones on malignant melanoma. Tumori 1990;76:439-49.
menstrual cycles (RR = 1.50) and who were first exposed 7. La Vecchia C, Franceschi S. Reproductive factors and colorectal
cancer. Cancer Causes Control 1991;2:193-200.
20 or more years ago (RR = 1.54).4 Melanoma, thyroid, 8. Negri E, Dal Maso L, Ron E, La Vecchia C, Mark SD, Preston-
cervical, and colon cancer sites had fewer than 5 exposed Martin S, et al. A pooled analysis of case-control studies of thyroid
cancers and did not show an increased cancer risk with cancer, II: menstrual and reproductive factors. Cancer Causes
gonadotropin use (RRs from 0.9-1.39). Control 1999;10:143-55.
Although our study had a number of strengths, there 9. Berrington G, Sweetland S, Green J. Comparison of risk factors
for squamous cell and adenocarcinomas of the cervix: a meta-
were some notable limitations. Although larger than analysis. Br J Cancer 2004;90:1787-91.
previously published studies, the total number of can- 10. La Vecchia C, Ron E, Franceschi S, Dal Maso L, Mark SD,
cers for each site was small. Given the retrospective Chatenoud L, et al. A pooled analysis of case-control studies of
nature of the study, we were unable to locate 20% of the thyroid cancer, III: oral contraceptives, menopausal replacement
study population and 11% did not agree to release their therapy and other female hormones. Cancer Causes Control
1999;10:157-66.
medical records. In addition, 41% of located subjects 11. Ron E, Lunenfeld B, Menczer J, Blumstein T, Katz L, Oelsner G,
did not complete a questionnaire, potentially leading to et al. Cancer incidence in a cohort of infertile women. Am J
a variety of selection biases that may have affected our Epidemiol 1987;125:780-90.
results. However, we were unable to detect any system- 12. Rossing MA, Daling JR, Weiss NS, Moore DE, Self SG. Risk of
atic selection biases in the current or previous analyses cutaneous melanoma in a cohort of infertile women. Melanoma
Res 1995;5:123-7.
of data from this cohort.3-5 Information on ovulation- 13. Rossing MA, Daling JR, Weiss NS, Moore DE, Self SG. In situ
stimulating drugs, although more complete than in most and invasive cervical carcinoma in a cohort of infertile women.
studies, was still less than optimal. Even though infor- Fertil Steril 1996;65:19-22.
mation about later drug use was obtained via question- 14. Breslow NE, Day NE. Statistical methods in cancer research.
naire, we could not account for drugs subsequently Volume IIdthe design and analysis of cohort studies. IARC Sci
Publ 1987;1-406.
prescribed by other providers among women who did 15. Brinton LA, Lamb EJ, Moghissi KS, Scoccia B, Althuis MD,
not complete the questionnaire. Finally, the pattern and Mabie JE, et al. Causes of infertility as predictors of subsequent
dose of drug exposures for many women that we cancer risk. Epidemiology 2005;16:1-8.
evaluated were quite different from those in current use. 16. Brinton LA, Melton LJ III, Malkasian GD Jr, Bond A, Hoover R.
In summary, after accounting for independent Cancer risk after evaluation for infertility. Am J Epidemiol
1989;129:712-22.
contributions of causes of infertility and reproductive 17. Modan B, Ron E, Lerner-Geva L, Blumstein T, Menczer J,
status, findings from this study do not suggest an Rabinovici J, et al. Cancer incidence in a cohort of infertile
association between ovulation-stimulating drug use and women. Am J Epidemiol 1998;147:1038-42.
674 Althuis et al
18. Rossing MA, Daling JR, Weiss NS, Moore DE, Self SG. Ovarian 25. Kuppens E, Bergman W, Welvaart K, Bruijn JA, Scheffer E.
tumors in a cohort of infertile women. N Engl J Med 1994;331:771-6. Multiple primary melanomas in a patient with familial-type
19. Venn A, Watson L, Bruinsma F, Giles G, Healy D. Risk of cancer DNS during clomiphene-induced pregnancy. Melanoma Res 1992;
after use of fertility drugs with in-vitro fertilisation. Lancet 2:71-4.
1999;354:1586-90. 26. Fuller PN. Malignant melanoma of the ovary and exposure to
20. Klip H, Burger CW, Kenemans P, van Leeuwen FE. Cancer risk clomiphene citrate: a case report and review of the literature. Am J
associated with subfertility and ovulation induction: a review. Obstet Gynecol 1999;180:1499-503.
Cancer Causes Control 2000;11:319-44. 27. Young P, Purdie D, Jackman L, Molloy D, Green A. A study of
21. Potashnik G, Lerner-Geva L, Genkin L, Chetrit A, Lunenfeld E, infertility treatment and melanoma. Melanoma Res 2001;11:
Porath A. Fertility drugs and the risk of breast and ovarian 535-41.
cancers: results of a long-term follow-up study. Fertil Steril 28. Kolonel LN, Hankin JH, Wilkens LR, Fukunaga FH, Hinds MW.
1999;71:853-9. An epidemiologic study of thyroid cancer in Hawaii. Cancer
22. Rossing MA, Daling JR, Weiss NS, Moore DE, Self SG. Risk of Causes Control 1990;1:223-34.
breast cancer in a cohort in infertile women. Gynecol Oncol 29. Kampman E, Potter JD, Slattery ML, Caan BJ, Edwards S.
1996;60:3-7. Hormone replacement therapy, reproductive history, and colon
23. Venn A, Watson L, Lumley J, Giles G, King C, Healy D. Breast cancer: a multicenter, case-control study in the United States.
and ovarian cancer incidence after infertility and in vitro fertilisa- Cancer Causes Control 1997;8:146-58.
tion. Lancet 1995;346:995-1000. 30. Goldstein SR, Siddhanti S, Ciaccia AV, Plouffe L Jr. A pharma-
24. Titus-Ernstoff L. An overview of the epidemiology of cutaneous cological review of selective oestrogen receptor modulators. Hum
melanoma. Clin Plast Surg 2000;27:305-16, vii. Reprod Update 2000;6:212-24.
www.ajog.org
First glimpse of the functional benefits of

clitoral hood piercings
Vaughn S. Millner, PhD,* Bernard H. Eichold II, MD, DR PH, Thomasina H. Sharpe, MD,
Sherwood C. Lynn Jr, MD
University of South Alabama, Mobile, AL
Received for publication November 15, 2004; accepted February 2, 2005
KEY WORDS In this exploratory study, we identify a positive relationship between vertical clitoral hood
Genital piercing piercing and desire, frequency of intercourse and arousal. There were no dramatic differences in
Female sexuality orgasmic functioning. Clinicians can play key roles in educating patients about potential
outcomes and risks of genital piercing.
Female genital piercing is an emerging form of body Orleans, La, between September 2001 and July 2003.
art in western culture.1 Vertical clitoral hood piercing, The participants were offered an opportunity to complete
one of several genital piercing techniques, is apparently the Female Sexual Functioning Inventory2 both before
popular and functional. A review of the literature piercing and 7 weeks after piercing. The instrument
reveals little more than anecdotal information, including requested information about the domains of arousal,
the belief that it is useful in the treatment of types of desire, lubrication, pain, orgasm, and overall sexual
female sexual dysfunction such as anorgasmia. satisfaction.
Of the 77 participants initially enrolled in the study,
33 (42%) completed and returned both questionnaires.
Objective and study design The mean age was 31 (SD = 9.4) and ages ranged from
19 to 55. Over half (55%) were between the ages of 19
The investigators conducted an exploratory study to and 29. Nine (27%) were between the ages of 31 and 40
describe the population and determine whether vertical years with an additional 6 (18%) between the ages of 41
clitoral hood piercing had associations with female sexual and 55 years. Fourteen (42%) of the participants were
functioning. In a University of South Alabama Institu- married and 14 (42%) were single. Three participants
tion Review Board–approved study, 33 female partici- (9%) were divorced and the remainder were separated
pants were recruited from a piercing studio in New or widowed. The majority (85%) were white and heter-
osexual (76%). The sample was generally well educated,
with 11 (33%) participants possessing a college degree
and 4 (12%) with some college education. Four (12%)
* Reprint requests: Vaughn S. Millner, PhD, Behavioral Studies
and Educational Technology, UCOM 3700, University of South
participants had masters degrees and 4 (12%) had
Alabama, Mobile, AL 36688-0002. doctorates. Eight (24%) earned a high school degree;
E-mail: vmillner@usouthal.edu 1 (3%) less than high school.
doi:10.1016/j.ajog.2005.02.130
676 Millner et al
The piercing materials were implant-grade surgical and stimulating the clitoris. At the same time, the clitoris
stainless steel, implant-grade titanium or 18-karat gold. enlarges and becomes erect elevating the clitoral hood,
The majority of piercings were steel. All participants which in turn acts as a direct stimulant to the clitoris via
kept the piercing. friction. Penetration causes movement and stretching of
the labia minora that pull down on the prepuce and
further stimulate the clitoris.
Results Piercing of the clitoral hood, either vertically or
We calculated 2-tailed bivariate analyses to test the horizontally, acts to lift the clitoral hood. The intention
degree of relationship between the before and after is to enhance sexual sensitivity and stimulation of the
scores. For the individual items comprising the domains, clitoris, the only organ in the human body that serves
we observed increased frequency of sexual desire (0.451, the single purpose of providing pleasure. In addition to
P = .008), level of desire (0.394, P = .023), and level of lifting and retracting the clitoral hood, piercings can be
arousal (0.429, P = .013). We observed significance in fitted with jewelry meant to directly stimulate or orna-
only 1 of the domains, desire (0.414, P = .017). Other ment the clitoris. Lesbians and heterosexual women
sexual functioning indexes were negligible. We expected experience both coital and noncoital sexual activity,
to see a change in the orgasm frequency and/or satis- including self masturbation and other types of genital
faction. Yet, contrary to popular belief, we saw no stimulation, wherein clitoral hood piercing can perform
dramatically significant difference in orgasm. a role.
The lack of significant differences between some of This limited survey encourages further study on the
the sexual functioning domains before and after piercing desire enhancing benefits of vertical clitoral hood pierc-
may reflect a true lack of difference or may be the result ing and the role of medical practitioners in treating and
of insufficient statistical power to detect significant counseling female patients seeking intimate piercing. We
relationships. We considered P ! .05 to be statistically found no indications to counsel against clitoral hood
different. An interim power analysis using a power of piercing, but urge practitioners to inform patients of the
80% indicated the need for a sample size greater than obvious health risks association with body piercing such
200 to detect significant differences in the other domains as infection and viral hepatitis as well as the possibility
of sexual functioning. Desire’s statistical significance is of streptococcal toxic shock syndrome associated with
noteworthy, considering the small sample size. piercing mucosal surfaces. We caution against general-
izing outcomes until other studies produce deeper
understanding.
Comment
We thank Elayne Angel with Rings of Desire for her
Vertical clitoral hood piercing’s influence on desire can assistance with data collection as well as Beth Mitchell,
be understood by considering the physiology involved in CRNP, and Peggy Hamlin, CRNP.
the sexual response cycle established by Masters and
Johnson as well as bearing in mind the more subjective
element of desire that so often eludes sexuality re- References
searchers. Genital piercing may address both objective
and subjective elements of desire. 1. Stirn A. Body piercing: medical consequences and psychological
The clitoral hood or prepuce serves a protective motivations. Lancet 2003;361:1205-15.
2. Rosen R, Brown C, Heiman J, Leiblum S, Meston C, Shabsigh R,
function and plays a critical role in sexual activity. et al. The female sexual function index (FSFI): a multidimensional
During sexual arousal, the labia minora become en- self-report instrument for the assessment of female sexual function.
gorged with blood pulling down on the clitoral hood J Sex Marital Ther 2000;26:191-208.
www.ajog.org
Arcus tendineus fascia pelvis: A further understanding

Todd S. Albright, DO, Alan P. Gehrich, MD, Gary D. Davis, MD, Farzaneh L. Sabi, MD,
Jerome L. Buller, MD
Division of Female Pelvic Medicine and Reconstructive Surgery, Department of Obstetrics/Gynecology, National
Capital ConsortiumdNIH (National Naval Medical Center, Uniformed Services University of Health Sciences,
Walter Reed Army Medical Center), Bethesda, MD
Received for publication July 24, 2004; accepted February 14, 2005
KEY WORDS Objective: The study was undertaken to further define the anatomy of the arcus tendineus fascia
Arcus tendineus fascia pelvis (ATFP).
pelvis Study design: Thirty cadavers were dissected to find the average length, SD, and range of the
White line ATFP. Comparisons were made to height and pelvis type. The average distance between the
Anatomy ischial spine and the attachment of the fascia of the rectovaginal septum (RVF) to the ATFP was
Rectovaginal septum measured.
Rectovaginal fascia Results: The average length, SD, and range in centimeters for the ATFP are 9.0, 0.70, and 7 to
10.5, respectively. The length of the ATFP increased with height. No associations could be made
regarding pelvis type. The average distance between the ischial spine and the attachment of the
RVF to the ATFP is 2.15 cm with a SD and range of 0.21 and 1.75 to 2.5, respectively.
Conclusion: In this study, an average length for the ATFP is established and the distance between
the ischial spine and the attachment of the RVF to the ATFP is redefined.
The arcus tendineus fascia pelvis (ATFP), originally is sparse in anatomic texts and only 2 articles have been
termed the ‘‘white line,’’ is a fibrous thickening that is dedicated to defining its anatomy.2,3 However, neither
made up of parietal fascia from surrounding muscles.1 study has accurately determined its length or attempted
These muscles are the pubococcygeous and iliococcy- to correlate its length to the pelvic morphology. Even
geous portions of the levator ani and the obturator some of the most widely used anatomic texts do not give
internus. The ATFP runs along the pelvic sidewall from an average length.4-7
its origin near the pubic symphysis to its insertion on the The ATFP has 3 functional partsdthe anterior,
ischial spine bilaterally. Discussion regarding the ATFP middle, and posterior segments. It is customary to refer
to the anterior segment as the most proximal and the
posterior segment as the most distal. The anterior
segment is attached to the lower posterior side of the
Supported by The Chief, Navy Bureau of Medicine and Surgery, body of the pubic bone approximately 1 cm from the
Washington, DC, Clinical Investigation Program. pubic symphysis and extends posterior for approxi-
The views expressed in this article are those of the authors and do
not reflect the official policy or position of the Department of the
mately 3 cm. It has attachments to the proximal urethra
Navy, Department of Defense, or the US government. and anterior vaginal wall. It functions as a lateral
Reprints not available from the authors. support for these structures. This portion of the ATFP,
doi:10.1016/j.ajog.2005.02.129
678 Albright et al
including the histology, has been extensively studied by Department of Clinical Investigation exempted this study
DeLancey.8,9 from Institutional Review Board approval.
The middle segment is approximately 3 cm long. It is An abdominal incision allowed access to the pelvis.
attached anteriorly to the anterolateral vagina. This The dome of the bladder was removed, and the lateral
attachment becomes less prominent toward its midpoint. attachments of the midline organs were dissected, allow-
At the midpoint of the middle segment, vessels from the ing visualization of the tendon’s entire path along
internal iliac artery running to the obturator internus the pelvic sidewall. The path originates 1 cm lateral to
muscle lie laterally. When placing sutures in this area, the pubic symphysis on the posterior, inferior border
care must be taken to avoid going too deep into the of the pubic bone, proceeds lateral to the urethra in a
obturator internus muscle. At the distal end of the middle posterior direction, extends along the pelvic sidewall,
segment are the fascial attachments of the arcus tendi- and ends at its insertion on the ischial spine. In many
neous levator ani (ATLA) and the fascia of the rectova- cases, the tendon was not clearly defined along its entire
ginal septum (RVF).10 These attachments distinguish the length. However, by using remnants of the tendon and
middle from the posterior segment of the ATFP. by using the ischial spine as an anatomic landmark
The posterior segment is approximately 2 to 2.5 cm in allowed for accurate measurement. No cadaver was
length and terminates on the ischial spine. It functions found to have detachment of the tendon from its origin
as the anchor for the fascia and all of its attachments. on the pubis.
DeLancey noted that the distal attachment of the First, the length of the ATFP was measured bilaterally
posterior segment to the ischial spine is avulsed in in centimeters while maintaining the curve of the pelvis.
96% of parous women.11 This avulsion occurs primarily Keeping the measuring tape in place, the distance be-
in childbirth and it is remarkable that the effects of tween the midpoint of attachment of the RVF to the
prolapse are not seen for decades. This may in part be ATFP and the ischial spine was measured. The RVF
due to the ‘‘anchors’’ provided by the ATLA and the attaches to the ATFP throughout its length. However,
RVF. Aging, estrogen deficiency, muscle atrophy, and/ this fascia condenses posteriorly. We like to refer to this
or lifestyle factors involving chronic increases in intra- condensation as the arcus tendineus rectovaginalis. By
abdominal pressure may ultimately be responsible. our observations, this posterior condensation of the
The ATFP is an important factor in the support of RVF is approximately 1 to 1.5 cm in width. It is the
pelvic structures. In 1912, White first noted its role in the midpoint of this condensation that attaches to the ATFP
support of the bladder and proposed a technique for and is where we began our measurement. Furthermore,
cystocele repair.12 Many different modifications have the obstetric conjugate and biischial spine diameter were
occurred since White’s proposal. Richardson enhanced measured. The obstetric conjugate was measured from
our surgical correction with his concept of the site- the posterior inferior margin of the pubic symphysis to
specific repair. He noted that the wide variety in the sacral promontory. The biischial spine diameter was
presentations of prolapse correlated with defects in the taken as the distance between the tips of the bilateral
different segments of the ATFP. He proposed the repair ischial spines. Finally, the patient’s height was obtained
of both cystocele and rectocele by correcting site-specific measuring heel to crown. Measurements of the obstetric
defects in the ATFP.13-15 Given the increasing impor- conjugate and the biishcial spine diameter along with
tance of the ATFP in the repair of pelvic organ prolapse, vaginal evaluation at dissection were used to diagnose
a better understanding of this fascia is needed. The goal pelvis type. Information regarding pelvis type and height
of this study is to further define the anatomy of the were consistently obtained following the measurements
ATFP and from these observations draw conclusions of the ATFP, and demographic data regarding the
concerning its function and repair. cadavers were collected after all measurements were
taken. Demographic data were limited to the information
Material and methods available to the anatomy laboratory.
Given the reported length of the ATFP to be 10 cm,
Seven fresh and 23 embalmed cadavers in the anatomic we assumed a significant difference in length of 0.5 cm
teaching laboratory at the Uniformed Services University and a SD of 1 cm. Setting power at 80% and alpha at
of Health Sciences were dissected to visualize the ATFP. .05, the necessary sample size was calculated to be 32
Visualization of the ATFP allowed for the following using the above parameters. However, because no prior
measurements to be taken: length of the ATFP bilater- SD for the ATFP existed and noting a SD of 0.7 after
ally, point of attachment of the RVF, obstetric conjugate, obtaining our measurements, it was found that only
biischial spine diameter, and patient height. Information 16 cadavers were needed for statistical significance.
on patient age and race was also noted. Because this Therefore, our results from 30 cadavers exceed the
information was obtained while teaching pelvic anatomy number needed to meet statistical significance. The
to medical students and is consistent with the intent of the same is true for the distance at which the RVF attaches
donors, the Anatomic Materials Use Committee and the to the ATFP.
Albright et al 679
Table Cadaver demographics and measurements

Measurements (cm)
Biischial Obstetric ATFP ATFP RVF RVF
Age Race Pelvis Height (in) spine conjugate Right left right left
1 89 W G 62 11 11 9 8.25 2.25 2.25
2 68 W G 65 11.5 12 9 8.5 2.5 2.5
3 77 W G 61 10.5 10 8 7.5 2.25 2
4 76 W G 66 12.5 13 9.5 9 2.5 2.25
5 94 W G 59 10 10 7 7 1.75 2
6 91 W A 62.5 10 12.5 8 8.5 2 2
7 80 W A 64 10.5 12.5 9 8.5 2 2.25
8 85 W A 67.5 11 12.5 10 10 2.25 2.25
9 93 W G 66 12.5 12 9.5 9.5 2 2
10 81 W G 71 13 12.5 9.5 10 2.5 2.5
11 64 W G 64 11.5 11 8.5 8 2 2
12 77 W G 64.5 11.5 10.5 8.5 8.5 2 2
13 91 W G 66 11.5 10 9.25 10 2.25 2.25
14 85 W G 69 12.5 12 9.5 9.5 2.25 2.5
15 82 W G 62 12.25 10.5 10 9.5 2 2
16 84 B A 66 12.5 14 9.5 9.5 2.25 2
17 80 W G 64 12 12.5 9 9 2 2
18 85 W G 69 12 12 8.75 9 2 2.25
19 90 W G 63 13 13 9.25 10 2 2
20 93 B G 63 11.5 12 9 8.5 2 2
21 73 W G 67 12.5 12 10.5 10 2.5 2.25
22 91 W G 64.5 12 11.5 8.5 9 2 2
23 98 W G 61 10 11 9 8.5 2 1.75
24 68 W G 62 11.5 11 9.25 9 2.25 2
25 92 W AND 63 9.25 10 8.5 8.5 2 2
26 89 W G 65 10.5 11.5 9 9 2.5 2.5
27 74 W G 62.5 10.5 11 9.25 9 2 2
28 76 B A 63 10 11.5 9.25 9.5 2 2.25
29 83 W G 66 10.5 12 9.25 9.25 2.5 2.5
30 71 W G 62 10 11.5 8.5 8.75 2.25 2.25
Average 64.35 11.317 11.617 9.025 8.958 2.158 2.150
SD 2.668 1.044 1.006 0.674 0.740 0.213 0.203
Range 59-71 9.25-13 10-14 7-10.5 7-10 1.75-2.5 1.75-2.5
Average of right and left measurements 8.991 2.154
SD of right and left measurements 0.703 0.206
W, White; B, black; G, gynecoid; A, anthropoid; AND, android.
The average length of the ATFP and the average loid pelvis types were diagnosed. Cadaver demographic
distance of the attachment of the RVF to the ATFP information can be seen in the Table.
were computed from the measurements. SDs and ranges The average length of the ATFP was 8.99 cm with a
were then computed. Comparisons between the lengths SD of 0.703 cm and a range of 7 to 10.5 cm. Measure-
of the ATFP and patient height and pelvis type were ments of the right and left ATFP were obtained for each
made. Also, to account for possible differences between cadaver and can be seen in the Table. Of note, the length
the fresh and embalmed cadavers, their average length of the ATFP on the right and left sides showed varia-
of the ATFP was compared. Statistical computations bility. The difference in length between the 2 sides
were performed with SPSS 12.0 (SPSS Inc, Chicago, Ill). ranged from 0 to 0.75 cm. The average length of the
ATFP was then compared with the height of the
Results cadavers. Although an exact association between height
and length could not be determined, a trend toward
The 30 cadavers used for dissection were made up of 27 increasing length of the ATFP can be seen in the Table as
white and 3 black women. Their average height was cadaver height increases. As for comparing length and
64.35 inches with a range of 59 to 71 inches. A total of pelvis type, there were not enough anthropoid, android,
24 gynecoid, 5 anthropoid, 1 android, and no platypel- or platypelloid pelvises to draw any conclusions. The
680 Albright et al
Figure 2 Artistic impression of the ATFP with attachment of

the ATLA and the RVF.
Figure 1 Picture of the ATFP with attachment of the arcus

tendineus levator ani and the rectovaginal fascia. RVF,
Rectovaginal fascia; ATLA, arcus tendineus levator ani; C,
point of convergence of the ATLA and RVF onto the ATFP;
IS, ischial spine; P, body of pubic bone; V, vagina. (Modified
from Leffler et al Am J Obstet Gynecol 2001;185:41-3.)
average length of the ATFP of the fresh and embalmed

cadavers was 8.88 and 9.02, respectively, P = .678. By
using P ! .05 as statistically significant, there was no
statistically significant difference between the fresh and
embalmed cadavers.
The average distance of the attachment of the RVF
to the ATFP as measured from the ischial spine was Figure 3 Diagram of ATFP highlighting observed vectors of
attachments.
2.154 cm with a SD of 0.206 cm and a range of 1.75 to
2.5 cm. Measurements of the right and left sides of
attachment were obtained for each cadaver and can be length was not stated, but adding the anterior and
seen in the Table. Again, small discrepancies were posterior segments would give a length of 8.55 cm. This
observed between the left and right distances. more closely resembles our finding of 8.99 cm, which is
Regarding the general pelvic measurements, the av- significantly less than the often-quoted 10 cm.
erage length of the biischial spine diameter was 11.317 The rectovaginal septum separates the vagina from
cm with a SD of 1.044 cm and a range of 9.25 to 13 cm. the rectum. Its origin has been vigorously debated in
The average length of the obstetric conjugate is 11.617 earlier literature.16-20 Here, we are concerned with its
cm with a SD of 1.006 cm and a range of 10 to 14 cm. attachment to the ATFP. We now know that defects in
The individual measurements can be seen in the Table. this attachment result in rectocele formation. Leffler
No comparison was made between the average length of et al10 obtained an average distance of 4.8 cm for the
the ATFP and these measurements. Rather, these mea- attachment of the RVF to the ATFP as measured from
surements helped in the diagnosis of the pelvis type of the ischial spine. However, our observations in this
the cadaver and are noted as points of interest. study place this point at 2.15 cm from the ischial spine.
It might be that Leffler et al chose a different point of
Comment reference than what was indicated in the article. Our
illustration, Figure 1, is a modification of the picture in
The ATFP is a point of attachment for many gynecologic the article by Leffler et al.10 It should be noted that the
procedures. As this tendon is increasingly used in pro- rectovaginal septum has minor attachments to the
lapse and incontinence procedures, it becomes more ATFP all along its lateral border. We measured only
important to have a thorough understanding of its the condensation of these attachments at its distal end.
anatomic course and attachments. Ocelli et al3 dissected We were able to consistently identify this condensation
2 cadavers and determined the ATFP length to be 10 cm. in all cadavers. Also, we noted that condensations of the
This is the length that is quoted by most gynecologists in ATLA and the RVF attach at the same point on the
discussions regarding the ATFP. Leffler et al10 dissected ATFP (Figure 2). This observation makes inherent sense
a total of 24 cadavers and determined the lengths for from an engineering standpoint (Figure 3). The attach-
anterior and posterior segments of the ATFP. A total ment of the RVF at a distance of 4.8 cm from the ischial
Albright et al 681
spine would change the forces applied to the ATFP, redefines the attachment of the RVF. By establishing an
making the tendon less efficient. It also places the average length for the ATFP and points of attachment,
attachment of the RVF to the ATFP very near, if not computer modeling of the pelvis will become more
directly over, vessels leading to the obturator internus accurate and in turn will help improve on medical and
muscle. This is also stated by Ocelli et al.3 Therefore, surgical therapies.
placing sutures in an attempt to reattach RVF to the
ATFP during a posterior repair would entail a high risk References
of potential catastrophic bleeding. With our findings,
such risky placement would not be warranted. Our 1. Weber A, Walters M. Anterior vaginal prolapse: review of anat-
results indicate no need to place sutures beyond this omy and techniques of surgical repair. Obstet Gynecol
attachment during a paravaginal repair. This issue has 1997;89:311-8.
2. Pit M, De Ruiter M, Lycklama A, Nijeholt AA, Marani E,
previously been questioned by Cornella.9 We surmise Zwartendijk J. Anatomy of the arcus tendineus fasciae pelvis in
that there may be no structural benefit of placing the females. Clin Anat 2003;16:131-7.
sutures as far back as the ischial spine. 3. Occelli B, Narducci F, Hautefeuille J, Francke JP, Querleu D,
Making associations to cadaver height further adds Crepin G, et al. Anatomic study of arcus tendineus fasciae pelvis.
information when building a model of the pelvis. Even Eur J Obstet Gynecol Reprod Biol 2001;97:213-9.
4. Netter F, Colacino S. Atlas of human anatomy. Summit, NJ:
though a direct correlation to height could not be made, CIBA-GEIGY Corporation; 1996.
an obvious trend toward increasing length with increas- 5. Williams P, Warwick R, Dyson M, et al. Gray’s anatomy. New
ing height was seen. Because this seems to follow York: Churchill Livingstone; 1993.
reasonable logic, we believe the exceptions are most 6. Agur A, Lee M. Grant’s atlas of anatomy. Philadelphia: Lippin-
cott Williams and Wilkins; 1999.
likely caused by differences in cadaver morphology, such
7. Moore K, Dalley A. Clinically oriented anatomy. Philadelphia:
as fat, atrophy, and edema. Lippincott Williams and Wilkins; 1999.
There are aspects of this study that could be im- 8. DeLancey J, Starr R. Histology of the connection between the
proved. First, our measurements were taken to the vagina and levator ani muscles: implications for urinary tract
nearest quarter of a centimeter. Taking the measure- function. J Reprod Med 1990;35:765-71.
ments to the nearest tenth of a centimeter would have 9. DeLancey J. Structural support of the urethra as it relates to stress
urinary incontinence: the hammock hypothesis. Am J Obstet
increased accuracy. However, the change in the average Gynecol 1994;170:1713-23.
value would have been small. Second, slicing the rele- 10. Leffler K, Thompson J, Cundiff G. Attachment of the rectovaginal
vant area of the pelvis into sections would provide for a septum to the pelvic sidewall. Am J Obstet Gynecol 2001;85:41-3.
more detailed measurement, but this would have added 11. DeLancey J. Fascial and muscular abnormalities in women with
significant time and cost to the study. Our dissections urethral hypermobility and anterior vaginal wall prolapse. Am
J Obstet Gynecol 2002;187:93-8.
did allow complete visualization for these measure- 12. White G. An anatomical operation for the cure of cystocele. Am
ments. Third, when comparing the length of the ATFP J Obstet Gynecol Dis Women Child 1912;65:286-90.
to cadaver height, one must consider that osteoporosis 13. Richardson A, Lyon J, Williams N. A new look at pelvic
would have affected the results in this age group. It relaxation. Am J Obstet Gynecol 1976;126:568-73.
would have been more accurate to compare the length of 14. Richardson A, Edmonds P, Williams N. Treatment of stress
urinary incontinence due to paravaginal fascial defect. Obstet
the ATFP with a hip to heel length measurement of the Gynecol 1981;57:357-62.
cadavers. Fourth, even though our study reaches statis- 15. Richardson A. The rectovaginal septum revisited: Its relationship
tical significance, studies involving a larger number of to rectocele and its importance in rectocele repair. Clin Obstet
dissections that are more diverse in ethnicity and pelvic Gynecol 1993;36:976-83.
types would truly define a standard length of the ATFP. 16. Curtis A, Anson B, Beaton L. The anatomy of the subperitoneal
tissues and ligamentous structures in relation to surgery of the
Finally, the study would have been more complete by female pelvic viscera. Surg Gynecol Obstet 1940;70:643-56.
measuring the distance of all the attachments to the 17. Uhlenhuth E, Wolfe W, Smith E, et al. The rectovaginal septum.
ATFP as these measurements are needed for a complete Surg Gynecol Obstet 1948;86:148-63.
reconstruction of a pelvic model. 18. Ricci J, Thom C. The myth of a surgically useful fascia in vaginal
plastic reconstructions. Q Rev Surg Obstet Gynecol 1954;2:253-61.
Our dissection of 30 cadavers is the largest study to
19. Uhlenhuth E, Nolley G. Vaginal fascia, a myth? Obstet Gynecol
date to specifically examine the length of the ATFP. We 1957;10:349-58.
therefore propose our finding of approximately 9 cm to 20. Milley P, Nichols D. A correlative investigation of the human
be used as the current standard length. This study also rectovaginal septum. Anat Rec 1969;163:443-51.
www.ajog.org
Characterization of vaginal microflora of healthy,

nonpregnant women by chaperonin-60
sequence-based methods
Janet E. Hill, PhD,a Swee Han Goh, PhD,c Deborah M. Money, MD, FRCSC,d,*
Melissa Doyle,a Andra Li, BSc,a William L. Crosby, PhD,b Matthew Links, BSc,b
Amy Leung,c Debbie Chan,c Sean M. Hemmingsen, PhDa
National Research Council of Canada, Plant Biotechnology Institutea; Department of Computer Science,
University of Saskatchewan,b Saskatoon, Saskatchewan, Canada; Department of Pathology and
Laboratory Medicine and UBC Centre for Disease Control, University of British Columbiac;
and Department of Obstetrics and Gynecology, Children’s and Women’s Health Centre
of British Columbia,d Vancouver, British Columbia, Canada
Received for publication May 28, 2004; revised January 6, 2005; accepted February 14, 2005
KEY WORDS Objective: The purpose of this study was to use a novel method that was based on the application
Vaginal flora of chaperonin-60 sequencing to describe the vaginal microflora of 16 healthy women.
Chaperonin-60 Study design: Asymptomatic women consented for vaginal swabs to be collected at the time of a
Molecular method clinical pelvic examination. Total genomic DNA was isolated from the vaginal swabs.
Degenerate, universal polymerase chain reaction primers were used to amplify an approximately
555 base pair region of the universal chaperonin-60 gene, which is found in all eubacteria and
eukaryotes, from the total genomic DNA and libraries of cloned polymerase chain reaction
products were constructed. Library clones were sequenced, and the resulting sequences were
assigned to taxonomic groups on the basis of similarity to reference sequence data. Presence of
Chlamydophila psittaci sequences in the samples was confirmed by species-specific polymerase
chain reaction.
Results: Sixteen of the 23 women who were enrolled had normal flora by Nugent’s score of !4
and had adequate polymerase chain reaction product for assessment. Vaginal flora libraries were
dominated by a variety of sequences with similarity to Lactobacillus spp L crispatus, L iners,
L gasseri, L jensenii, and L buchneri. Other sequences that were identified included representatives
of Gardnerella spp, sequences with similarity to Porphyromonas spp and Megasphaera spp and
sequences identical to C psittaci.
* Reprint requests: Deborah M. Money, MD, FRCSC, Assistant Professor & Head, Division of Maternal Fetal Medicine, University of BC,
BC Women’s Hospital, Rm 2H30, 4500 Oak St, Vancouver, BC V6H 3N1.
E-mail: dmoney@cw.bc.ca
doi:10.1016/j.ajog.2005.02.094
Hill et al 683
Conclusion: Culture-independent, chaperonin-60 sequence-based molecular methods can lead to

the identification of greater diversity within defined taxa compared with those that are identified
by standard culture-based methods and to the identification of novel organisms that were not
previously associated with vaginal flora.
Human vaginal flora plays a profound role in repro- Chaperonin-60 is a molecular chaperone essential for
ductive health and disease. However, our primitive the folding and assembly of proteins and protein com-
understanding of the complex microbial ecosystem of plexes in all eubacteria and in the plastids and mitochon-
the genital tract greatly hampers our ability to develop dria of eukaryotes. The gene encoding chaperonin-60
appropriate, focused therapies for genital infections. (cpn60) offers several advantages over the widely used 16S
Given the current limitations in our diagnostic abilities, recombinant RNA (rRNA) gene as a target for microbial
it is naive to assume that we know all of the organisms species identification and phylogenetics.14 A robust mo-
that are involved in genital tract health. It is likely that lecular method for the identification of microorganisms,
the microorganisms that are responsible for reproduc- which is based on the amplification of a 549– to 567–base
tive health and disease remain to be discovered. To pair (bp) portion of the cpn60 gene (the ‘‘universal target’’)
date, no exhaustive, culture-independent survey has been with universal degenerate polymerase chain reaction
done of this important microbial community. (PCR) primers,15 and the comparison of amplified se-
The use of conventional culture repeatedly has been quences to a reference database of cpn60 sequences has
found to be unhelpful, because approximately 5% of been applied previously to phylogenetic studies,16 the
normal flora is comprised of multiple organisms that are identification of clinical isolates,17-19 and studies of the
implicated as genital pathogens; thus, their presence microbial ecologic condition of the animal gastrointesti-
alone is insufficient information. In addition, many are nal tract.20,21 The cpn60 universal target region generally
extremely difficult or impossible to culture routinely, provides more discriminating and phylogenetically infor-
and there may be organisms that have yet to be mative data than the 16S rRNA target, particularly
detected.1 The organisms that are associated most often between closely related species.16 Sequence variation
with bacterial vaginosis include anaerobic bacteria, in extends quite uniformly throughout the cpn60 coding
particular Bacteroides spp, Peptostreptococcus spp, region, whereas variable regions of 16S rRNA genes are
Gardnerella vaginalis, and Mycoplasma hominis.2 The dispersed between regions of highly conserved sequence
fastidious nature of these organisms makes culture- that result in stable secondary structure and facilitates
based methods impractical. This major limitation of PCR artifacts. Cpn60 genes generally are present in a
culture-based methods has been described as ‘‘the great single copy in prokaryotic genomes, which makes an
plate count anomaly’’3 because only a small fraction of attractive target for quantitative methods. The relatively
microorganisms that are present in a population can be small size of the universal target facilitates high through-
cultured. Studies that are based on culture have char- put sequencing approaches. Finally, a reference database
acterized vaginal lactobacilli4 or other specific organisms of cpn60 sequences is available. 14
of interest, such as G vaginalis.5 A few new organisms We describe the application of cpn60 sequence-based
have been recognized,6-8 but these generally are associ- methods to the characterization of vaginal flora of 16
ated with disease, not with healthy flora. There is a healthy nonpregnant women. The sample size was
reasonable expectation that organisms that are impor- chosen in consideration of the cost of the sequencing
tant to reproductive health may have evaded detection of a large number of clones and our previous experience
with standard methods. with cpn60 sequence-based microbial ecologic studies.
The development of culture-independent, gene-based Our previous results demonstrate that the size of the
methods has facilitated small-scale studies of a wide current pilot-scale study will be sufficient to identify
variety of complex microbial communities.9 Molecular potentially interesting targets that can then be moni-
methods have been applied in previous studies to iden- tored in a larger group of subjects.20,21
tify and enumerate vaginal organisms. However, the
relatively small scale and often generally descriptive
nature of these studies leaves us with a somewhat
superficial understanding of vaginal flora.4,10-12
Recently, results of a larger study demonstrated the Subjects
potential usefulness of the application of high through-
put molecular methods to the characterization of This project received University of British Columbia
vaginal microflora.13 Research Ethics Board approval. Women who attended
684 Hill et al
an outpatient Sexually Transmitted Diseases clinic, at

Table I Numbers of clones and unique sequences from 16
the British Columbia Centre for Disease Control, vaginal flora cpn60 libraries
Vancouver, British Columbia, who self-identified as not
Number of unique
having symptoms were offered enrollment. Written in-
Number of clones nucleotide
formed consent was obtained; at the time of the standard
Library sequenced sequences
speculum examination, an additional swab for vaginal
secretions was taken from the posterior fornix of the hvf3233 430 5
hvf3238 410 1
vagina. All women had clinical specimens that were
hvf3244 436 1
evaluated for bacterial vaginosis by standardized Nugent hvf3245 440 3
scoring at the British Columbia Centre for Disease hvf3246 441 2
Control clinical laboratory. As part of routine clinical hvf3247 439 6
assessments, samples were taken for routine light micros- hvf3257 431 9
copy assessment for yeast and Trichomonas vaginalis, hvf3258 446 5
Chlamydia trachomatis PCR, and Neisseria gonorrhoeae hvf3265 434 12
culture. hvf3266 449 1
hvf3267 338 3
hvf3268 386 1
Total genomic DNA isolation hvf3269 419 1
hvf3271 451 3
Each clinical Dacron swab was processed within 24
hvf3272 466 17
hours of receipt from the clinic. The contents of each hvf3273 367 4
swab were extracted in 800 mL of DNAzol (MRC Inc,
Cincinnati, Ohio). The extract was vortexed vigorously,
with pulsing for 2 minutes. Ethanol (600 mL) was added,
mixed, and incubated at room temperature for 5 Template preparation and sequencing
minutes. Precipitated nucleic acid was pelleted by cen-
trifugation and washed twice with 1 mL of 75% ethanol. The sequencing template was prepared by the TempliPhi
One hundred microliters of 8 mmol/L NaOH was used system (Amersham Pharmacia, Piscataway, NJ) at a 1 to
to solubilize the nucleic acid, which was followed by the 8 recommended reaction scale in 384-well plates (total
addition of 3 mL of 1 mol/L HEPES to neutralize the reaction volume 2.5 mL). Sequencing reactions were
purified DNA solution. performed by direct addition of ET terminator reaction
mix (Amersham Pharmacia) to 384-well plates that
PCR and creation of PCR product libraries contained the TempliPhi products. Sequencing reactions
were thermocycled according to the manufacturer’s
Each DNA sample (2 mL) was used as template in PCR recommended protocol. Sequencing reactions were re-
reactions with 0.5 mg of each universal cpn60 PCR solved on an ABI PRISM 3730XL DNA Analyzer system
primer (H279 50 -GAI III GCI GGI GAY GGI ACI ACI (Applied Biosystems, Foster City, Calif) at the McGill
AC-30 and H280 50 -YKI YKI TCI CCR AAI CCI GGI University and Genome Québec Innovation Centre.
GCY TT-30 ), 50 mmol/L KCl, 10 mmol/L Tris (pH 8.3),
1.5 mmol/L MgCl2, 200 mmol/L of each deoxynucleo- Sequence data processing and bioinformatics
side triphosphate, 2 U Taq DNA polymerase in a final
volume of 50 mL. After the addition of paraffin oil, PCR Raw sequence data was processed using the Phred22
amplification with a robocycler (Stratagene, La Jolla, program, which assigns quality values to the bases and
Calif) was carried out for 3 minutes at 95(C for 1 cycle, trims poor quality regions. The resulting sequences were
followed by 40 cycles of 1 minute at 94(C, 2 minutes at clustered on the basis of sequence identity with the
40(C, 5 minutes at 72(C, and completed with 1 cycle of d2_cluster.23 Clusters of identical sequences were assem-
10 minute at 72(C. PCR products from each template bled, with the use of Phrap,24 which incorporates the base
were agarose gel purified and ligated into T-A cloning quality information from Phred. Manual confirmation of
vector pCR2.1-TOPO (Invitrogen, Carlsbad, Calif). contig assembly was done with Gap4 (version 4.6) in
Ligation mixtures were used to transform Escherichia the Staden software package (release 2000.0; J. Bonfield,
coli strain JM109. The 16 resulting libraries were plated K. Beal, M. Betts, M. Jordan, R. Staden, 2000). Sequence
on Luria Broth (LB)/ampicillin/X-gal, and 480 white data, template information and similarity results were
colonies were picked for each library. Colonies were placed in a mySQL database (mySQL AB, Uppsala,
picked into 96-well plates that contained 100 mL of LB Sweden) for storage and further analysis. Sequence
with 100 mg/mL ampicillin. After overnight incubation manipulations, such as format changes and amino acid
at 37(C, 100 mL of 30% (vol/vol) glycerol was added to translations, were done with the European Molecular
each well, and the cultures were stored at 80(C. Biology Open Software Suite software.25 Sequence
Hill et al 685
Figure 1 Phylogenetic relationships of unique cpn60 sequences from 16 vaginal flora libraries. The tree was calculated with the use
of a maximum likelihood distance calculation, followed by neighbor-joining. Phylogenetic clusters L1-L6, A1-A4, B1, CH1, and
CL1 are indicated within the major taxonomic groups.
alignments were done using Clustal W.26 To determine calculated with the maximum likelihood option of
the putative taxonomy of each contig and singleton that dnadist. Dendrograms were constructed from distance
arises from the assembly step, each sequence was com- data by neighbor-joining with neighbor. Consensus trees
pared with a reference set of cpn60 sequences with the were calculated with consense, and branch lengths were
sequence alignment program FASTA.27 superimposed on consensus trees using fitch.
Phylogenetic analysis was done with programs in the
Phylogeny Inference Package software package (version Species-specific PCR primer design and
3.5c; Distributed by the author [Felsenstein J], 1993, amplification of C psittaci cpn60
Department of Genetics, University of Washington,
Seattle, Wash). Specifically, alignments were sampled Primers H1520 (50 - GCT CAG GTA GCC ACC ATT
for bootstrap analysis with seqboot; distances were TC 30 ) and H1521 (50 - GCT AGA AAG GTA TCC
686 Hill et al
Table II Taxonomic identification of cloned cpn60 sequences, based on comparison to reference sequence data
Nearest reference Range of sequence Number of sequences Number of clones
Taxonomic group sequence neighbor identity (%) recovered recovered
A1 G vaginalis ATCC14018 89 1 1
A2 G vaginalis ATCC14018 88-95 12 629
B1 Porphyromonas levii ATCC29147 72 1 1
CH1 C psittaci 6BC ATCCVR-125 100 1 678
CL1 Megasphaera elsdenii ATCC25940 70 1 1
L1 L crispatus A3M75 93-100 8 2393
L2 L gasseri ATCC9857 93-97 6 393
L3 L jensenii ATCC25258 93-96 2 65
L4 L gasseri ATCC9857 92 1 2
L5 L iners A3M7 98-99 12 2503
L6 L buchneri ATCC4005 83-84 3 5
TCG GTT G 30 ) were designed with Oligo6 (Molec- samples were used had completely normal flora by all
ular Biology Insights Inc, Cascade, Colo) and Signature standard clinical testing.
Oligo (LifeIntel, Port Moody, BC, Canada). Amplifica-
tions were performed in a 50 mL reaction that contained
20 mmol/L Tris-HCl pH 8.0, 50 mmol/L KCl, 1.5 Cpn60 gene sequences amplified
mmol/L MgCl2, 0.2 mmol/L deoxyribonucleoside tri- from vaginal swabs
phosphate, 20 pmol of H1520, 20 pmol of H1521, and 2
High-quality sequence data were obtained for 6869 of
units Taq DNA polymerase. Reactions were incubated
the 7680 clones that were picked randomly from 16
at 95(C for 5 minutes, followed by 40 cycles of 30
libraries. Data from the remaining 811 clones were
seconds at 95(C, 30 seconds at 62(C, and 30 seconds at
excluded from the analysis because of incomplete or
72(C. A final extension of 10 minutes at 72(C followed
partially ambiguous sequences. As summarized in Table
the last cycle. Amplifications were performed with a
I, pairwise comparisons of the sequences that were
thermocycler instrument (BioRad iCycler; Bio-Rad
determined for each library indicated that from 1 to 17
Laboratories, Hercules, Calif).
different nucleotide sequences were identified in each of
the 16 libraries. Pooling of the sequence data from all 16
libraries resulted in the identification of a total of 57
Results different nucleotide sequences (GenBank accessions
Clinical samples AY581720-AY581776). Phylogenetic analysis of the 57
different nucleotide sequences that were recovered from
Twenty-three nonpregnant, sexually active women who the pooled library data resulted in the identification of
were being seed for a sexually transmitted infection 13 distinct clusters of sequences (Figure 1). Clusters were
screening examination, without genital symptoms, were classified on the basis of similarity of the sequences to
enrolled after written informed consent. Their ages reference sequence data (Table II). Most clones that
ranged from 19 to 35 years. All of the women underwent were analyzed (5361 clones, representing 32 distinct
the clinic’s standardized questionnaire and examination. sequences) fell into clusters L1 to L6. Sequences in L1
Two women had abnormal findings on pelvic examina- to L5 were at least 92% identical to Lactobacillus spp.
tion and were excluded immediately. All remaining Sequences in L6 were consistent with the Lactobacillales
women tested negative for N gonorrheae, and C tracho- family but had weaker sequence similarity (83%-84%)
matis. Yeast and T vaginalis were not detected by light to any reference Lactobacillus sp. Four distinct clusters
microscopy in any subjects. One sample was ‘‘interme- (A1-A4) of Actinobacteria-like sequences were identi-
diate’’ for bacterial vaginosis, with a Nugent’s score of 4, fied. These sequences (22 sequences, represented by 742
and was excluded. Two samples were eliminated because clones) were all at least 88% identical to G vaginalis
they had been delayed in transit to the research labora- ATCC 14018 (American Type Culture Collection,
tory by O24 hours, and 3 samples did not yield sufficient Manassas, Va). Single sequences were identified with
PCR product in the initial reactions to generate an similarity to the Clostridiales family (cluster CL1), the
adequate library. The remaining 16 women for whom Bacteroidetes family (cluster B1), and the Chlamydiales
Hill et al 687
Figure 2 Composition of 16 vaginal flora libraries. The number of clones that corresponded to each phylogenetic group are
presented. Phylogenetic groups are described in Figure 1 and in the text.
(cluster CH1, 100% identical to Chlamydophila psittaci hvf3258, and hvf3272 contained sequences in the A1 to
ATCC VR-125). A4 clusters (88%-100% identical to G vaginalis ATCC
14018). Sequences identical to the type strain of C
Comparison of sequence profiles psittaci were identified in libraries hvf3257, hvf3267
of individual libraries and hvf3269.
Figure 2 shows the taxonomic composition of the 16 Variation within G vaginalis sequence cluster
vaginal flora libraries. Nine of the libraries were
composed exclusively of Lactobacillus-like sequences. Twenty-two sequences (represented by 742 clones) with
Library hvf3273 contained one sequence with similarity strong similarity to G vaginalis were identified. A phy-
to the Bacteroidetes family and one sequence with logenetic analysis of these sequences and other related
similarity to the Clostridiales family and was otherwise Actinobacteria showed that these sequences reliably
composed completely of Lactobacillus-like sequences. cluster with G vaginalis (Figure 3). Pairwise nucleotide
The major Lactobacillus constituent in 6 libraries sequence identities within this group of sequences
(hvf3233, hvf3238, hvf3245, hvf3267, hvf3268 and ranged from 86% to 99%. A multiple sequence align-
hvf3273) was the L1 cluster (93% identical to L crisp- ment of the 22 G vaginalis-like sequences resulted in the
atus). The L5 group (98%-99% identical to L iners) was identification of 107 positions of difference in the 552-bp
most abundant in hvf3244, hvf3246, hvf3266 and alignment (data not shown). Seven of these differences
hvf3271. Library hvf3265 was the only library to be were found in codon position 1; 1 difference was found
dominated by L2 type sequences (93%-97% identical to in position 2, and 99 differences were found in the third
L gasseri ATCC 9857). Libraries hvf3247, hvf3257, codon position.
688 Hill et al
Figure 3 Phylogenetic tree shows the relationships of Gardnerella vaginalis–like library sequences to closely related actinobacteria.
The tree was calculated from 500 bootstrap iterations with maximum likelihood distance calculation and neighbor-joining.
Bootstrap values (of 500) for major branch points in the tree are indicated. Scale bar indicates 0.1 substitutions per site.
Specific amplification of C psittaci-like closely related species such as C abortus (95% identical
sequences to C psittaci in the 555-bp amplified region of cpn60).
Primers for the specific amplification of a 174-bp region
Phylogenetic analysis of the C psittaci-like sequence that of the C psittaci cpn60 gene (from positions 160-333 of
were derived from the vaginal flora libraries and refer- the 555-bp cpn60 universal target) were designed on the
ence sequence data from additional Chlamydiales family basis of a multiple sequence alignment of partial cpn60
members was performed to confirm that C psittaci sequences from C pneumoniae J138 (Genbank accession
formed a distinct taxon based on the 555-bp region of NC_002491), C muridarum (NC_002620), C trachomatis
cpn60 (Figure 4) and could be discriminated from D/UW-3/CX (NC_000117), C pneumoniae CWL029
Hill et al 689
Figure 4 Phylogenetic tree shows the relationships of Chlamydophila psittaci–like library sequences to closely related Chlamy-
diaceae. The tree was calculated from 500 bootstrap iterations with maximum likelihood distance calculation and neighbor-joining.
Bootstrap values (of 500) for major branch points in the tree are indicated. Scale bar indicates 0.1 substitutions per site.
(NC_000922), C pneumoniae AR39 (NC_002179), C felis sequence analysis (data not shown). All sequences were
FEIS (AF448139), C pecorum (AF109789), C abortus 100% identical to library clone sequences.
B577 (AF109790), C caviae GPIC (NC_003361), C abor-
tus AB7 (AY052785), C suis R27 (AY581778), C suis H7 Comment
(AY581779) and C psittaci ATCC VR-125 (AY581777).
Figure 5 shows the results of PCR reactions that were Previous studies of the vaginal flora of healthy individ-
performed with C psittaci-specific primers on total ge- uals that were based on culture or sequence-based
nomic DNA samples that were used to generate the methods have led to the understanding of this microbial
vaginal flora libraries. The expected product size was community as relatively homogenous and dominated by
obtained from samples 3247, 3267, and 3269. Product was a small subset of the Lactobacillus acidophilus complex,
also detected in template 3246. The identity of amplified particularly L crispatus, L gasseri, L jensenii, and L iners.
products from these 4 templates was confirmed by The shortcomings of exclusively culture-based studies
690 Hill et al
Figure 5 Chlamydia psittaci–specific PCR. PCRs performed with C psittaci–specific primers on total genomic DNA samples that
were used to generate the vaginal flora libraries. A 174-base pair product was obtained from samples 3246, 3247, 3267, and 3269.
Lane N is a negative control reaction that contains no template DNA. M, Size markers.
are illustrated by the case of L iners. This organism, that only 1 sequence was recovered in some libraries. In a
unlike other Lactobacillus spp, can be cultured only on much more diverse community (pig feces), most sequen-
blood agar and was thus overlooked in most studies of ces were detected at a frequency of approximately 0.1%
vaginal lactobacilli that rely on Man, Rogosa, and (1 occurrence in 1125 sequences),20 a level that would
Sharpe agar. It is now apparent that L iners is a major make the sequences likely undetectable in a smaller
component of the vaginal flora.4 study. It is also likely that there is bias in the PCR
Consistently with previous findings, we found that reaction, in which some templates are favored on the
most of the cpn60 sequence libraries that were constructed basis of composition (especially the guanine-cytosine
in the current study were dominated by sequences with content) or priming efficiency. To address these issues
strong similarity to the L acidophilus complex, specifically and identify rarer, potentially unculturable organisms in
L crispatus, L gasseri, L jensenii and L iners. Most of the vaginal flora, technical advances that include subtraction
libraries were found to be composed of representatives of methods and modified PCR protocols are being pursued.
1 or 2 of the Lactobacillus clusters, frequently L1 In addition to identifying gross taxonomic clusters,
(L crispatus) and L5 (L iners; Figure 2). A similar result we also identified a large amount of variation within
was obtained in a study of the vaginal Lactobacillus flora these defined taxa. The 6 identified Lactobacillus-like
of 23 healthy Swedish women with the use of randomly clusters, L1 to L6, each contained from 1 to12 distinct
amplified polymorphic DNA analysis, where most indi- sequences (Table II; Figure 1). This potentially biolog-
vidual samples were found to contain only 1 or 2 ically significant ‘‘intraspecies’’ diversity would not be
randomly amplified polymorphic DNA patterns, which apparent with the use of culture-based methods or
indicated the dominance of 1 or 2 species.4 The results of a molecular methods (such as denaturing gradient gel
recent 16S rRNA sequence-based study of the vaginal electrophoresis) in which banding patterns are often
flora of 3 healthy subjects support the observation that identical for closely related species of Lactobacillus.11
healthy vaginal flora is dominated by 1 or 2 Lactobacillus Similar ‘‘intraspecies’’ variation was observed in the
spp.13 G vaginalis-like taxa, clusters A1 to A4. G vaginalis can
Five of the libraries that we examined contained only be isolated from the vaginal flora of individuals with
1 sequence each (Table I). We do not suggest that these healthy vaginal ecosystems and individuals who receive
subjects are colonized only by 1 organism but rather a diagnosis with bacterial vaginosis, although the
consider that this apparent lack of diversity is the result reported proportions of healthy individuals harboring
of a combination of factors that are related to the G vaginalis varies widely.5,28,29 Results of previous
application of a PCR-based method to a microbial studies of G vaginalis isolates from the vagina suggest
community that is dominated largely by a small number that the ‘‘biotypes’’ of G vaginalis that are associated
of species. The relative abundances of organisms in any bacterial vaginosis are distinct from the G vaginalis that
complex microbial community can vary over many are found to varying degrees in healthy individuals.5 The
orders of magnitude so that, in a total DNA preparation data presented in Figure 3 certainly support the idea
from the community, genomes of the most abundant that there is tremendous variability within the
organisms far outnumber those of rare organisms and G vaginalis taxon and that a quantitative assessment of
will be over represented correspondingly in the PCR the occurrence of these organisms in healthy and non-
product pool. Given that we sequenced only a few healthy vaginal ecosystems certainly would provide
hundred clones from each library, it is not surprising clues to the significance of the variability.
Hill et al 691
The most surprising finding in the current study was 5. Aroutcheva AA, Simoes JA, Behbakht K, Faro S. Gardnerella
the detection of sequences that are identical to C psittaci vaginalis isolated from patients with bacterial vaginosis and from
patients with healthy vaginal ecosystems. Clin Infect Dis
in 3 of the libraries. Of the 8 characterized serovars of 2001;33:1022-7.
C psittaci, serovars A, C, D, and E have been identified 6. Rodriguez Jovita M, Collins MD, Sjoden B, Falsen E. Character-
as human pathogens.28 However, C psittaci has not been ization of a novel Atopobium isolate from the human vagina:
found previously in the vaginal mucosal flora. Two description of Atopobium vaginae sp nov. Int J Syst Bacteriol
other species of the family Chlamydiaceae have 1999;49:1573-6.
7. Collins MD, Jovita MR, Hutson RA, Ohlen M, Falsen E.
been isolated from human vaginal flora. Chlamydia Aerococcus christensenii sp nov, from the human vagina. Int J
trachomatis causes trachoma, sexually transmitted dis- Syst Bacteriol 1999;3(49 Pt):1125-8.
ease, some types of arthritis, neonatal conjunctivitis, and 8. Shukla SK, Bernard KA, Harney M, Frank DN, Reed KD.
pneumonia; Chlamydophila abortus has been found in Corynebacterium nigricans sp nov: proposed name for a black-
sporadic zoonotic infections that cause abortion in pigmented corynebacterium species recovered from the human
female urogenital tract. J Clin Microbiol 2003;41:4353-8.
women who work with sheep.30 However, all of these 9. Morris CE, Bardin M, Berge O, Frey-Klett P, Fromin N, Girardin
species can be distinguished readily by cpn60 sequence H, et al. Microbial biodiversity: approaches to experimental design
(Figure 4). The detection of C psittaci-like cpn60 and hypothesis testing in primary scientific literature from 1975 to
sequences raises the question of the role of C psittaci 1999. Microbiol Mol Biol Rev 2002;66:592-616.
in the vagina and also suggests that the application of 10. Pavlova SI, Kilic AO, Kilic SS, So JS, Nader-Macias ME, Simoes
JA, et al. Genetic diversity of vaginal lactobacilli from women in
molecular methods will likely lead to the identification different countries based on 16S rRNA gene sequences. J Appl
of other organisms that have not been associated pre- Microbiol 2002;92:451-9.
viously with vaginal flora in culture-based studies. 11. Burton JP, Reid G. Evaluation of the bacterial vaginal flora of 20
This study presents an evaluation of a small number postmenopausal women by direct (Nugent score) and molecular
of healthy women and demonstrates the usefulness of (polymerase chain reaction and denaturing gradient gel electro-
phoresis) techniques. J Infect Dis 2002;186:1770-80.
cpn60 sequence-based methods to enhance detailed the 12. Tarnberg M, Jakobsson T, Jonasson J, Forsum U. Identification
evaluation of human vaginal flora. The application of of randomly selected colonies of lactobacilli from normal vaginal
this method resulted in the identification of ‘‘intraspe- fluid by pyrosequencing of the 16S rDNA variable V1 and V3
cies’’ sequence variation that likely would be undetect- regions. APMIS 2002;110:802-10.
able with the use of culture-based or sequence-based 13. Verhelst R, Verstraelen H, Claeys G, Verschraegen G, Delanghe J,
Van SL, et al. Cloning of 16S rRNA genes amplified from normal
methods that use the 16S rRNA target. Clearly, further and disturbed vaginal microflora suggests a strong association
studies of larger populations of women with and with- between Atopobium vaginae, Gardnerella vaginalis and bacterial
out normal flora in concert with technical developments vaginosis. BMC Microbiol 2004;4:16.
to improve the detection of rarer sequences will be 14. Hill JE, Penny SL, Crowell KG, Goh SH, Hemmingsen SM.
needed to expand our understanding of this complex CpnDB: a chaperonin sequence database. Genome Res 2004;14:
1669-75.
microbial community. The specific sequence data that 15. Goh SH, Potter S, Wood JO, Hemmingsen SM, Reynolds RP,
were generated in these studies can be used to quantify Chow AW. HSP60 gene sequences as universal targets for micro-
and monitor individual population members so that bial species identification: studies with coagulase-negative staphy-
their contributions to the function of vaginal microflora lococci. J Clin Microbiol 1996;34:818-23.
can be assessed and understood. 16. Brousseau R, Hill JE, Prefontaine G, Goh SH, Harel J,
Hemmingsen SM. Streptococcus suis serotypes characterized by
analysis of chaperonin 60 gene sequences. Appl Environ Micro-
Acknowledgments biol 2001;67:4828-33.
We thank Jennifer Town for excellent technical assis- 17. Goh SH, Santucci Z, Kloos WE, Faltyn M, George CG, Driedger
D, et al. Identification of Staphylococcus species and subspecies
tance and Tony Rees, RN, for clinical support and using the chaperonin-60 gene identification method and reverse
subject recruitment. checkerboard hybridization. J Clin Microbiol 1997;35:3116-21.
18. Goh SH, Driedger D, Gillett S, Low DE, Hemmingsen SM, Amos
References M, et al. Streptococcus iniae, a human and animal pathogen:
specific identification by the chaperonin-60 gene identification
1. Hillier SL. Diagnostic microbiology of bacterial vaginosis. Am J method. J Clin Microbiol 1998;36:2164-6.
Obstet Gynecol 1993;169:455-9. 19. Goh SH, Facklam RR, Chang M, Hill JE, Tyrrell GJ, Burns EC,
2. Thorsen P, Jensen IP, Jeune B, Ebbesen N, Arpi M, Bremmelgaard et al. Identification of Enterococcus species and phenotypically
A, et al. Few microorganisms associated with bacterial vaginosis similar Lactococcus and Vagococcus species by reverse checker-
may constitute the pathologic core: a population-based microbi- board hybridization to chaperonin 60 gene sequences. J Clin
ologic study among 3596 pregnant women. Am J Obstet Gynecol Microbiol 2000;38:3953-9.
1998;178:580-7. 20. Hill JE, Seipp RP, Betts M, Hawkins L, Van Kessel AG, Crosby
3. Staley JT, Konopka A. Measurement of in situ activities of WL, et al. Extensive profiling of a complex microbial community
nonphotosynthetic microorganisms in aquatic and terrestrial by high-throughput sequencing. Appl Environ Microbiol 2002;68:
habitats. Annu Rev Microbiol 1985;39:321-46. 3055-66.
4. Vasquez A, Jakobsson T, Ahrne S, Forsum U, Molin G. Vaginal 21. Hill JE, Hemmingsen SM, Goldade B, Zijlstra R, Goh SH,
lactobacillus flora of healthy Swedish women. J Clin Microbiol Klassen J, et al. Characterization and quantification of ileum
2002;40:2746-9. microflora of pigs fed corn, wheat or barley-based diets using
692 Hill et al
chaperonin-60 sequencing and quantitative PCR. Appl Environ 27. Pearson WR, Lipman DJ. Improved tools for biological sequence
Microbiol 2005;71:867-75. comparison. Proc Natl Acad Sci U S A 1988;85:2444-8.
22. Ewing B, Hillier L, Wendl MC, Green P. Base-calling of auto- 28. Burton JP, Dixon JL, Reid G. Detection of Bifidobacterium species
mated sequencer traces using phred I: accuracy assessment. and Gardnerella vaginalis in the vagina using PCR and denaturing
Genome Res 1998;8:175-85. gradient gel electrophoresis (DGGE). Int J Gynaecol Obstet
23. Burke J, Davison D, Hide W. D2_cluster: a validated method for 2003;81:61-3.
clustering EST and full-length cDNA sequences. Genome Res 29. Totten PA, Amsel R, Hale J, Piot P, Holmes KK. Selective
1999;9:1135-42. differential human blood bilayer media for isolation of
24. Gordon D, Abajian C, Green P. Consed: a graphical tool for Gardnerella (Haemophilus) vaginalis. J Clin Microbiol 1982;15:
sequence finishing. Genome Res 1998;8:195-202. 141-7.
25. Rice P, Longden I, Bleasby A. EMBOSS: the European Mole- 30. Everett KD, Bush RM, Andersen AA. Emended description of the
cular Biology Open Software Suite. Trends Genet 2000;16:276-7. order chlamydiales, proposal of Parachlamydiaceae fam nov and
26. Thompson JD, Higgins DG, Gibson TJ. Clustal W: improving Simkaniaceae fam nov, each containing one monotypic genus,
the sensitivity of progressive multiple sequence alignment through revised taxonomy of the family Chlamydiaceae, including a new
sequence weighting, position-specific gap penalties and weight genus and five new species, and standards for the identification of
matrix choice. Nucleic Acids Res 1994;22:4673-80. organisms. Int J Syst Bacteriol 1999;49:415-40.
www.ajog.org
A national probability survey of American Medical

Association gynecologists and primary care
physicians concerning menopause
Betsy Singh, PhD,a,* Xiao-Dong Liu, PhD,a Claudia Der-Martirosian, PhD,a
Mary Hardy, MD,b Vijay Singh, BS,a Neil Shepard, MA,c Sonal Gandhi, MD,a
Raheleh Khorsan, MAa
Southern California University of Health Sciences, Whittier, Calif,a University of California, Los Angeles,
Los Angeles, CAb; Bowling Green State University, Bowling Green, OH c
Received for publication August 9, 2004; revised February 14, 2005; accepted February 14, 2005
KEY WORDS Objective: This survey intended to clarify physicians’ understanding of the issues surrounding
Survey women, menopause, alternative medicine, and drug therapy for the treatment of menopause.
Complementary and Study design: This study was designed as a national probability sample survey of primary care
alternative physicians and gynecologists nationwide. Its focus was to identify major concerns and issues
medicine identified by patients about menopause and perceived communication with effectiveness how to
Traditional therapy communicate with their patients. Physicians were also asked to rate their comfort level in
Menopause recommending the use of herbal remedies and which herbal remedy they felt comfortable
Health care recommending to interested patients.
professionals Results: Data indicated that a patient’s complaint about menopausal symptoms was the most
common factor leading to discussion of menopausal issues with physicians (91%) and that the
primary concern to the patient was management of menopausal symptoms. Other factors were
controversies about hormone replacement therapy, long-term health implications of menopause,
and hormone replacement therapy. Eighty percent of the physician found confusing messages
with regard to menopause to be the most challenging aspect in patient communication. The
second most challenging issue is ‘‘inconclusive data about hormone replacement therapy’’ (56%).
Seventy-six percent of the physicians found ‘‘showing sympathy’’ to be the most important factor
for the physicians to communicate effectively with patients, whereas ‘‘being honest and open’’ was
the most important patient attitude cited for the same purpose. When it comes to herbal therapy
for menopause symptom control, only 4% of the physicians indicated that none of their patients
take any remedies. Only 18% were not very comfortable in discussing or recommending herbal
therapies, whereas the rest ranged from fairly comfortable to completely comfortable.
Conclusion: This study has provided data with regard to physician understanding of menopause
treatment options and their primary interaction with patients on this issue. More in-depth studies
concerning efficacy and/or side effects of each available treatment will be the relevant next step,
given the controversies about both hormone replacement therapy and alternative therapies. The
* Reprint requests: Dr. Betsy B. Singh, Professor and Dean of Research, Southern California University of Health Sciences, 16200 East Amber
Valley Drive, Whittier, CA 90604.
E-mail: betsysingh@scuhs.edu
0002-9378/$ - see front matter Ó 2005 Elsevier Inc. All rights reserved.
doi:10.1016/j.ajog.2005.02.080
694 Singh et al
relative efficacy, safety, and cost-effectiveness of different treatments should also be put into the
context of both clinical diagnosis and physicians’ clinical judgment. Attention to comments by
physicians and patients with regard to communication may produce better information exchange
and trust between patient and physician.
Ó 2005 Elsevier Inc. All rights reserved.
Menopause is a universal phenomenon that typically prolonged estrogen exposure to the corresponding
occurs between the ages of 40 and 58 years in the Western tissues.13 The addition of progestin (or other progester-
world, with an average age of 51 years.1 Factors that one analogs) to estrogen in the so-called HRT protects
influence the course, timing, duration, level, and severity against estrogen-induced endometrial hyperprolifera-
of associated symptoms or health risks, from symptoms tion, which is linked to malignancy. Yet prolonged use
as commonly experienced as hot flashes and night sweats of HRT with combined progesterone and estrogen
to osteoporosis and cardiovascular diseases1,2 are: age (more than 5 to 10 years) is still associated with
of onset, the cause of menopause (natural or induced), increased risk of breast and endometrial cancers.10,12
health and nutritional status, physical fitness, sexual As a remedy in response to postmenopausal drop of
activity, emotional state, stress level, use of medication, testosterone, testosterone has also been included in
or alternative supplements.1,3-8 Perimenopausal symp- HRT to increase muscle strength, appetite, physical
toms, a transition phase that immediately precedes true well-being, and sexual desire.7
menopause and is marked by irregular menstrual periods, In recent years a new treatment regimen has been
include vasomotor responses such as hot flashes, night developed: selective estrogen receptor modulators
sweats, headaches, palpitations, and insomnia; and neu- (SERMs). SERMs have the advantage of combining
rological or mental responses such as irritability, depres- the positive skeletal protective effects of estrogen with
sion, anxiety, mood swings, and memory loss.2,7 Other estrogen-antagonistic effects on sex tissues as to reduce
manifestations include joint and muscle pain, dryness of the risk of breast cancer.14,15
skin and vagina, dyspareunia, and increased facial hair. There has been rising use of alternative therapies,
Most of the symptoms wane over time but can persist in a especially herbals, in recent years. Commonly used
small percentage of females. herbals include black cohosh for hot flashes, night sweats,
Women’s attitudes toward menopause range from insomnia, and vaginal dryness; valerian for sleep distur-
feeling relieved, happy, and having more control over bance; chaste berry for heavy menses and hot flashes;
health to anxious, depressed, or plagued by physical and dong quai for easing common menopausal symptoms;
mental aggravation brought on by prolonged sympto- Gingko biloba for memory loss; licorice for support of
matic periods.9 Regardless, the long-term consequence adrenal gland in producing low level of sex hormones; St.
of reduced estrogen, progesterone, and other female John’s wort for depression; red clover, soybeans, and
hormones following menopause, complicated by the flaxseeds as sources of isoflavones/phytoestrogens, which
natural course of aging, are inevitably linked to more has estrogen-like effects; and wild yam for regulating
severe health conditions such as osteoporosis and car- progesterone/estrogen ratio in body.6,7,16,17 Use of
diovascular diseases.10 Some factors that play important phytoestrogens are an important option according to
roles in long-term health in postmenopausal years research.16,18
are: balanced nutrition, especially adequate intake of Facing the overwhelming available treatment options,
calcium, vitamin D, and vitamin E; physical activity, physicians have traditionally been playing the most
particularly weight-bearing exercises5,11; use of hormone significant role in channeling the appropriate information
replacement therapy; and use of herbal supplements.7 to female patients concerning menopause, including the
Estrogen or hormone replacement therapy (HRT) discussed modifiable lifestyle-related risk or benefits,
has long been used to help relieve menopause-related supplement use, and HRT. Low compliance has been
symptoms such as hot flashes and night sweats. Long- known for HRT, especially in cases in which strong side
term usage has also been shown to decrease serum low- effects are encountered.19,20 Additional reasons include
density lipoprotein and cholesterol and reduce risk of risks of cancer associated with long-term use and lack of
heart disease, osteoporosis, colorectal cancer, Alzheimer’s long-term study of newer HRT alternatives such as
disease, and macular degeneration and delay aging of SERMs.22 Under such circumstances, there has been a
skin.1,10,12 Common side effects include bloating, breast rapidly increasing use of alternative medicine among
tenderness, irritability, depression, and sometimes pro- postmenopausal women for various symptom relief.6
longed menstrual bleeding. Furthermore, long-term use However, it has been shown that physicians are poor
of estrogen replacement has also been known to increase at providing information on complementary and alterna-
risk of breast cancer and endometrial cancer because of tive therapies to ease postmenopausal symptoms and
Singh et al 695
diseases.21,22 There have been limited regional studies on mean years in practice were 23.5, and more than two
these subjects,8,9,21,23,24 whereas national probability thirds of the sample (67%) were male physicians.
sample surveys are still scarce. Relevant information Almost half (49%) worked in group practices, 38%
will greatly help to guide researchers and policy makers had a solo practice, and 11% worked in a hospital
to steer relevant research directions. setting. Size of practice was measured by number of
This study was designed to facilitate more under- patients seen daily. The average number was 20 patients.
standing of menopausal topics and communication The mean daily working hours were 7.6.
between patients and physicians by carrying out a
national probability survey with a hierarchical sample Patient profiles and general attitudes and
drawn proportionally from 2 American Medical Asso- concerns about menopause
ciation membership groups (primary care physicians
The average age of menopausal women in the surveyed
and gynecologists) and by geographical region of the
practice was 56 years, according to the responses.
United States.
Physicians were also asked to report the percentage of
menopausal women who were on HRT. Only 1% of the
Methods physicians surveyed indicated that none of their patients
were on HRT. Thirty-one percent indicated that 1% to
The survey sample was from the American Medical 48% of their patients were on HRT. Twelve percent of
Association membership list (2000) of family practice, physicians reported that 50% of their patients were on
general practice, and internal medicine physicians. Using HRT. Almost one quarter reported between 51% and
a power calculation set at a 95% confidence level and 75% being on HRT. The remaining 40% of the sample
a confidence interval of 4 (G 4) a sample of 1137 was indicated that 76% to 100% of their patients were on
required. A stratified hierarchal sampling technique was HRT. Overall, physicians reported a high compliance
adopted. Multiple waves of mailed questionnaires were rate. The majority of physicians (92%) indicated that
sent out. The questionnaires were administered from May their patients take products as prescribed.
2001 to September 2001. Table I lists the most common factors that lead to
Two hundred seventy-two physicians responded to discussion of menopause. Menopausal symptom relief is
the mail questionnaire. The physician response rate was of great concern to most women (91%), and 95% of the
24%. The response was lower than desired. However, physicians indicated that they get unsolicited questions
physicians constitute a relatively homogenous group, concerning menopause from asymptomatic patients.
compared with the general public, and therefore the Eighty-three percent of physicians bring up menopause
required sample size to ensure external validity is to all patients of a certain age. Seventy-four percent of
smaller.25,26 patients have questions about HRT; 63% have ques-
The questionnaire consisted of 23 questions covering tions about information they received from friends,
various issues about menopause. Respondents were family, and the media; and 55% of patients have
asked to discuss major concerns and issues that they questions about herbal supplements or over the counter
receive from patients about menopause. Several ques- (OTC) preparations.
tionnaire items addressed the ways in which physicians Table II ranks the top 3 concerns and/or questions
can communicate effectively with their patients. Physi- patients have about menopause. At the top of the list,
cians were also asked to rate the qualities for effective managing menopause symptoms ranks first at 82%.
patient communication and discuss how patients can Questions about HRT were ranked second at 50%, with
communicate more effectively. The study also addressed long-term health implications at third place at 33%.
recommendations for treatment of menopause symp- The following factors rated 16% or below at second or
toms. Respondents were asked to rate their comfort level third rank: weight management, quality-of-life concerns,
in using herbal remedies and to indicate which herbal herbal supplements and OTC supplements, other alter-
remedy they would feel comfortable recommending to native therapy, and reactions to information from the
interested patients. The last set of questions addressed media.
the quality of information available to patients regarding
menopause and alternative treatment. Challenges to communication
In the survey, physicians were asked to indicate which of
the following (listed in Table III) poses challenges to
Results
communicating effectively with patients. ‘‘Confusing
Physician profiles messages from media’’ (80%) was the most commonly
reported challenge. The second most common response
Among the 128 primary care physicians and 144 was ‘‘inconclusive data about HRT’’ (56%). At third and
gynecologists who responded to the questionnaire, the fourth were ‘‘complicated issues surrounding treatment
696 Singh et al
Table I Regular checkup: genesis of discussion about meno- Table II Top three concerns/questions from patients about
pause menopause
Patient complains about symptoms 91% Ranked first
I bring it up with all my patients who are 83% Specific menopause symptom 82%
a certain age management/relief
Patient questions about hormonal replacement 74% Hormone replacement therapy 8%
therapy Weight management 5%
Patient questions about information they received 63% Long-term health implications 3%
from friends, media, family Ranked second
Patient seeking information on herbal 55% Hormone replacement therapy 50%
supplements and over-the-counter treatments Long-term health implications 18%
I don’t bring it up unless the patient raises 3% Weight management 16%
a concern Herbal supplements and over-the-counter 7%
Other 10% treatments
Reactions to news stories, information on 6%
the Internet
options’’ (45%) and ‘‘physician time constraints’’ (43%). Ranked third
A ‘‘lack of available information about treatment op- Long-term health implications 33%
tions’’ and ‘‘patients are not open,’’ rated at 27% and Reactions to new stories, information on 21%
23%, respectively. Seventeen percent or less of physicians the Internet
Quality-of-life issues 13%
ranked the next 3 statements as a challenge: ‘‘patients
Herbal supplements and over-the-counter 11%
appear to manage menopause on their own,’’ ‘‘physi-
treatments
cians’ lack of knowledge about alternative treatments,’’ Hormone replacement therapy 9%
and ‘‘patients are embarrassed to ask questions.’’ Alternative therapy other than herbal options 6%
Table IV ranks the elements of physician attitudes
that are important for effective communication with
patients. Seventy-six percent of the physicians agreed
that showing sympathy is very important. Providing Table III Challenges to effective communication between
useful information to help the patient make decisions physicians and patients
was reported to be very important by 66% of the sample. Confusing messages from media, health 80%
Half of the physicians rated ‘‘allowing plenty of time to care professionals, and friends
discuss the patients’ concerns’’ and ‘‘not interrupting Inconclusive data about hormone 56%
replacement therapy
when the patient is speaking’’ as being very important.
Complicated issues surrounding 45%
‘‘Probing physical and emotional symptoms carefully’’
treatment options
(49%) and ‘‘making the patient feel her experiences are My time constraints 43%
normal’’ (45%) were also considered very important. Lack of available information about 27%
The last column of Table IV represents physicians’ alternative treatment options
self-rating with respect to the same set of qualities. Half Patients are not open about what they 23%
of the physicians rated themselves ‘‘excellent’’ for show- know/don’t know about menopause
ing sympathy toward their patients, and 42% of the Patients appear to be managing their 17%
physicians rated themselves as ‘‘excellent’’ on providing menopause on their own
useful information to help the patient make decisions. My lack of knowledge about alternative 17%
The range of ‘‘excellent’’ rankings for self-rating of treatment options
Patients are embarrassed to ask questions 14%
physicians was 50% for believing that they showed
sympathy and 24% for believing that they do not
interrupt when a patient was speaking.
Table V shows how physicians responded to questions considered to be important by 27% of the physicians.
addressing patient attitudes in facilitating discussion. In Similarly, having a list of goals and concerns was consid-
terms of the elements of patient attitudes that are impor- ered not important for 16% of the sample.
tant for effective communication, 76% of physicians Physicians were also asked to rate their patients for
reported ‘‘being honest and open’’ as a very important each of these communication qualities, as listed in the
communication quality from patients. To ‘‘understand last column of Table V. Only 17% of the physician
why a certain therapy is recommended’’ was ranked sample gave the patients an ‘‘excellent’’ rating for being
second by physicians (58%). Sufficient time to discuss ‘‘honest and open.’’ The range for physicians’ rating
questions and concerns (55%) was ranked third. Sharing their patients as excellent was 19% for ‘‘willing to trust
information read, heard, or seen on the Internet was not my advice’’ and 9% at ‘‘having a list of goals and
Singh et al 697
Table IV Important physician qualities for good physician/patient communication and physician self-rating of behavior
Physician
Very Not self-rating
important Important important ‘‘excellent’’
(%) (%) (%) (%)
Showing sympathy 76 21 2 50
Providing useful information to help the patient make decisions 66 32 2 42
Allowing plenty of time to discuss the patient’s concerns 50 47 2 28
Not interrupting when a patient is speaking 50 38 9 24
Probing physical and emotional symptoms carefully 49 47 3 34
Making the patient feel her experience is normal 45 48 6 35
Table V Important patient qualities for good physician/patient communication and physician rating of patients’ behavior
Very Not Physician
important Important important rating of patient
(%) (%) (%) ‘‘Excellent’’ (%)
Be honest and open 76 23 0.8 17
Have a list of goals and concerns 35 47 16 9
Share information read, heard, or seen on the Internet 21 39 27 9
Be willing to trust me and follow my advice 33 45 12 19
Understand why a certain therapy is recommended 58 40 1 15
Sufficient time to discuss questions and concerns 55 42 2 d
concerns’’ and ‘‘sharing information from media.’’ This Table VI Types of herbal remedies and/or supplements
demonstrates a discrepancy in physicians’ attitude to- recommended by physicians
ward their own communication skills and their patients’
Soy products 62%
communications skills.
Vitamin E 52%
Recommendations for herbal remedies Black cohosh (RemiFemin) 31%
St. John’s wort 29%
When asked what percentage of menopausal women Evening primrose berry 22%
takes some type of herbal remedy or OTC treatment to Red clover (Promensil) 16%
relieve symptoms, only 4% of the physicians indicated Combination products (Estroven) 13%
Valerian 11%
that none of their patients take herbal remedies. Forty-
Dong quai 7%
four percent of the physicians indicated that less than Chaste tree berry 5%
10% of their patients take herbal remedies. Twenty-three
percent of the physicians indicated that 11% to 20% take
herbal remedies. The remaining physicians indicated that
in purchasing herbals. The 2 most common responses
more than 21% of their patients take herbal remedies.
were that patients ‘‘learn about ingredients and their
Physicians were also asked how comfortable they were
safety/effectiveness’’ and ‘‘carefully read labels.’’
discussing or recommending herbal remedies. Only 16%
reported being ‘‘completely comfortable.’’ Thirty percent Sources and quality of menopause treatment
of the physicians reported being ‘‘very comfortable,’’ 32%
reported being ‘‘fairly comfortable,’’ and the rest reported The majority of physicians (81%) would feel more com-
being ‘‘not very comfortable.’’ fortable in discussing or recommending herbal remedies
Among the herbal remedies that physicians are com- to patients if there were more information from clinical
fortable recommending, soy products (62%) and vitamin trials and other studies about herbal remedies. For these
E (52%) were on the top of the list. Black cohosh (31%), respondents, information about the following items
St. John’s wort (29%), and evening primrose berry would be most helpful: safety (82%), efficacy (73%),
(22%) ranked next. Red clover, combination products dosing (64%), effectiveness (61%), quality (49%), and
(Estroven), valerian, dong quai, and chaste tree berry all reproducibility (42%).
ranked at or below 16% (Table VI). When asked to rank the top 3 qualities that are
Physicians were also asked what course of action most important when recommending any treatment
they would recommend to patients who are interested for menopause, safety was ranked first by 65% of the
698 Singh et al
Table VII Physicians’ report on patients’ source of information on menopause and quality of information on herbal/OTC menopause
supplements
Physician thinks Excellent, Very good, Satisfactory, Not good,
Source of information patients rely on info, % % % % %
Internet 44 0 15 41 33
Magazine articles 70 0 8 54 32
Pamphlets in pharmacy/doctor office 30 4 40 37 14
Newspaper 74 0 6 40 45
Friends/family 85 0 3 23 57
Television 57* 0 3 31 53
Advertisements in magazines 0 3 31 52
Advertisements in journals 0 12 52 32
Journal articles 0 8 33 43 10
Doctor 61 13 37 39 9
Other health care professionals 36y 2 30 49 15
* Question for patients’ source of information combined television or print advertisements.
y
Of the 36%, 14% was from the subfield of ‘‘nurse.’’
respondents. Being supported by science or research was ceive patients relying on most as being of poor quality,
ranked second by 38%, and secondarily providing pro- suggesting that the physicians may view many patients as
tection from other diseases was ranked third by 39% of misinformed about menopause.
the physicians.
In terms of the satisfaction rating of physicians of Future concerns for menopause
currently available treatments or remedies for menopause symptom management
symptoms, approximately 34% of the physicians reported
being ‘‘somewhat satisfied,’’ and only 13% reported being When physicians were asked what improvements they
‘‘very satisfied.’’ The rest of the sample reported being would like to see in menopause symptom management
‘‘somewhat dissatisfied’’ (31%) and 16% ‘‘very dissatisfied.’’ remedies, responses were divergent. The majority focused
Sources of information about menopause and treat- on the following few issues: (1) better understanding of the
ments were the last topic of the questionnaire. Physicians benefits and side effects of HRT and SERMs; HRT
were asked to indicate where they thought patients get alternatives with fewer side effects (especially the risk of
their information on menopause. The most common breast cancer) and the same potency; (2) better studies in
response was friends or family (85%). The second and alternative treatment effectiveness that would limit unsub-
third most common response was newspaper articles or stantiated claims; (3) more research in alternative medicines
news programs (74%) and magazines (70%). Doctors with more holistic approaches that take into account the
were ranked fourth as a source of information at 61%. condition of the whole person, instead of just symptoms;
Media, such as television or print advertising (57%) or the (4) research-based guidelines to individualize or customize
Internet (44%), were also ranked by physicians. Pam- treatment options and dosage for different patients; and
phlets in doctors’ offices and other health care profes- (5) better patient education for efficacy and safety.
sionals ranked less than 30%. When asked when and how herbal treatments or OTC
Rating of the quality of these sources of information remedies are recommended or discussed, physician re-
was the last item on the questionnaire. The ‘‘doctor’’ sponses were more unanimous: herbal therapies or OTC
category received an ‘‘excellent’’ rating (13%). Only treatments are discussed when HRT is contraindicated
1 source of information, pamphlets in a pharmacy or (eg, breast cancer, adverse reaction); a patient is reluctant
doctor’s office, received a ‘‘very good’’ rating from 40% of in taking HRT; or a patient brings up the topic.
the sample. The majority of the categories received ratings
of ‘‘satisfactory’’ or ‘‘not good.’’ Magazine articles (54%), Differences in responses between gynecologists
advertisements in journals (52%), other health care and primary care physicians
professionals (49%), journal articles (43%), the Internet
(41%), and doctors (39%) received satisfactory ratings When comparing the responses to the SENSEI ques-
from physicians. Friends and family (57%), television tionnaire, some notable differences were found between
(53%), advertisements in magazines (52%), and news- the 2 groups (primary care versus gynecology physi-
papers (45%) received ratings of ‘‘not good’’ (see Table cians). First, there were significant differences in average
VII). It is interesting to note that physicians consider age. Gynecologists were older (mean 68 years), and the
those sources of information that they commonly per- mean age for primary physicians was 48 years (P ! .0001).
Singh et al 699
Given the older age of the gynecologists, they had percent of physicians found ‘‘showing sympathy’’ to be
been in practice for a longer time (27 years versus 18 the most important factor for physicians to communicate
years for primary). This difference was significant at effectively with patients, whereas patients being ‘‘being
P ! .0001. There were significant differences in type honest and open’’ was the most important patient attitude
of practice as well, with 56% of primary care physicians for the same purpose.
in group practice and 18% in hospital based. For When it comes to herbal therapy for menopause
gynecologists 90% were either in solo or group practice symptom control, only 4% of the physicians indicated
(P = .0016). With regard to number of hours in prac- that none of their patients take any remedies. Only18%
tice, average age of menopausal women in their practice of physicians are not very comfortable in discussing or
or whether the majority of patients take products as recommending herbal therapies, whereas the rest range
prescribed, no significant differences were found. For from fairly comfortable to completely comfortable. Soy
gynecologists, more patients reported being on a hor- products turned out to be the most commonly recom-
mone replacement product than primary care physicians mended herbal remedies by physicians for menopause
(P = .0003). (62%) and vitamin E the second (52%). Eighty-one
In terms of self-rating, gynecologists consistently percent of physicians would feel more comfortable in
scored higher on showing empathy, allowing plenty of recommending herbal products to patients if there were
time to discuss patients’ concerns, and making the patient more information from clinical trials or other studies.
feel her experience is normal. When asked about ‘‘when Safety was ranked as the most important quality (65%)
recommending treatment for menopause symptoms, when recommending herbals to patients, and support by
what are the top 3 qualities that are most important to research was ranked the second (38%). Only 13% of the
you?’’ no significant differences were found. Gynecolo’ists physicians are ‘‘very satisfied’’ by the currently available
reported being more satisfied with available remedies/ treatment options for menopause, and 34% are ‘‘some-
treatments for menopause symptom management what satisfied.’’ In the physicians’ opinion, patients most
(P = .0024). In terms of how comfortable they feel likely get their information on menopause from friends
discussing or recommending herbal remedies, no signif- and family, newspapers, and magazines.
icant differences were found between primary care physi- There were 2 open-ended questions soliciting opin-
cians and gynecologists. In terms of what course of action, ions on what improvement they would like to see in
if any, they would recommend to patients in purchasing menopause management and when they would recom-
herbals, no significant differences were found in their mend herbal remedies. The answers to the first question
responses. were boiled down to a few key issues: better under-
standing of the pros and cons of HRT and SERMs,
better HRT alternatives, and more research on alterna-
Comment tive medicines. The physicians were more in agreement
about the second question: They most likely recommend
In this national probability sample study, we investi- herbal products when HRT is contraindicated. This
gated the opinions of gynecologists and primary care study has provided data concerning physician under-
physicians with regard to menopause. Findings in this standing of menopause treatment options and their
study presented primary evidence of physicians’ first- primary interaction with patients on this issue.
hand experience with menopause issues in daily practice Some limitations to this study need to be addressed.
as well as their opinions on a variety of menopause- First, a larger response rate would have been preferable,
related topics. It also allowed physicians a chance to although cited literature indicates that high compliance is
reflect on patient needs and attitudes toward menopause unusual among physicians. Second, this survey was lim-
and to have their findings collectively analyzed statisti- ited in scope by addressing only gynecologists and
cally. It will serve as an important feedback tool to guide primary care physicians for reasons of cost. Other
physicians’ communication with patients with regard to members of the medical field are solicited for advice by
menopausal issues. women facing menopause. Their attitudes about their
Data indicated that patients’ complaint about meno- patients, menopause, and menopausal treatment would
pausal symptoms is the most common lead to discussion have added to the generalizability of physicians’ attitudes
of menopausal issues with physicians (91%), and that of and behaviors toward menopausal issues reported here.
top concern to patients are management of menopausal Regardless, more in-depth studies regarding efficacy
symptoms, controversies around HRT, and long-term and/or side effects of each available treatment will be the
health implications of menopause and HRT. Confusing relevant next step, given the controversies around both
messages with regard to menopause was found by 80% of HRT and alternative therapies. The relative efficacy,
the physicians to be the most challenging aspect in patient safety, and cost-effectiveness of different treatments
communication. The second most challenging issue is should also be put into the context of both clinical
‘‘inconclusive data about HRT’’ (56%). Seventy-six diagnosis and physicians’ clinical judgment.
700 Singh et al
References 15. Komm BS, Lyttle CR. Developing a SERM: stringent preclinical
selection criteria leading to an acceptable candidate (WAY-
1. The North American Menopause Society (TNAMS). Menopause 140424) for clinical evaluation. Ann N Y Acad Sci 2001;949:
guidebook: helping women make informed healthcare decisions 317-26.
through perimenopause and beyond. Cleveland (Ohio): North 16. Seidl MM, Stewart DE. Alternative treatments for menopausal
American Menopause Society; 2001. symptoms. Systematic review of scientific and lay literature.
2. Sheehy G. Menopause: the silent passage. New York: Simon & Can Fam Physician 1998;44:1299-308.
Schuster; 1991. 17. McKenna DJ, Jones K, Humphrey S, Hughes K. Black cohosh:
3. Clinical challenges of perimenopause: consensus opinion of The efficacy, safety, and use in clinical and preclinical applications.
North American Menopause Society. Menopause 2000;7:5-13. Altern Ther Health Med 2001;7:93-100.
4. Freedman RR, Woodward S. Behavioral treatment of menopausal 18. Tham DM, Gardner CD, Haskell WL. Clinical review 97: poten-
hot flushes: evaluation by ambulatory monitoring. Am J Obstet tial health benefits of dietary phytoestrogens: a review of the
Gynecol 1992;167:436-9. clinical, epidemiological, and mechanistic evidence. J Clin Endo-
5. Hammar M, Berg G, Lindgren R. Does physical exercise influence crinol Metab 1998;83:2223-35.
the frequency of postmenopausal hot flushes? Acta Obstet Gynecol 19. Reynolds RF, Walker AM, Obermeyer CM, Rahman O,
Scand 1990;69:409-12. Guilbert D. Discontinuation of postmenopausal hormone ther-
6. Ewies AA. A comprehensive approach to the menopause: so far, apy in a Massachusetts HMO. J Clin Epidemiol 2001;54:
one size should fit all. Obstet Gynecol Surv 2001;56:642-9. 1056-64.
7. Montemuro S. Relieving the symptoms of menopause: from herbs 20. Reynolds RF, Obermeyer CM, Walker AM, Guilbert D. Side
to hormones. B C Med J 2001;43:452-7. effects and sociobehavioral factors associated with the discontin-
8. Conboy L, Domar A, O’Connell E. Women at mid-life: symptoms, uation of hormone therapy in a Massachusetts health maintenance
attitudes, and choices, an Internet based survey. Maturitas organization. Menopause 2001;8:189-99.
2001;38:129-36. 21. Vashisht A, Domoney CL, Cronje W, Studd JW. Prevalence of
9. Utian WH, Boggs PP. The North American Menopause Society and satisfaction with complementary therapies and hormone
1998 Menopause Survey. Part I: postmenopausal women’s per- replacement therapy in a specialist menopause clinic. Climacteric
ceptions about menopause and midlife. Menopause 1999;6:122-8. 2001;4:250-6.
10. Burkman RT, Collins JA, Greene RA. Current perspectives on 22. Gordon NP, Sobel DS, Tarazona EZ. Use of and interest in
benefits and risks of hormone replacement therapy. Am J Obstet alternative therapies among adult primary care clinicians and adult
Gynecol 2001;185(2 Suppl):S13-23. members in a large health maintenance organization. West J Med
11. Hu FB, Stampfer MJ, Manson JE, Grodstein F, Colditz GA, 1998;169:153-61.
Speizer FE, et al. Trends in the incidence of coronary heart disease 23. Factor-Litvak P, Cushman LF, Kronenberg F, Wade C, Kalmuss
and changes in diet and lifestyle in women. N Engl J Med D. Use of complementary and alternative medicine among women
2000;343:530-7. in New York City: a pilot study. J Altern Complement Med
12. Gaspard U. Risks, benefits and costs of hormone replacement 2001;7:659-66.
therapy in menopause. Rev Med Liege 1998;53:298-304. 24. Seidl MM, Stewart DE. Alternative treatments for menopausal
13. Ariyo AA, Villablanca AC. Estrogens and lipids. Can HRT symptoms. Qualitative study of women’s experiences. Can Fam
designer estrogens, and phytoestrogens reduce cardiovascular risk Physician 1998;44:1271-6.
markers after menopause? Postgrad Med 2002;111:23-30; quiz 23. 25. Evans RW, Evans RI, Sharp MJ. The physicians survey of post-
14. Pfeilschifter J. Hormone replacement and selective estrogen recep- concussion and whiplash syndromes. Headache 1994;34:268-74.
tor modulators (SERMS) in the prevention and treatment of 26. Babbie E. The practice of social research, 7th ed. Belmont (CA):
postmenopausal osteoporosis. Orthopade 2001;30:462-72. Wadsworth; 1994.
www.ajog.org
GENERAL OBSTETRICS AND GYNECOLOGY: OBSTETRICS
A randomized trial of amnioreduction versus septostomy in

the treatment of twin-twin transfusion syndrome
Kenneth J. Moise Jr, MD,a,* Karen Dorman, MS,a Georgine Lamvu, MD, MPH,a George
R. Saade, MD,b Nicholas M. Fisk, MD, PhD,c Jan E. Dickinson, MD,d R. D. Wilson, MD,e
Alain Gagnon, MD,e Michael A. Belfort, MD,f Richard O. O’Shaughnessy, MD,g
Usha Chitkara, MD,h Sonia S. Hassan, MD,i Anthony Johnson, DO,i Anthony
Sciscione, DO,j Daniel Skupski, MDk
Departments of Obstetrics and Gynecology of the University of North Carolina School of Medicine, Chapel Hill, NCa;
University of Texas Medical Branch at Galveston, Galveston, TX b; Queen Charlotte’s and Chelsea Hospital & Imperial
College London, London, United Kingdomc; King Edward Memorial Hospital for Women, Mumbai, Indiad; University of
British Columbia, Vancouver, British Columbia, Canadae; University of Utah School of Medicine, Salt Lake City, UT f;
Ohio State University, Columbus, OH g; Stanford University Medical Center, Stanford, CAh; Wayne State University,
Ann Arbor, MI i; Christiana Hospital, Newark, DEj; and New York Hospital Weill Cornell Medical Center, New York,
NY k
Received for publication October 21, 2004; revised January 6, 2005; accepted January 25, 2005
KEY WORDS Objective: Left untreated, severe twin-to-twin transfusion syndrome (TTTS) presenting in the
Twin-to-twin early second trimester of pregnancy is often associated with significant maternal morbidity and
transfusion almost universal perinatal loss. Removal of excessive amounts of amniotic fluid through serial
syndrome amniocenteses (amnioreduction) has been the mainstay of therapy. We sought to compare
TTTS amnioreduction to intentional perforation of the intervening twin membrane (septostomy).
Amnioreduction Study design: Pregnant women with TTTS before 24 weeks’ gestation were randomly assigned to
Septostomy serial amnioreduction or septostomy. A single puncture technique under ultrasound guidance was
used for the septostomy. The primary outcome measure was survival to neonatal discharge, and
was assessed based on the number of pregnancies or the number of fetuses as appropriate.
Results: The study was terminated at the planned interim analysis stage after 73 women were
enrolled. This was because the rate of survival of at least 1 infant was similar in the
amnioreduction group compared to the septostomy group (78% vs 80% of pregnancies,
respectively; RR = 0.94, 95%CI 0.55–1.61; P = .82). Patient undergoing septostomy were more
likely to require a single procedure for treatment (64% vs 46%; P = .04).
Conclusion: Although overall perinatal survival is not enhanced, septostomy offers the advantage
of often requiring a single procedure compared to serial amnioreduction in the treatment of severe
twin-to-twin transfusion syndrome.
* Reprint requests: Kenneth J. Moise Jr, MD, 214 MacNider Bldg, CB #7516, University of North Carolina School of Medicine, Chapel Hill, NC
27599-7516.
E-mail: kmoisejr@med.unc.edu
doi:10.1016/j.ajog.2005.01.067
702 Moise et al
Although monochorionic twin placentation occurs in Women with monochorionic twin gestations at less
only one fifth of all twin gestations, they account for the than 24 weeks’ gestational age with ultrasound evidence
majority of the resultant perinatal morbidity and of TTTS were offered enrollment in the trial. Mono-
mortality. Approximately 15% are complicated by chorionicity was defined as the presence of concordant
discordance of amniotic fluid volumedthe ‘‘donor’’ fetal sex, a single placenta, and the lack of a projection
fetus becomes ‘‘stuck’’ with minimal to absent amniotic of the chorionic tissue at the interface between the
fluid, while its ‘‘recipient’’ sibling is surrounded by intertwin membrane and the chorionic surface. Chorio-
excessive amount of fluid.1 This condition has been nicity was confirmed after delivery by examination of
described as twin-to-twin transfusion syndrome (TTTS), the placenta. Standard inclusion criteria for TTTS
reflecting the proposed causation of unequal blood included polyhydramnios in 1 amniotic cavity (deepest
volumes mediated by unbalanced intertwin transfusion vertical pocket of O8 cm at less than 20 weeks’
through placental anastomoses. Monochorionic placen- gestation, O10 cm at 20 to 22 weeks’ gestation, and
tas that result in TTTS are less likely to have compen- O12 cm after 22 weeks’ gestation), and oligohydram-
satory bidirectional vascular channels, and recent nios in the second amniotic cavity (deepest vertical
reports document resolution of the syndrome after laser pocket of !2 cm).
ablation of deep unilateral intraplacental anastomo- Exclusion criteria were fetal structural anomalies,
ses.2,3 premature contractions associated with cervical change,
Left untreated, severe TTTS presenting in the second premature rupture of the membranes, suspected cho-
trimester has been associated with a perinatal survival of rioamnionitis, or the presence of other indications for
0% to 30%.4,5 Until recently, serial amnioreduction has delivery.
been the mainstay of treatment, with survival rates of Computer randomization was undertaken in advance
approximately 60%.6 Amnioreduction aims to normalize utilizing a block size of 10. No stratification was used.
amniotic fluid volume and reduce the rate of preterm Each participating center was given access to a secure
labor and amniorrhexis, and there is some evidence that it web site for randomization as previously described.10
may improve uteroplacental perfusion and, on occasion,
lead to regression of the TTTS.7 In 1998, Saade et al8 Amnioreduction procedure
proposed the concept of intentionally perforating the
intertwin membrane (septostomy) to allow equilibration Amnioreductions were performed using real-time ultra-
of the amniotic fluid volume in severe cases of TTTS. sound guidance within the recipient’s amniotic sac. An
Indeed, allowing the donor twin to swallow excess effort was made to avoid the intertwin membrane and,
amniotic fluid produced by the recipient is one explana- thus, the gestational sac of the donor twin so as not to
tion for the extreme rarity of TTTS in monoamniotic perform an unintentional septostomy. An 18-gauge
twins. In a feasibility study of 12 women treated with needle was introduced into the amniotic cavity with
septostomy at 5 centers, the overall perinatal survival was polyhydramnios, and then connected to either wall
86%, with a mean gain of 8.3 weeks in utero. All suction or a syringe attached to extension tubing in an
procedures were successful on the first attempt. In only effort to actively drain the amniotic fluid, thereby
3 of the 12 cases was an amnioreduction necessary reducing the total procedure time. Fluid was removed
concurrently with the septostomy; a single amnioreduc- until the deepest vertical pocket was less than or equal to
tion performed at a later date was necessary in only 2 of 6 cm, or a total of 5 L was removed. Ultrasound
these cases. Johnson et al9 later reported a retrospective assessment of amniotic fluid was then repeated weekly.
series of 14 women treated with septostomy or amnior- A repeat amnioreduction was undertaken when the
eduction. Pregnancy was prolonged an average of 12 woman developed symptoms of excessive uterine activ-
weeks in the septostomy group compared to 6.5 weeks in ity or maternal respiratory compromise, or when the
the amnioreduction group (P ! .01). deepest vertical amniotic fluid pocket met the original
These data led us to consider a multicenter random- inclusion criteria.
ized trial to compare the standard therapeutic option of
amniodrainage with the newer technique of septostomy Septostomy procedure
for the treatment of TTTS.
The intertwin membrane was purposefully perforated
Material and methods under ultrasound guidance with a single puncture using
a 22-gauge needle. This was usually introduced though
The study was approved as a multicenter trial by the the donor twin’s gestational sac into the recipient twin’s
human investigational review board of the University of amniotic cavity, as preliminary experience before the
North Carolina School of Medicine, as well as each of trial indicated that visualization of the needle tip for
the local institutional review boards of the 11 partici- confirmation of perforation was easier with this ap-
pating centers. proach.
Moise et al 703
Figure 1 Distribution of women in the trial. Data in parentheses indicate numbers of patients. Information in brackets indicates
additional description of procedures for patients in that category. Amniored, amnioreduction.
All women undergoing septostomy underwent a re- sound was repeated 48 hours after successful septos-
peat ultrasound in 48 hours. If reaccumulation of tomy. If the oligohydramnios had not resolved in the
amniotic fluid in the donor twin’s amniotic cavity did donor twin’s sac, or if the deepest vertical pocket in the
not occur, a repeat septostomy was undertaken. Occa- recipient’s twin’s sac had increased by 30% over baseline
sionally, the treating clinician decided that a second value, crossover to the amnioreduction treatment arm
amnioreduction was also necessary at the time of the was allowed.
repeat septostomy. If a repeat septostomy was accom- If the amniotic fluid volume in the polyhydramnios
panied by an amnioreduction at the same setting, this sac was significant enough to be associated with
was considered a single additional procedure. Ultra- maternal symptoms, a salvage amnioreduction was
704 Moise et al
interim analysis, 36 women were enrolled in the amnior-

eduction arm and 37 women in the septostomy arm. A
decision was made to terminate the trial by the Data
Safety Monitoring Officer after the interim analysis
because of slower than projected enrollment that had
been anticipated, and almost identical perinatal survival
in the 2 treatment arms of the trial, rendering it unlikely
that the primary end point would be achieved.
The mean gestational age at enrollment in the
amnioreduction group was 20.9 G 2.6 weeks compared
to 20.8 G 2.7 weeks in the septostomy group (P = .85).
The trial was initiated before the description of the
Quintero staging system, and therefore cases were not
specifically staged at the time of enrollment in the trial.13
Figure 2 Survival for no infants, 1 infant, and both infants.
However, a review of the ultrasound parameters at entry
Solid bars represent women (pregnancies) in the aminoreduc-
tion group; plain bars represent women (pregnancies) in the
revealed that only 2 pregnancies (one in each arm of the
septostomy group. The vertical axis represents the percentage trial) were complicated by the finding of a hydropic fetus
of pregnancies in each treatment group. Actual number of at the time of entry (stage 4 disease). Therefore, 97% of
women (pregnancies) is denoted at the top of each vertical bar. the gestations in the trial represented stage 1 to 3
disease.
undertaken concurrently with the septostomy. In these Figure 1 depicts the total number of women in the
cases, the 22-gauge needle used to create the septostomy trial. Two women were lost to follow-up in the
was inserted through the 18-gauge needle used for the septostomy arm. One patient was randomized at 23
amnioreduction in an effort to create only 1 puncture of weeks’ gestation, was treated with 2 septostomies, and
the amniotic cavity. then moved to Korea at 24 2/7 weeks’ gestation. The
New therapeutic interventions such as laser ablation second patient was randomized to septostomy but all
of anastomotic vessels and umbilical cord occlusion data, including follow-up information, were lost. In the
were introduced into clinical care during the time course amnioreduction arm, 11 women required only 1 pro-
of this study.3,11 These modalities were employed in cedure. In the remaining 25 women, there were a total of
a minority of women who had progression of TTTS 80 procedures: 77 amnioreductions (range: 1-12), 1 laser
despite amnioreduction or septostomy. Outcomes of all therapy and 2 umbilical cord occlusions. In the septos-
women enrolled in the trial were analyzed on an intent- tomy arm of the trial, 18 (51%) of the women required
to-treat basis. only a single septostomy for treatment. One woman
randomized to the septostomy arm received a single
Statistical analysis
amnioreduction as her only therapeutic procedure. In 16
The primary outcome variable for the study was at least of the 35 women (46%), a salvage amnioreduction was
1 infant surviving until hospital discharge. Based on the undertaken in conjunction with the initial septostomy.
pilot trial of septostomy, we utilized a perinatal survival Sixteen women in this treatment group required a total
rate of 85% for the septostomy group and 65% for the of 54 additional procedures: 8 repeat septostomies (3 of
amnioreduction group.6,8 Using an alpha value of 0.05 these accompanied by an amnioreduction), 44 additional
and a 1-beta of 0.80, a sample size of 140 subjects (70 in amnioreductions (range: 1-8), 1 umbilical cord occlusion,
each arm) was required. An interim analysis was and 1 laser therapy.
planned at the midway point of the trial utilizing the Perinatal survival in the trial is shown in Figure 2.
O’Brien-Fleming stopping rule12 for discontinuing en- The survival of at least 1 infant occurred in 78% of the
rollment. pregnancies in the amnioreduction group vs 80% in the
The Shapiro-Wilk test was employed to test for septostomy group (RR = 0.94, 95%CI 0.55–1.61;
normality of distribution for continuous variables. Data P = .82). Survival of a single infant only occurred in
were analyzed using contingency tables, Student t test, 28% vs 20% of pregnancies (RR = 1.22, 95%CI 0.75–
Mann-Whitney test, and multiple logistic regression. 2.00; P = .44; amnioreduction vs septostomy), while
A P value of less than .05 was considered statistically survival of both infants was noted in 50% vs 60% of
significant. Relative risks and 95%CIs were calculated. pregnancies (RR = 0.82, 95%CI 0.52–1.30; P = .40;
amnioreduction vs septostomy). Fetal and neonatal
Results deaths occurred with equal frequency in the amnio-
reduction and septostomy groups (P = .40). There was
Between September 1997 and July 2002, a total of 73 no difference in the frequency of fetal or neonatal deaths
women were enrolled in the trial. At the time of the in either the donor or recipient twin (Table).
Moise et al 705
Table Outcome data

Amnioreduction Septostomy P value
Number 36 35
GA at delivery (wk) 29.5 (3.5) 30.7 (5.0) .24
Days gained 59.9 (26.4) 69.2 (37.5) .12
Birth weight of donor twin (g) 996 (408) 1291 (731) .12
Birth weight of recipient twin (g) 1075 (146–3189) 1455 (214–3335) .27
Fetal deaths
Donor 5 (14%) 4 (11%) .76
Recipient 4 (11%) 5 (14%) .69
Neonatal deaths
Donor 9 (25%) 6 (17%) .42
Recipient 8 (22%) 6 (17%) .59
No. of additional procedures 2 (1–12) 2 (1–9) .07
More than 1 procedure necessary 25 (69%) 16 (46%) .04
Data provided as mean (SD), median (range), or numbers (%).
Secondary outcome analysis revealed that women In the present investigation, the survival of at least 1
undergoing septostomy were less likely to require more infant did not differ between amnioreduction and
than the initial additional procedure for treatment septostomy (78% vs 80%, respectively). However, other
(P = .04). findings in the study would point to a slowing of the
Minimal complications were reported for both am- disease process of twin-to-twin transfusion with septos-
nioreduction and septostomy. There were no cases of tomy. More than half of the women in the septostomy
placental abruption or chorioamnionitis. Total disrup- arm required only the initial invasive intervention.
tion of the intervening membrane creating a monoamni- The mechanism by which septostomy potentially
otic twin gestation occurred in 1 case in each treatment retards the progression of TTTS is speculative. One
arm of the study. However, entanglement of the umbilical must first consider what causes the rapid reaccumulation
cords was not observed at delivery in either case. of amniotic fluid around the donor twin. The rapid time
In multivariate analysis, perinatal survival was not course of this finding points to movement of amniotic
related to treatment group after adjusting for gestational fluid from the polyhydramnios sac to the oligohydram-
age at randomization, gestational age at delivery, birth nios sac. The law of fluid dynamics of Pascal indicates
weight, and donor vs recipient twin status (adjusted that 2 compartments in the same vessel separated by an
RR = 1.1; 95%CI 0.60–1.94). elastic membrane should exhibit equal pressures. Thus,
fluid should not move between the amniotic sacs in
TTTS even after the membrane has been breached. In
Comment practical terms, however, once the intervening mem-
brane becomes transposed against the stuck twin, it
Severe polyhydramnios due to second-trimester TTTS is serves as an inelastic wall within the uterus, thereby
often associated with preterm labor and preterm pre- creating 2 separate compartments. A minor difference in
mature rupture of the membranes, leading to perinatal hydrostatic pressure is therefore the most likely expla-
loss. This has led many investigators to propose serial nation for the movement of fluid through the septos-
amnioreduction as the treatment of choice. Multiple tomy from the higher-pressure polyhydramnios sac to
cases have been reported in which a single amnioreduc- the lower-pressure oligohydramnios sac. This proposed
tion was the only therapeutic maneuver required for mechanism is consistent with a case reported by Bruner
TTTS. In one series, 17% of cases presenting in the and Crean.15 Indigo carmine injected at the time of
second trimester responded to a single amnioreduc- aminoreduction of the recipient’s sac was rapidly
tion.14 This was seen in the present study in almost detected in the amniotic fluid surrounding the donor
one third of the 38 cases enrolled in the amnioreduction twin. Examination of the intertwin membrane at de-
group. Saade et al8 proposed that the likely explanation livery revealed a 3-cm rent.
for this effect is that the dividing membrane, which is Equilibration of the amniotic fluid volumes may then
compressed against the uterine wall and therefore not have 1 of 2 subsequent effects. Polyhydramnios in TTTS
visualized, is unintentionally perforated during amnio- has been associated with alterations in the pulsatility
reduction. This led these authors to propose the use of index of the uterine artery due to increased amniotic
an intentional fenestration of the intervening membrane fluid pressure, reflecting reduced uterine perfusion, and
as a treatment option for severe TTTS. there is an association between raised amniotic fluid
706 Moise et al
pressure and abnormal fetal blood gases.16 Amniore- therefore, still remain viable treatment options for stage
duction or equilibration of amniotic volumes may I and II disease. Septostomy offers the advantage over
improve blood flow to the uterus. Alternatively, high amnioreduction of more frequently requiring only
amniotic fluid pressure may be transmitted to placental a single procedure with minimal complications. If pro-
plate vessels, causing increased uteroplacental resistance gression of disease is noted after a septostomy or
in the recipient twin. Doppler evidence of fetal hyper- amnioreduction, laser therapy would seem the treatment
tension in the recipient twin has been recently docu- of choice. However, we acknowledge that septostomy
mented.17 A reduction in amniotic fluid pressure due to may rarely impede subsequent laser therapy treatment if
resolution of the polyhydramnios could alleviate this the resultant floppy membrane hinders visualization of
increased afterload. the chorionic plate. We suggest that a randomized trial
Septostomy was associated with minimal complica- of combination amnioreduction/septostomy vs laser is
tions in this trial. Critics of septostomy have cautioned indicated in early stage TTTS.
that purposeful fenestration of the intertwin membrane
could lead to creation of a monoamniotic gestation with
a significant risk for entangled umbilical cords, and References
subsequent perinatal loss. Sporadic case reports of
disruption of the dividing membrane have been reported 1. Jain V, Fisk NM. The twin-twin transfusion syndrome. Clin
after genetic amniocentesis using a single puncture Obstet Gynecol 2004;47:181-202.
2. Denbow ML, Cox P, Taylor M, Hammal DM, Fisk NM. Placental
technique.18 However, large series of amniocentesis in angioarchitecture in monochorionic twin pregnancies: relationship
twins have failed to identify this as a complication.19 In to fetal growth, fetofetal transfusion syndrome, and pregnancy
the present trial, only 2 cases of disruption of the outcome. Am J Obstet Gynecol 2000;182:417-26.
intervening membrane occurred, one in each treatment 3. Hecher K, Plath H, Bregenzer T, Hansmann M, Hackeloer BJ.
arm. In both cases, entanglement of the umbilical cords Endoscopic laser surgery versus serial amniocenteses in the
treatment of severe twin-twin transfusion syndrome. Am J Obstet
was not seen at birth. Nevertheless, both amnioreduction Gynecol 1999;180:717-24.
and septostomy present the potential for intertwining of 4. Gul A, Aslan H, Polat I, et al. Natural history of 11 cases of twin-
the umbilical cords, which could lead to perinatal loss. twin transfusion syndrome without intervention. Twin Res 2003;6:
One of the limitations of the current trial was that 263-6.
long-term neurologic outcome was not assessed. Since it 5. Berghella V, Kaufmann M. Natural history of twin-twin trans-
fusion syndrome. J Reprod Med 2001;46:480-4.
was first introduced in 1990 by De Lia,20 proponents of 6. Mari G, Roberts A, Detti L, Kovanci E, Stetos T, Bahado-
laser ablation of the communicating vessels for the Singh RO, et al. Perinatal morbidity and mortality rates in
treatment of TTTS have suggested that both amnio- severe twin-twin transfusion syndrome: results of the Interna-
reduction and septostomy do not treat the underlying tional Amnioreduction Registry. Am J Obstet Gynecol 2001;185:
condition. Although perinatal outcomes with amnio- 708-15.
7. Bower SJ, Flack NJ, Sepulveda W, Talbert DG, Fisk NM. Uterine
reduction are reasonable in early stage (I/II) disease, the artery blood flow response to correction of amniotic fluid volume.
outcome is poorer with advanced and progressive Am J Obstet Gynecol 1995;173:502-7.
disease, such that other treatment modalities are now 8. Saade GR, Belfort MA, Berry DL, Bui TH, Montgomery LD,
warranted. In a recent randomized trial, laser coagula- Johnson A, et al. Amniotic septostomy for the treatment of twin
tion of communicating vessels in TTTS was reported to oligohydramnios-polyhydramnios sequence. Fetal Diagn Ther
1998;13:86-93.
improve both perinatal survival, as well as short-term 9. Johnson JR, Rossi KQ, O’Shaughnessy RW. Amnioreduction
neurologic outcome.21 This is supported by a large versus septostomy in twin-twin transfusion syndrome. Am J Obstet
comparative series showing that laser was associated Gynecol 2001;185:1044-7.
with better survival in stage III and IV disease than 10. Dorman K, Saade GR, Smith H, Moise KJ Jr. Use of the world
amnioreduction, albeit with worse survival in early stage wide web in research: randomization in a multicenter clinical trial
of treatment for twin-twin transfusion syndrome. Obstet Gynecol
disease.22 2000;96:636-9.
The best therapeutic option for all women with severe 11. Taylor MJ, Shalev E, Tanawattanacharoen S, Jolly M, Kumar S,
second-trimester TTTS has yet to be determined. In Weiner E, et al. Ultrasound-guided umbilical cord occlusion using
a recent editorial that accompanied the report of the bipolar diathermy for stage III/IV twin-twin transfusion syn-
randomized trial of laser therapy, Fisk and Galea23 drome. Prenat Diagn 2002;22:70-6.
12. O’Brien PC. Procedures for comparing samples with multiple
proposed that although laser treatment of TTTS endpoints. Biometrics 1984;40:1079-87.
appeared promising, further investigation particularly 13. Quintero RA, Morales WJ, Allen MH, Bornick PW, Johnson PK,
in early-stage disease is warranted. In addition, one Kruger M. Staging of twin-twin transfusion syndrome. J Perinatol
study of the progression of early stage TTTS indicated 1999;19:550-5.
that only 31% of stage I and 20% of stage II cases 14. Elliott JP, Urig MA, Clewell WH. Aggressive therapeutic amnio-
centesis for treatment of twin-twin transfusion syndrome. Obstet
advanced to a higher stage disease, and there is evidence Gynecol 1991;77:537-40.
that adverse neurologic outcome is associated with 15. Bruner JP, Crean DM. Equalization of amniotic fluid volumes
advanced disease.6,24 Amnioreduction and septostomy, after decompression amniocentesis for treatment of the twin
Moise et al 707
oligohydramnios-polyhydramnios sequence. Fetal Diagn Ther 20. De Lia JE, Cruikshank DP, Keye WR Jr. Fetoscopic neodym-
1999;14:80-5. ium:YAG laser occlusion of placental vessels in severe twin-
16. Fisk NM, Vaughan J, Talbert D. Impaired fetal blood gas status in twin transfusion syndrome. Obstet Gynecol 1990;75:1046-53.
polyhydramnios and its relation to raised amniotic pressure. Fetal 21. Senat MV, Deprest J, Boulvain M, Paupe A, Winer N, Ville Y.
Diagn Ther 1994;9:7-13. Endoscopic laser surgery versus serial amnioreduction for severe
17. Mahieu-Caputo D, Salomon LJ, Le Bidois J, Fermont L, Brunhes twin-to-twin transfusion syndrome. N Engl J Med 2004;351:
A, Jouvet P, et al. Fetal hypertension: an insight into the 136-44.
pathogenesis of the twin-twin transfusion syndrome. Prenat Diagn 22. Quintero RA, Dickinson JE, Morales WJ, Bornick PW, Bermudez
2003;23:640-5. C, Cincotta R, et al. Stage-based treatment of twin-twin trans-
18. Feldman DM, Odibo A, Campbell WA, Rodis JF. Iatrogenic fusion syndrome. Am J Obstet Gynecol 2003;188:1333-40.
monoamniotic twins as a complication of therapeutic amniocen- 23. Fisk NM, Galea P. Twin-twin transfusiondas good as it gets?
tesis. Obstet Gynecol 1998;91:815-6. N Engl J Med 2004;351:182-4.
19. Sebire NJ, Noble PL, Odibo A, Malligiannis P, Nicolaides KH. 24. Taylor MJ, Govender L, Jolly M, Wee L, Fisk NM. Validation of
Single uterine entry for genetic amniocentesis in twin pregnancies. the Quintero staging system for twin-twin transfusion syndrome.
Ultrasound Obstet Gynecol 1996;7:26-31. Obstet Gynecol 2002;100:1257-65.
www.ajog.org
Chorioamnionitis with a fetal inflammatory response is

associated with higher neonatal mortality, morbidity,
and resource use than chorioamnionitis displaying
a maternal inflammatory response only
Jacqueline Lau, BSc,a,b Fergall Magee, MD, FCRCP(C),a,c Zhenguo Qiu, PhD,a
Jill Houbé, MD, FRCPC, MPhil,a,d Peter Von Dadelszen, MBChB, MRCOG, FRCSC,
DPhil,a,e Shoo K. Lee, MBBS, FRCPC, PhDa,d,*
Centre for Healthcare Innovation and Improvement,a School of Medicine, Queen’s Universityb; Departments of
Pathology,c Paediatrics,d and Obstetrics and Gynaecology,e University of British Columbia, Vancouver, British
Columbia, Canada
Received for publication July 27, 2004; revised December 16, 2004; accepted January 11, 2005
KEY WORDS Objective: This study was undertaken to evaluate whether the proximity of infection of the
Chorioamnionitis chorion/amnion and fetal vessels affects neonatal outcomes.
Neonatology Study design: We examined all (n = 2012) infants admitted to the British Columbia’s Children’s
Outcome Hospital Neonatal Intensive Care Unit, from January 1996 to October 1997. We included infants
Morbidity with a placental examination (n = 1296), and stratified those with histologic chorioamnionitis
Resources into cases displaying a maternal inflammatory response only and cases also displaying a fetal
inflammatory response (funisitis and/or fetal surface vessel angiitis).
Results: Histologic evidence of chorioamnionitis was present in 31% of placentas. Of those, 38%
exhibited maternal inflammation only, whereas 62% also exhibited fetal inflammation. Neonatal
mortality (9.2% vs 7.2%), morbidity, and resource use were significantly (P ! .05) higher when
fetal inflammation was present compared with when only maternal inflammation was present.
Conclusion: Chorioamnionitis with a fetal inflammatory response is associated with higher
neonatal mortality, morbidity, and resource use than when only a maternal inflammatory
response is present.
Supported by the Medical Research Council of Canada and the BC Research Institute for Children’s and Women’s Health.
* Reprint requests: Shoo K. Lee, MBBS, FRCPC, PhD, Centre for Healthcare Innovation and Improvement, 4480 Oak St, Room E414,
Vancouver, BC, Canada V6H 3V4.
E-mail: shool@interchange.ubc.ca
doi:10.1016/j.ajog.2005.01.017
Lau et al 709
The term chorioamnionitis refers to acute inflamma- Information abstracted

tion of the placental membranes (amnion and chorion)
and is a leading cause of maternal and fetal complica- Abstracted information included specific placental and
tions, including preterm birth and neonatal infection.1-3 birth characteristics as well as neonatal outcomes.
The inflammatory process is generally regarded as Maternal, pregnancy, placental, and neonatal character-
a continuum. During the early stage, the neutrophils istics were abstracted and recorded in tabular format.
involved in this inflammatory response are usually
maternal (migrating from intervillous sac and/or Definition of terminology
decidual vessels) in origin. During later stages, fetal
neutrophils (migrating from fetal surface vessels of the Chorioamnionitis was defined histologically as an in-
chorionic plate or umbilical cord) are involved.4,5 The filtration of the chorion and/or amnion (membranes) by
incidence of histologically identified chorioamnionitis polymorphonuclear leucocytes. These polymorphs are
among infants born preterm has been reported to be maternal and may migrate from the decidual vessels and/
about 33%.1 Although chorioamnionitis among infants or intervillous space. Diagnosis was based on histologic
admitted to the neonatal intensive care unit (NICU) has features (not gross pathology) and required neutrophil
been previously reported, few studies have examined the infiltration involving at least 50% of membrane section
correlation between the proximity of the inflammatory submitted. The cases of choriamnionitis were stratified
response (fetal vs maternal inflammatory response) in into those that showed the inflammatory response
chorioamnionitis and neonatal outcomes.6-8 The objec- confined to membranes only (maternal response) and
tive of this study was to compare outcomes of infants those that showed both inflammation of the membranes
whose placenta exhibited histologic chorioamnionitis with a fetal surface vessel angiitis and/or umbilical vessel
with fetal response with those exhibiting maternal inflammation (maternal and fetal response).4,10
inflammatory response only. Indices of neonatal outcome were defined according
to the Canadian Neonatal Network Data Abstractor
Manual.9 Gestational age was defined as the best
obstetric estimate that was based on early prenatal
ultrasound, obstetric examination, and obstetric history,
Study population unless the postnatal pediatric estimate of gestation
differed from the obstetric estimate by more than 2
The study cohort included all infants (n = 2012) weeks. In that case, the pediatric estimate was used
admitted to the British Columbia’s Children’s Hospital instead. An infant was defined as small-for-gestational
NICU from January 1996 to October 1997. Infants age (SGA) if the birth weight was less than the third
were identified from standardized data collected by the percentile for gestational age according to the British
Canadian Neonatal Network as part of a larger study Columbia provincial growth charts established by Whit-
of 17 tertiary NICUs across Canada, and which has field et al in 1992.11 SNAP-II is a neonatal illness
been previously described.9 Trained research assistants severity score calculated from 6 empirically weighted
abstracted patient information from the mothers’ and physiologic measurements made during the first 12
infants’ charts at each participating hospital on a daily hours of admission to the NICU.12 NTISS13 is a score
basis. Data were entered directly into laptop computers of neonatal therapeutic intensity calculated from
at the bedside by using a customized data entry a checklist of 63 NICU therapies used in a 24-hour
program with built-in error checking and a standard period, weighted according to invasiveness and cost.
manual of operations and definitions. Identified infants Intraventricular hemorrhage was defined according to
were matched with placental histopathology reports the criteria of Papile et al14 from head ultrasound
from the Department of Pathology database. The performed before 14 days of life. Necrotizing enteroco-
pathology slides were read by 4 surgical pathologists litis was defined according to the criteria of Bell et al15
in the hospital Department of Pathology and reviewed (stage II or higher) and was classified as medical (clinical
by a single perinatal surgical pathologist. Acute in- symptoms and signs plus evidence of pneumatosis on
flammation of the membranes was reported as present abdominal radiographs) or surgical (histologic evidence
or absent. A further comment was then made as to on surgical specimen of intestine). Retinopathy of pre-
whether there was evidence of a fetal inflammatory maturity was defined according to the International
response (fetal surface vessel angiitis, funisitis). Pathol- Classification for Retinopathy of Prematurity16 and the
ogy reports were reviewed by using a standardized data Reese Classification of cicatrical disease.17 Nosocomial
form and data definitions by a single reviewer who infection was defined by using blood and cerebrospinal
was blinded to the maternal and neonatal outcomes. fluid culture results according to criteria by Freeman
Institutional ethics approval for the study was et al.18 Primary infection was defined as positive single
obtained. organism cultures from blood or cerebrospinal fluid
710 Lau et al
Table I Characteristics and outcomes of infants with no chorioamnionitis, maternal inflammation and fetal inflammation (n = 1296)
Percent (unless otherwise specified) No chorioamnionitis Maternal inflammation Fetal inflammation P value*
Number of infants, n (%) 893 (68.9) 153 (11.8) 250 (19.3)
Mean birth weight, g (SE) 2124 G 28 2028 G 74 1894 G 65 ! .01
Mean gestational age at birth, wk (SE) 33.7 G 0.1 33.0 G 0.4 31.7 G 0.3 ! .01
Male sex (%) 55.2 59.2 53.2 .50
Prenatal care 99.1 97.8 98.7 .46
Cesarean section (%) 50.7 40.5 33.2 ! .01
Outborn (%) 13.3 16.3 15.6 .47
Multiple births (%) 30.3 14.4 10.0 ! .01
Antenatal steroids (%) 42.1 47.8 63.5 ! .01
Maternal hypertension (%) 17.6 13.7 2.4 ! .01
SGA (3rd percentile) 7.0 4.6 0.8 ! .01
Apgar !7 at 5 min 11.3 21.5 21.3 ! .01
SNAP-II score (mean) 8.9 12.0 14.1 ! .01
NTISS score (mean) 12.1 12.5 14.9 ! .01
RDS 28.7 29.8 41.8 ! .01
Chronic lung disease 9.3 14.6 17.1 ! .02
Primary infection 0.9 1.3 4.8 ! .01
Nosocomial infection 7.2 12.4 14.0 ! .01
Necrotizing enterocolitis 1.1 3.4 4.1 ! .01
Severe IVH 3.3 8.0 7.8 ! .05
Seizures 4.5 5.9 4.8 .77
Retinopathy of prematurity (Ograde 3) 1.9 3.7 5.6 .40
Patent ductus arteriosus 8.9 15.1 20.2 ! .01
Death 3.6 7.2 9.2 ! .01
Survival without major morbidity 81.9 75.2 62.0 ! .01
Mean duration of NICU stay (days G SE) 22.6 G 1.1 33.5 G 4.6 37.8 G 3.2 ! .01
Need for assisted ventilation 36.2 43.1 54.8 ! .01
Need for oxygen treatment 49.8 49.7 66.8 ! .01
Antibiotic use 67.4 75.2 92.8 ! .001
Villous edema 9.5 9.2 20.0 ! .001
* ANOVA was used to examine duration of NICU stay, SNAP, and NTISS. A likelihood-ratio test was used for other analyses.
obtained from an infant with signs or risk factors for These groups were also compared with those who had
sepsis during the first 48 hours of life. Patent ductus a placental examination but did not have chorioamnio-
arteriosus was defined as clinical diagnosis plus treat- nitis. Descriptive statistics were used to examine the
ment with indomethacin or surgical ligation or both. data. Analysis of variance and likelihood ratio tests were
Respiratory distress syndrome was defined as the used to compare differences in the characteristics and
presence of respiratory symptoms, such as grunting outcomes of infants with maternal and fetal inflamma-
and chest retraction, typical chest x-ray findings, or tion, maternal inflammation, and those cases with no
treatment with surfactant and the need for mechanical histologic evidence of chorioamnionitis. Multivariable
ventilation for greater than 24 hours. Chronic lung logistic regression analysis was used to examine the
disease was defined as oxygen dependency at 36 weeks’ relationship between the extent of inflammation and
corrected gestational age for an infant who was born at neonatal outcomes after adjusting for confounding
32 weeks’ or less gestation.19 Seizures were defined as factors, and multivariable linear regression modeling
clinically significant episodes witnessed by a nurse or was used to analyze resource consumption and length of
physician and for which anticonvulsant treatment was NICU stay.
given. Outborn infants were those born at a hospital
different from the hospital in which the NICU was
located. Results
Data analysis Of the 2012 cases, 1296 had a record of placental

pathology, representing 64% of all NICU admissions.
Infants with histologic evidence of chorioamnionitis Histologic evidence of a maternal inflammatory
were stratified into those with maternal inflammation response was identified in 403 cases, or 31% of those
only and those with maternal and fetal inflammation. with a pathology report. Of those 403 cases, 153 (38%)
Lau et al 711
Table II Risk factors for outcomes among NICU infants, using multivariable logistic regression analysis (n = 1296) (ORs and 95% CIs
shown)
Survival without Primary Nosocomial Need for assisted Need for
Variables Death in NICU major morbidity infection infection ventilation oxygen
Fetal inflammation 2.9 (1.6-5.1) 0.4 (0.3-0.6) 5.5 (2.2-13.7) 2.3 (1.5-3.6) 1.4 (1.1-2.0) 1.4
(1.1-2.0)
Maternal 2.1 (1.1-4.2) NS NS 1.9 (1.1-3.3) NS NS
inflammation
Antenatal steroids NS 0.3 (0.2-0.4) NS NS 5.2 (3.9-6.9) 2.3
(1.8-3.0)
Apgar !7 at 5 min NS 0.2 (0.2-0.3) NS NS 5.6 (3.7-8.3) 1.9
(1.4-2.7)
Outborn status 2.6 (1.5-4.6) 0.5 (0.4-0.8) NS 1.7 (1.1-2.7) 2.5 (1.7-3.7) NS
SGA (3rd 2.5 (1.1-6.2) 0.4 (0.2-0.7) NS 3.0 (1.6-5.8) NS NS
percentile)
Multiple birth NS NS NS NS NS NS
Cesarean section NS NS NS NS 1.4 (1.1-1.8) NS
Patent ductus Necrotizing Chronic lung Severe IVH RDS Villous
Variables arteriosus enterocolitis disease edema
Fetal inflammation 2.6 (1.7-3.8) 5.6 (2.1-14.9) 2.3 (1.4-3.9) 2.4 (1.1 - 5.6) 1.9 (1.3-2.5) 2.1
(1.4-3.1)
Maternal 1.8 (1.1-3.0) 4.0 (1.3-12.4) NS NS NS NS
inflammation
Apgar !7 at 5 min NS NS NS 2.2 (1.1-4.8) 1.9 (1.4-2.7) NS
Outborn status 1.7 (1.1-2.6) NS NS NS 1.6 (1.2-2.3) 1.6
(1.1-2.5)
SGA (3rd NS 3.9 (1.1-14.1) 3.9 (1.8-8.2) NS NS NS
percentile)
Multiple birth NS 2.7 (1.1- 6.6) NS NS 1.5 (1.1-2.0) NS
Cesarean section NS NS NS NS 1.4 (1.1-1.8) 0.6
(0.3-0.9)
Maternal NS NS NS NS 1.8 (1.3-2.6) NS
hypertension
Male sex NS NS 0.6 (0.4-0.9) NS NS NS
had maternal inflammation only, whereas 250 (62%) (31.7 G 0.3 weeks, 33.0 G 0.4 weeks, 33.7 G 0.1 weeks,
also displayed evidence of fetal inflammation. Infants respectively) and birth weight (1894 G 65 g, 2028 G 74 g,
who had histologic examination of their placentas were 2124 G 28 g, respectively) than those with only
significantly (P ! .05) more likely than infants who did a maternal inflammatory response or no chorioamnio-
not have a placental examination to be inborn (86% vs nitis. They also had significantly (P ! .01) lower
76%), male (44.3% vs 38.5%), of lower mean gesta- incidence of maternal hypertension (2.4%, 13.7%, and
tional age (33.2 G 4.3 weeks vs 37.3 G 3.1 weeks) and 17.6%, respectively), cesarean section (33.2%, 40.5%,
lower mean birth weight (2.1 G 0.9 kg vs 3.0 G 0.8 kg), and 50.7%, respectively), and multiple births (10%,
receive antenatal corticosteroid treatment (46.8% vs 14.4%, and 30.3%, respectively) and were more likely to
6.7%), born of a mother with maternal hypertension have received antenatal corticosteroid treatment
(15.5% vs 8.6%), and born by cesarean section (46.2% (63.5%, 47.8%, and 42.1%, respectively) than infants
vs 30.4%). However, the incidence of multiple births with a maternal inflammatory response and no cho-
was similar between the groups (12.5% vs 11.7%, rioamnionitis.
P = .61). Cases with histologic evidence of fetal in- Table I compares outcomes and resource use of
flammatory response were associated with a higher infants without chorioamnionitis, with chorioamnionitis
incidence of villous edema as identified by histology involving a maternal response only, and with chorioam-
(20.0% vs 9.2%, odds ratio [OR] = 2.48; 95% CI 1.32- nionitis also displaying a fetal response. Mortality,
4.67), when compared with those displaying maternal morbidity, and resource use were highest among infants
inflammation only. with evidence of a fetal inflammatory response and
Infants with a fetal inflammatory response had lowest among those with no evidence of chorioamnio-
significantly (P ! .01) lower mean gestational age nitis.
712 Lau et al
Table II shows independent predictors of neonatal Jeffrey.26 Our finding that increasing stages of cho-
outcomes, after adjustment for baseline population rioamnionitis are correlated with chronic lung disease is
characteristics by using multivariable logistic regression also consistent with previous reports by Watterberg
analysis. Maternal only inflammatory response was et al27 and others, and suggests that the severity of
independently predictive of mortality, nosocomial in- pulmonary injury may be associated with severity of
fection, necrotizing enterocolitis, and patent ductus chorioamnionitis.
arteriosus. However, fetal inflammatory response was Although chorioamnionitis with fetal response has
even more highly predictive than maternal inflammation usually been regarded as a later stage of infection than
only for these same outcomes, as well as for primary chorioamnionitis with maternal response only,26 it is
infection, respiratory distress syndrome, need for assis- unclear whether this is really the case, or whether they
ted ventilation, need for oxygen treatment, chronic lung might be manifestations of different infections (eg, type
disease, severe intraventricular hemorrhage, and de- of organism, route of infection), modification by treat-
creased survival without major morbidity. A propor- ment (eg, antenatal corticosteroids), and host defenses of
tional odds model showed that fetal (but not maternal both the mother and fetus. It is also unclear whether
only) inflammation was also significantly associated infants whose placentas show fetal inflammation should
with higher NTISS scores, after adjustment for baseline be treated differently than those with maternal inflam-
population risk factors. A multivariate linear regression mation only because of the higher associated mortality,
model demonstrated that fetal inflammation (but not morbidity, and resource use. It is possible that more
maternal only) was associated with increased risk aggressive treatment of chorioamnionitis or earlier de-
adjusted duration of NICU stay. livery of an infant before the development of a fetal
inflammatory response may result in improved out-
Comment comes. Clarification of these questions through pro-
spective clinical trials is important for determining
The importance of placental pathology for the diagnosis optimum management of chorioamnionitis and reducing
and management of neonatal conditions has long been associated mortality and morbidity. Placental histologic
recognized. As early as 1892, a Scottish obstetrician, findings may have a role to play in risk assessment and
Ballantyne stated that ‘‘During the intrauterine life, the prognostication. Addition of placental findings to known
fetus, the membranes, the cord, and the placenta form demographic risk factors and clinical presentation may
an organic whole, and disease of any part must react improve our ability to more accurately predict neonatal
upon and affect the others.’’20 Our findings that 31% of outcomes and counsel parents. Further research is also
placentas examined in our cohort showed histologic needed to determine whether clinical correlates of
evidence of chorioamnionitis, and that chorioamnionitis maternal and fetal inflammation among infants with
is associated with higher incidence of adverse neonatal chorioamnionitis can provide further insights into iden-
outcomes are consistent with reports by previous tification of antenatal risk factors that might guide
authors.5,21 interventions aimed at reducing neonatal mortality and
Our finding of increased mortality and morbidity morbidity.
among infants with increasing stages of chorioamnioni- This study was a single institution report and
tis is consistent with previous reports by Van Hoevan therefore the results reflect the unique diagnostic pat-
et al22 and others.23-25 This is in contrast to a recent report terns and pathology referral practices of one hospital.
by Lahra and Jeffrey,26 who found that chorioamnioni- Although neonatal data were abstracted by a single
tis with a histologic fetal response was more prevalent individual using standardized definitions and protocols,
among preterm survivors of 25 to 34 weeks’ gestation placental pathology was reported by 4 pathologists.
(but not 20-24 weeks’ gestation) compared with cases of However, interobserver variability was minimized be-
perinatal death. Lahra and Jeffrey26 attributed their cause half the placental specimens were read by a single
finding to increased secretion of cortisol production perinatal surgical pathologist, who also reviewed the
secondary to intrauterine exposure to infection, which specimens read by the other pathologists. Because only
facilitates fetal lung maturation in preterm infants and 64% of the cohort had histologic examination of the
reduces respiratory distress syndrome during the neo- placenta, there might be bias in the results. However, the
natal period. However, because their cohort included incidence of chorioamnionitis in our study was consis-
infants born between 1984 and 1999, the majority of tent with other studies.
infants studied were from the presurfactant era. It is also
unclear to what extent antenatal corticosteroid treat- Acknowledgments
ment was used in the cohort. Finally, their results were
not risk-adjusted for either population risks or use of We acknowledge the assistance of Ruth Little in pre-
antenatal corticosteroids. These factors could account paring this article, and the Canadian Neonatal Network
for the difference in findings reported by Lahra and for use of their data.
Lau et al 713
References 14. Papile LA, Munsick-Bruno G, Schaefer A. Relationship of

cerebral intraventricular hemorrhage and early childhood handi-
1. Ogunyemi D, Murillo M, Jackson U, Hunter N, Alperson B. The caps. J Pediatr 1983;103:272-7.
relationship between placental histopathology findings and peri- 15. Bell MJ, Ternberg JL, Feigin RD, Keating JP, Marshall R, Barton
natal outcome in preterm infants. J Matern Fetal Neonatal Med L, et al. Neonatal necrotizing enterocolitis: therapeutic decisions
2003;13:102-9. based on clinical staging. Ann Surg 1987;187:1-7.
2. Romero R, Espinoza J, Chaiworapongsa T, Kalache K. Infection 16. American Academy of Pediatrics. Committee members: an In-
and prematurity and the role of preventative strategies. Semin ternational Committee for the Classification of the Late Stages of
Neonatol 2002;7:259-74. Retinopathy of Prematurity. An international classification of
3. Morales WJ, Washington SR 3rd, Lazar AJ. The effect of retinopathy of prematurity. Pediatrics 1984;74:127-33.
chorioamnionitis on perinatal outcome in preterm gestation. 17. Reese AB, Ling MJ, Owens WC. A classification of retrolental
J Perinatol 1987;7:105-10. fibroplasias. Am J Opthalmol 1953;36:1333.
4. Naeye RL. Functionally important disorders of the placenta, 18. Freeman J, Epstein MF, Smith NE, Platt R, Sidebottom DG,
umbilical cord, and fetal membranes. Hum Pathol 1987;18:680-91. Goldmann DA. Extra hospital stay and antibiotic usage with
5. Lee SD, Kim MR, Hwang PG, Shimm SS, Yoon BH, Kim CJ. nosocomial coagulase-negative staphylococcal bacteremia in two
Chorionic plate vessels as an origin of amniotic fluid neutrophils. neonatal intensive care unit populations. Am J Dis Child
Pathol Int 2004;54:516-22. 1990;144:324-9.
6. Rhone SA, Magee F, Remple P, Money D. The association of 19. Shennan AT, Dunn MS, Ohlsson A, Lennox K, Hoskins EM.
placental abnormalities with maternal and neonatal clinical findings: Abnormal pulmonary outcomes in preterm infants: prediction
retrospective cohort study. J Obstet Gynaecol Can 2003;25:123-8. from oxygen requirement in the neonatal period. Pediatrics
7. Blackwell S, Romero R, Chaiworapongsa T, Kim YM, Bujold E, 1988;82:527-32.
Espinoza J, et al. Maternal and fetal inflammatory responses in 20. Ballantyne JW. The diseases and deformities of the foetus, vol. 1.
unexplained fetal death. J Matern Fetal Neonatal Med Edinburgh: Oliver and Boyd; 1892.
2003;14:151-7. 21. Hardt NS, Kostenbauder M, Ogburn M, Behnke M, Resnick M,
8. Yoon BH, Romero R, Park JS, Kim M, Oh SY, Kim CJ, Jun JK. Cruz A. Influence of chorioamnionitis on long term prognosis in
The relationship among inflammatory lesions of the umbilical cord low birth weight infants. Obstet Gynecol 1985;65:5-10.
(funisitis), umbilical cord plasma interleukin 6 concentration, 22. Van Hoevan KH, Anyargbunam A, Hochster H, Whitty JE,
amniotic fliud infection, and neonatal sepsis. Am J Obstet Gynecol Distant J, Crawford C, et al. Clinical significance of increasing
2000;183:1124-9. histologic severity of acute inflammation in the fetal membranes
9. Lee SK, McMillan DD, Ohlsson A, Pendray M, Synnes A, Whyte and umbilical cord. Pediatr Pathol Lab Med 1996;16:731-44.
R, et al and The Canadian NICU Network. Variations in practice 23. Meuller-Heubach E, Rubinstein DN, Schwarz SS. Histologic
and outcomes in the Canadian NICU Network: 1996-1997. chorioamnionitis and preterm delivery in different patient popula-
Pediatrics 2000;106:1070-9. tions. Obstet Gynecol 1990;75:622-6.
10. Benirschke K, Kaufmann P. Pathology of the human placenta. 2nd 24. Zhang J, Kraus FT, Aquino TI. Chorioamnionitis: a comparative
ed. New York: Spinger-Verlag; 1990. Chap. 24. histologic, bacteriologic and clinical study. Int J Gynecol Pathol
11. Whitfield M. British Columbia provincial growth chart. Vancou- 1985;4:1-10.
ver: British Columbia Children’s Hospital; 1992. 25. Romero R, Salve CM, Athanassiadis AP, Hanaoka S, Mazor M,
12. Richardson DK, Corcoran JD, Escobar GL, Lee SK, for the Sepulveda W, et al. The relationship between acute inflammation
Canadian NICU Network, the Kaiser Permanente Neonatal Min- of the preterm placenta and amnionitic fluid microbiology. Am J
imum Data Set Wider Area Network, and the SNAP-II Study Obstet Gynecol 1992;166:1382-8.
Group. SNAP-II and SNAPPE-II: simplified newborn illness 26. Lahra MM, Jeffrey HE. A fetal response to chorioamnionitis is
severity and mortality risk scores. J Pediatr 2001;138:92-100. associated with early survival after preterm birth. Am J Obstet
13. Gray JE, Richardson DK, McCormick MC, Workman-Daniels K, Gynecol 2004;190:147-51.
Goldmann DA. Neonatal therapeutic intervention scoring system 27. Watterberg KL, Scott SM, Naeye RL. Chorioamnionitis,
(NTISS): a therapy-based severity of illness assessment tool. cortisol and acute lung disease in very low birth weight infants.
Pediatrics 1992;90:561-7. Pediatrics 1997;99:E6.
www.ajog.org
Early discharge from obstetrics-pediatrics at the

Hospital de Valme, with domiciliary follow-up
José Antonio Sainz Bueno,a,* Marı́a Ruiz Romano,b Rogelio Garrido Teruel,a
Antonio Gutiérrez Benjumea,b Ana Fernández Palacı́n,c Carmen Almeida
González,c Manuel Caballero Manzanoa
Obstetric and Gynecology Department,a and Pediatric Department,b Hospital de Valme, Sevilla, and
Social Health Science Service, Biostatistics Department, University of Sevilla, Spainc
Received for publication June 17, 2004; revised January 6, 2005
KEY WORDS Objectives: This study was undertaken to evaluate the advantages and disadvantages of
Early discharge a program of early obstetric-pediatric discharge (24 hours postpartum) with domiciliary follow-
Postpartum up, compared with the traditional postpartum hospital stay (more than 48 hours), according to
Newborn infants the criteria described by reviewers of the subject.
Mothers Study design: A randomized controlled trial of early obstetric discharge for healthy mothers and
term infants, with postpartum randomization, with no prenatal preparation and with
observational and clinical follow-up was performed. The participants were mothers with healthy,
term neonates (37-42 weeks) weighing more than 2500 g and produced via vaginal delivery and
with a verified normal evolution before discharge. The sample consisted of 430 cases (213 cases
with early discharge, and 217 control cases) in which the following variables were evaluated:
existence of complications in the mother and/or child that required rehospitalization or a medical
consultation, existence of maternal problems of fatigue or anxiety/depression after the birth,
continuity of lactation and its problems, satisfaction of the mother and family, and relative costs.
Conclusion: After demonstrating the homogeneity of the groups, no significant differences were
found in the rates of maternal rehospitalization (1.9% in the early discharge group vs 2.3% in the
control group, relative risk 0.81, 95% CI 0.21-3.03) or in the rates of rehospitalization of the
neonates (1.4% in the early discharge group vs 2.3% in the control group, relative risk 0.16, 95%
CI 0.15-2.56). No increases were observed in maternal or neonatal disease, puerperal fatigue, or
maternal anxiety/depression. A prolongation of maternal lactation to 3 months was observed in
the early discharge group (P = .016 !.05 Fisher exact test). When the cost of early discharge is
compared with that of traditional discharge with a minimum of 48 hours hospital stay, we find
a saving of 18% to 20%. The level of maternal satisfaction with early discharge is better than
90%.
* Reprint requests: José Antonio Sainz Bueno, Avd/Virgen de la Esperanza, No 29, Bloque 7-Piso 3C, 41012 Sevilla, Spain.
E-mail: ginjsb2@wanadoo.es.
doi:10.1016/j.ajog.2005.01.015
Sainz et al 715
The 2 reasons for postpartum hospitalization are to neonatal care and maternal lactation), at 7 to 10 days in
identify maternal or neonatal complications after the the practice and at 1, 3, and 6 months with a follow-up
birth and to offer professional assistance to mother and telephone consultation to the home.
child ensuring that the mother has recovered sufficiently
Nonintervention group
from the birth to be able to care for both herself and her
Puerperae discharged after the usual minimum of 48
child when they return home. For these purposes, in
hours postpartum, monitored at 7 to 10 days in the
cases of uncomplicated vaginal birth, the postpartum
practice and at 1, 3, and 6 months with a follow-up
hospital stay recommended by the American College of
telephone consultation to the home.
Obstetricians and Gynecologists (ACOG) and the
Variables measured are as follows:
American Academy of Pediatrics (AAP) is a minimum
of 48 hours.1 Primary results of mother (in the first 6 weeks):
Early discharge, in cases of vaginal delivery, may Proportion of readmissions, proportion of cases with
therefore be defined as the release of the mother and/or high scores for depression (Appendix A.)
the neonate in less than 48 hours postpartum.2 Primary results of newborn infant (in the first 28
Since the 1970s, there have been many programs of days): Proportion of neonates readmitted, propor-
early obstetric-pediatric discharge, and various reviews tion of cases with breastfeeding, exclusive or mixed
of the medical literature on this topic exist with 1 conclu- (evaluation at 1, 3, 6, and 9 months).
sion in common: The existence of important errors in the Secondary results of mother (in the first 6 weeks):
methodology of the studies, which devalues their con- Total duration of rehospitalization, proportion of
clusions.3-7 No evidence exists that supports or denies the cases requiring emergency consultation, proportion
usefulness of the practice, and with the available data, it of women with complaints of fatigue (Appendix B),
is not possible to define the ideal period of postpartum proportion of women with physical problems, such
hospitalization for mother and child, nor the necessity of as perineal.
domiciliary follow-up in the event of early discharge. Secondary results of newborn infant (in the first 28
Our objective was to perform a program of early days): Length of stay of readmissions, total duration
obstetric-pediatric discharge with domiciliary follow-up of rehospitalization, proportion of cases requiring
and to evaluate its advantages and disadvantages emergency consultations.
compared with the traditional postpartum hospital Evaluation of satisfaction (surveyed for satisfaction
stay, according to the established criteria for early.1,2,8 at 6 weeks).
Economic evaluation: Cost of stay from birth to
discharge, cost of the postnatal care required, such as
Material and methods health or lactation consultations, home visit pro-
grams, readmissions in hospitals (first 6 weeks), cost
Study population of the maintenance of the early discharge programs
(first 6 weeks).
The study consisted of puerperae and their neonates, at
the Hospital de Valme de Sevilla during the period of
April 1999 to April 2001, who fulfilled the criteria in Statistical method
Table I.
The sample size was fixed at 213 women in each group
Study description for a potential of 80%, an alpha error of 5%, and a
percentage difference in maternal or neonatal readmis-
The studies were nonexperimental with observational sion of 5.5% for a unilateral test. A randomization by
and clinical follow-up with no prenatal preparation and blocks (opaque sealed envelops) was performed within
with postpartum randomization. The 2 study groups, the 2 strata defined by the parity variable (primiparous,
intervention and nonintervention, received information, multiparous); the sample size within each group in these
and consent was obtained in writing. 2 strata was fixed, taking into account the distribution
Intervention group of the parity variable within our area of study.
Puerperae discharged in the first 24 hours postpartum, Statistical analysis
monitored at home over the next 24 to 48 hours by
a nurse qualified in puerperal and neonatal care (phys- Means and SDs or interquartile ranges for the numerical
ical revision of the mother with arterial pressure, oral variables were determined, whereas percentages were
temperature (T(), uterine involution, metrorrhagia, and obtained for the qualitative variables. The Student t test,
evaluation of the episiotomy; physical revision of the or the Mann-Whitney U test, was used to compare the
neonate with T(, and evaluation of icterus; informing numerical information in the 2 groups ‘‘early discharge’’
the mother and family members about puerperal and and ‘‘traditional discharge.’’ Significant results were
716 Sainz et al
Table I Criteria for early obstetric-pediatric discharge

There was compliance with the minimum criteria for early discharge according to the ACOGdAAP Committee on Fetus and Newborn;
Hospital stay for healthy term newborn infants8 and the requirements of the Guidelines for Perinatal Care,1 before the suspicion of any
pathology; there was no discharge without blood group analysis (and hence non compliance with the requirements for inclusion in
the study). Screening analyses and those for innate errors of metabolism in our health area are performed at the fifth day of life, and
therefore out of the hospital.
Gestation: Term gestation, primipara or multipara, with no important toxic habits. No existence of gestational pathology requiring
monitoring in the puerperium such as: Arterial hypertension-preeclampsia-eclampsia, gestational diabetes, Rh isoimmunization.
Where there was a good outcome, gestations of risk were included, such as threat of premature birth, previous cesarean.
Vaginal delivery: Presentation: cephalic-breech. Onset of birth: spontaneous or induced. Dilation: with or without usual medication
(oxytocin-analgesia-epidural anesthesia). Completion: spontaneous or instrument (forceps-suction-spatula). Birth: spontaneous or
manual expulsion but always complete. With or without episiotomy: excluding third- to fourth-degree tearing. Absence of fetal
suffering during birth.
Immediate puerperium: Adequate temperature (%38(C) and arterial pressure (%140-90), blood loss less than 500 mL and adequate
uterine involution.
Clinical puerperium: Puerpera in good general condition. Controlled pain and mobile. T( %38(C and arterial pressure %140-90, taken
every 8 h. Vulva-perineum-breasts in good condition. Moderate or minimal vaginal secretions of adequate characteristics. Normal
urine and presence of peristalsis. Correct uterine involution. Acceptable blood loss, with control hemogram (hematocrit !10% and
hemoglobin O9 g/mL). Know the blood groups of mother and neonate. Have supplies of anti-D gammaglobulin if necessary.
Adequate neonatal result: discharge from pediatrics: Term gestation with appropriate weight for gestational age. Normal
cardiorespiratory adaptation to extrauterine life. No evidence of sepsis or important jaundice. Adequate temperature (axilla 36.1(C-
37(C). Physical examination of the neonate by the doctor and a review within the 12 h before discharge that does not indicate any
necessity for additional observation and/or therapy in the hospital. Urinated and expelled meconium at least once. Received
vaccinations and necessary medications (eg, administration of vitamin K). Laboratory data available and reviewed, including
(maternal syphilis and hepatitis B, blood group of mother and neonate, Coombs’ test if indicated). Appointment to perform metabolic
screening. The mother is capable of providing routine care for the infant and is well aware of the signs of illness and other problems.
Acceptance of early discharge by the mother and family (signed the consent form), previous verification that the mother knows how to
care for the infant.
Residence in Dos Hermanas, Alcalá, Los Palacios, Seville (populations less than 20 km from the hospital) and with easy access to the
hospital.
complemented by confidence intervals for means to early discharge group (ED) and 2.3% in the control
95%. For qualitative variables, the tests used were the group (CO). Between the sixth week and 6 months,
c2 test for r ! s tables, and the c2 test with correction readmissions amounted to 0.94% (1.4% ED, 0.5% CO)
for continuity or Fisher exact test; in 2 ! 2 tables, the (Tables IV and V). We found puerperal pathology in
odds ratio and confidence intervals of the odds ratio 21.1% (16.6% ED, 22.9% CO) (Tables IV and V). If
were obtained for significant results. late puerperal pathology is included, then the rate rises
to 24.8% (19.9% ED, 25% CO) (Table V).
Results Anxiety-depressive pathology
Homogeneous study groups We found a score of more than 7 points on the hospital
anxiety and depression (HAD) scale in 15.4% of the
A total of 430 cases were evaluated (Table II); 15 cases cases at 1 week, 2.8% of cases (1.4% ED, 4.2% CO) with
(3.5%) did not wish to participate in the assigned group, a score of more than 10 points (requiring psychiatric
although they had previously given their consent; 14 monitoring), with 0.5% (0.5% ED, 0.5% CO) being
cases withdrew their consent (7 did not want the follow- serious cases (with psychiatric admission) (Table IV). At
up at 1 week, and 7 were scared of this type of practice). 1 month postpartum, only 2.3% (0.9% ED, 3.7% CO)
We consider a total of 22 cases lost (5.1%), those who were considered as pathologic and corresponded to the
did not accept the study group, and those who did not patients that were under psychiatric care (Tables IV).
want the follow-up at 1 week (Table III). All the cases
who did not accept the assigned study group were Puerperal fatigue
included in the results (analysis by intention to treat). Puerperal fatigue increased in the week of birth in
Maternal puerperal pathology 21.9% of the patients (20.1% ED, 23.5% CO), with
1.9% serious cases. At 1 month, only 1.2% of puerperae
In the study there was 2% maternal readmission (9 referred to an increased puerperal fatigue, none of which
cases) in the first 6 weeks postpartum, with 1.9% in the were serious (0.5% ED, 1.8% CO) (Table IV). The
Sainz et al 717
Table II Epidemiologic pregnancy-birth data

Total ED CO
Number 430 (100%) 213 (49.6%) 217 (50.4%)
Primiparous 160 (37.2%) 78 (18.2%) 82 (19%)
Multiparous 270 (62.8%) 135 (31.4%) 135 (31.4%)
Did not accept 15 (3.4%) 11 (2.6%) 4 (0.9%)
Epidemiologic data
Age (y)
% 19 18 (4.2%) 7 (3.3%) 11 (5%)
20-30 232 (53.9%) 117 (54.9%) 115 (53.1%)
31-40 173 (40.2%) 88 (41.3%) 85 (39.2%)
O40 5 (1.2%) 1 (0.5%) 4 (1.85%)
Education level
No studies 59 (13.7%) 32 (15%) 27 (12.4%)
Primary studies 291 (67.7%) 133 (62.4%) 158 (72.9%)
Secondary studies 57 (13.3%) 29 (13.6%) 28 (12.9%)
University studies 23 (5.3%) 19 (8.9%) 4 (1.8%)
Marital status
Single 12 (2.8%) 6 (2.8%) 6 (2.8%)
Married 418 (97.2%) 207 (97.2%) 211 (97.2%)
Birth data
History of cesarean section 7 (1.6%) 2 (0.4%) 5 (1.2%)
Average age at birth 29.47 G 1.79 29.51 G 4.94 29.43 G 5.8
Weeks of gestation 39.51 G 1.12 39.52 G 1.13 39.50 G 1.12
Onset of birth
Spontaneous 368 (85.6%) 179 (84%) 189 (87.1%)
Stimulated 34 (7.9%) 18 (8.5%) 16 (7.4%)
Induced 28 (6.5%) 16 (7.5%) 12 (5.5%)
Fetal presentation
Cephalic 427 (99.3%) 212 (99.5%) 215 (99.1%)
Pelvic 3 (0.7%) 1 (0.5%) 2 (0.9%)
Color of A. liquid
Clear 392 (91.2%) 191 (89.7%) 201 (92.6%)
Meconium 36 (8.4%) 21 (9.9%) 15 (6.9%)
Meconium CC 2 (0.4%) 1 (0.5%) 1 (0.5%)
T-dilation (h) 3.42 G 3.75 3.08 G 2.74 3.76 G 2.73
T-expulsion (min) 15.27 G 10.0 14.48 G 9.5 16.04 G 9.33
T-oxytocin (min) 199.5 G 164 179.3 G 160 219.4 G 166
T- of sack breakage 5.06 G 4.31 4.79 G 4.39 5.33 G 4.22
Completion
Normal 379 (88.1%) 187 (87.8%) 192 (88.5%)
Forceps/suction/spatula 51 (11.9%) 26 (12.2%) 25 (11.5%)
Tearing
No 392 (91.2%) 193 (90.6%) 199 (90.7%)
1 degree 10 (2.3%) 6 (2.8%) 4 (1.8%)
2 degree 28 (6.5%) 14 (6.6%) 14 (6.5%)
Sex of newborn infant
Male 211 (49.1%) 106 (49.8%) 105 (48.4%)
Female 219 (50.9%) 107 (50.2%) 107 (51.6%)
Newborn weight (g) 3349 G 384 3348 G 396 3335 G 372
Size (cm) 50.50 G 1.94 50.53 G 2.17 50.47 G 1.68
Time of stay (h) 39.3 G 14.5 27.23 G 9.12 51.25 G 7.29
Time of stay (d) 1.63 1.13 2.13
ED, Early discharge; CO, control group.
comparison of puerperal fatigue between the patients Health service consultations

who presented a puerperal pathology (mammary, in-
fectious, vascular, traumatic, psychiatric, or episiotomial Of the cases, 9% (8% ED, 10.1% CO) consulted a health
effects) and those who did not is expressed in Table V. service system in the 6 months postpartum (Table V).
718 Sainz et al
tained more information about the care of the infant,

Table III Lost cases
and themselves, with the home visit. Up to 90% of this
Total ED CO group indicated they would repeat the practice in
n = 430 n = 213 n = 217
a future birth.
Reasons
Fear 7 (1.63%) 7 (3.3%) 0 (0%) Evaluation of costs
Too far 1 (0.23%) 0 (0%) 1 (0.5%)
No to follow-up 7 (1.63%) 4 (1.9%) 3 (1.4%) In the GDR (Grupos de Diagnóstico Relationado)
Total 15 (3.5%) 11 (5.16%) 4 (1.84%) database9 of the COANH program (Contabilidad Ana-
Follow-up not 22 (5.1%) 8 (3.8%) 14 (6.5%) lı́tica de Hospitales), the unit GDR cost for hospital
attended obstetrics is $1725.20 US and $2269.50 US for pediatrics
ED, Early discharge; CO, control group. (average euro value $1.03 US with a range of 0.89-1.16.
Average estimated between July 1999 and August 2000.
Maternal lactation http://www.europa.eu.int). Starting with this GDR
value, the cost estimates for intervention types are given
Breastfeeding was monitored up to 9 months post- in Table VII.
partum, the results being expressed in Table IV. The The formula to calculate the daily cost per case is as
reasons for the abandonment of breastfeeding are as follows: GDR weighting ! GDR value/average stay.
follows: the infant was still hungry in 57.9% (162 cases), For the economic estimations in this study of early
working mother 11.4% (32 cases), the infant did discharge, we measured all those parameters recom-
not want the breast 7.9% (22 cases), the infant did mended by the reviewers.
not put on weight 6.4% (18 cases), and other reasons The results of the cost evaluations are expressed
12.8% (46 cases) (n = 280; distinct variables such as in Tables VIII, IX, and X. A saving of 18.3% was
type of discharge, parity, cultural level were not mea- obtained, representing $59,207.59 US over a total of 202
sured.) cases, and a saving of $293.11 US per patient by an early
discharge program with home visit and telephone
Neonatal pathology follow-up.
The rate of neonatal rehospitalization in the first 28 days
was 1.9% (1.4% ED, 2.3% CO) (relative risk [RR] 0.61
with 95% CI 0.15-2.56) (Table IV). The reasons for Comment
readmission in the early discharge group were as All the reviews of the practice of early discharge have
follows: 1 case of fever, 1 case of dehydration, and 1 conclusion in common, the existence of important
1 for icterus; in the control group: 3 cases of icterus, methodologic errors in the studies, which devalue their
1 case of fever, and 1 for important weight loss. The conclusions. Among the most common errors are that
neonatal pathologies that did not need readmission in many were observational, the lack of control groups, the
the first 28 days were as follows: 15 cases (7%) in the ED objectives were measured on a few occasions, the
group (2 cases of suspected cardiac pathology, 4 cases of majority of studies have too few cases to be statistically
infectious pathology, 6 cases of orthopedic pathology, significant, families were carefully selected for low
1 case of congenital spherocytosis, 1 case of cephalohe- socioeconomic risk, the inclusion of families with
matoma, and 1 case of hydrocele) and 10 cases (4.5%) in motives in favor of early discharge, and in many cases
the CO group (6 cases of infectious pathology, 2 cases of careful monitoring was applied after discharge.3-7 Var-
orthopedic pathology, 1 case of cephalohematoma, and ious authors describe important negative effects for the
1 case of umbilical hernia). neonate and mother in programs of early discharge.4,10-16
Evaluation of satisfaction Other authors, however, consider this practice to be
a family-based medical exercise that can potentially lead
In the satisfaction survey, a favorable valuation of this to lower risks of adverse effects.10,17-19
type of practice is observed: ‘‘early puerperal discharge For these reasons the reviewers indicate that well-
with home visit’’ (Table VI). Of those patients who were designed studies should be performed to evaluate the
in the CO group, more than 40% confirmed that they impact of different periods of postpartum hospitaliza-
would have been more satisfied with early discharge and tion, and that the studies should be randomized, pro-
a home visit. spective with control groups of mothers and neonates.
In the intervention group, more than 90% indicated Participants should be mothers with healthy neonates of
they were satisfied or very satisfied with the practice; more than 2500 g weight at term (37-42 weeks) and have
92% indicated that they felt better at home and that demonstrated normal development before discharge.
they preferred the home visit to the usual hospital The variables that should be evaluated in all studies
monitoring. Furthermore, they referred to having ob- on early discharge are the existence of complications in
Sainz et al 719
Table IV Maternal and neonatal results

Traditional Statistical
Results ED (%) discharge (%) RR 95% CI difference
Maternal readmission 4/213 (1.9%) 5/217 (2.3%) 0.81 0.21-3.03 None
Neonatal readmission 3/213 (1.4%) 5/217 (2.3%) 0.61 0.15-2.56 None
Maternal puerperal pathology 35/213 (16.6%) 50/217 (22.9%) 0.73 0.30-2.20 None
Maternal consultation 17/213 (8.0%) 22/217 (10.1%) 0.78 0.30-2.10 None
to the health system
Increase in anxiety-depressive 17/213 (8.0%) 27/217 (12.5%) 0.64 0.25-1.63 None
pathology at 1 wk
Increase in anxiety-depressive 2/213 (0.9%) 8/217 (3.7%) 0.30 0.33-3.20 None
pathology at 1 mo
Increase in puerperal 43/213 (20.1%) 51/217 (23.5%) 0.85 0.43-1.64 None
fatigue at 1 wk
Increase in puerperal 1/213 (0.5%) 4/217 (1.8%) 0.50 0.04-5.54 None
fatigue at 1 mo
Breastfeeding at 1 wk 205/213 (96.2%) 200/217 (92.2%) 0.48 0.14-1.65 None
Breastfeeding at 1 mo 190/213 (89.2%) 182/217 (83.9%) 0.58 0.25-1.36 None
Breastfeeding at 3 mo 141/213 (66.2%) 119/217 (54.8%) 0.62 0.42-0.91 Yes P = .016 !.05.
Fisher exact test
Breastfeeding at 6 mo 94/213 (44.1%) 76/217 (35.0%) 0.68 0.46-1.006 None
Breastfeeding at O 9 mo 42/213 (19.7%) 36/217 (16.6%) 0.81 0.49-1.32 None
ED, Early discharge; CO, control group.
the mother or child that require rehospitalization or In our study there was neither specific selection of
a medical consultation; the existence of maternal factors patients from a low-risk population nor exhaustive
such as fatigue, anxiety, level of self-confidence after the monitoring of the patients, aspects that have been
birth; problems of maternal lactation, satisfaction of the much criticized in other studies of this type3,4,6,19-21
mother and family; and an evaluation of the costs. (Table II). Studies of early discharge indicate the exis-
tence of 5% to 20% puerperal pathology.3,4,6,10,19-21 In
Homogeneous study groups the studies these pathologies are neither described nor
differentiated by clinical importance. In our study we
In our study, no differences were observed between the included all the puerperal pathologies that occurred and
intervention and control groups in epidemiologic char- found an incidence of 21.1% (Tables IV and V) from
acteristics (marital status, age, in the educational level which we found no statistically significant differences (in
a small unimportant difference existed in favor of the the evaluation of the diverse maternal pathologies,
intervention group), gestational histories, birth details average or late, possible in the puerperium) between
(weeks of gestation, onset to completion of birth, the 2 study groups. A diminution in health service
dilation times, expulsion or broken sack, reason for consultations for maternal reasons in the 6 months
induction), or details of the neonate (sex, Apgar score, postpartum is found in the early discharge group
weight) (Table II). It is demonstrated that both study (Tables IV and V).
groups were homogeneous and that there was no patient
selection by important variables such as cultural level,
Anxiety-depressive pathology
age, onset to evolution to completion of birth, or by
neonatal characteristics. In our study selections were In general, studies of early discharge report no increase
made after birth, and no previous contact existed with in anxiety-depressive pathology caused by early dis-
the patients, only 1 visit was made to the home, which is charge.22 Only Carty and Bradley20 and James et al23
considered indispensable in this type of discharge. Easy cite an increase in depressive problems in cases of early
access to health services was available to both the study discharge. Brown et al24 conclude that depression
groups. problems do not vary and value scores do not increase
with the length of hospital stay, but may do so with
Maternal puerperal pathology other variables, such as low economic level where there
may be dissatisfaction with a longer hospital stay, thus
The readmission rate (2%) in the first 6 weeks post- increasing the degree of maternal anxiety and discom-
partum for maternal reasons (Table IV) is similar to that fort and where there may have been a complicated
described in the literature on early discharge programs.3-6 pregnancy. Grullon and Grimes6 in their meta-analysis
720 Sainz et al
Table V Maternal results. Description of maternal pathology

Maternal risk in the first 6 wks Description
Early discharge (day of 1 case of postepisiotomial dehiscence (3)(P), 2 cases of puerperal fever (1 bartolin and
readmission) 1 endometritis) (2,5)(P, M) and 1 case of puerperal depression (10)(M)
Control (day of readmission) 1 case of postepisiotomial dehiscence (3)(P), 3 cases of puerperal fever (1 mastitis and 1
endometritis) (2,4,5)(P, P, M) and 1 case of puerperal depression (25)(P)
Distribution, by study groups, of puerperal pathology in the first 6 mo
Pathology in % ED CO Statistical difference
Mammary 6.1 8.4 None 11(5.2%) cases of cracking in ED and 14 (6.5%) in CO, 2(0.9%) mastitis in ED
and 3 (1.4%) in CO. In CO, 1 (0.5%) tumor case (fibroadenoma not
previously diagnosed)
Infectious 1.6 1.5 None 3(1,4%) cases of UTI in ED y 1(0,5%) in CO. 1(0,5%) case of endometritis in
each group (ED, CO) and 1(0,5%) of RI in CO
Vascular 0.5 0.5 None 1(0,5%) case per group (ED, CO) of superficial phlebitis and no case of DVT or PTE
Orthopedic 3.3 2.3 None 6(2,8%) cases of sciatica in ED and 4(1,8%) in CO, 1(0,5%) case per group
(ED, CO) of pubic diastasis
Psychiatric 1.4 4.2 None 3(1,4%) cases required psychiatric follow-up in ED and 9(4,2%) in CO
Episiotomial 3.7 6.0 None 6(2,8%) cases of open episiotomy in ED and 7(3,2%) in CO
changes
16.6 22.9 None
Late puerperal 3.3 4.1 None
pathology
19.9 25.0 None
Relation between puerperal fatigue and the existence or not of puerperal pathology. In the evaluation of maternal fatigue, the cases
that did not go to the follow-up (7 cases) were not included, because they did not complete the appropriate questionnaire.
FSS Puerperal pathology No puerperal pathology
Normal (7-12) 48 (51.6%) 265 (84.4%)
Slight-moderate (13-29) 41 (44.1%) 45 (14.3%)
Serious (30) 4 (4.3%) 4 (1.3%)
Lost 23.22 for not responding to the FSS and 1 lost to the system when making the statistical study
Statistical difference Pearson’s c2 P ! .0005
Maternal pathology between 6 wks and 6 mo from birth
ED CO Statistical
difference
Distribution 7 9 None In ED, 4 cases of mastitis (2P, 2M), 1 bartolin (M), 1 transitory dysarthria (M)
(3.3%) (4.1%) and 1 facial paralysis (M). In CO, 2 UTI (M,P), 1 repetitive UTI-renal
calculus-lithotripsy (M), 2 mastitis (2M), 1 RI (M), 1 moderate depression
(P), 2 late puerperal hemorrhages (2M)
Admission 3 1 In ED, 1 mastitis (2), 1 transitory dysarthria (10) and 1 bartolin (2). In CO,
(1.4%) (0.5%) 1 repetitive UTI - renal calculus-lithotripsy (5)
Consultations for maternal pathology between 6 wks and 6 mo
ED CO Statistical difference
Distribution 17 (8%) 22(10%) None
Health service consultations for neonate pathology in the first 28 days
ED CO Statistical difference
Distribution 18 (8%) 13 (6%) None
ED, Early discharge; CO, control group; FSS, fatigue severity scale; P, primiparous; M, multiparous.
refer to quality evidence (type II-2, recommendation C) test is that it is not easy to distinguish anxiety from
with respect to lack of increase in anxiety-depression depression with the factors that are measured.27 This
problems in cases of early discharge. point was not considered to be a problem because we
We used the HAD scale25 for measurement, as it is did not attempt to distinguish the 2 pathologies but only
easy to apply and gives rapid results with a high degree to identify any problem that could be assessed later by
of reliability and validity.26,27 The main criticism of this psychiatrists. In our study we found an incidence of
Sainz et al 721
Table VI Evaluation of satisfaction

Total ED CO
Were you satisfied with your postnatal treatment?
No 14 (4.7%) 2 (1.2%) 12 (9.6%)
Adequate 65 (22%) 28 (16.3%) 37 (29.6%)
Satisfied 131 (44%) 69 (40.1%) 62 (49.6%)
Very satisfied 87 (29.3%) 73 (42.4%) 14 (11.2%)
Were you satisfied with the answers to your question?
No 23 (7.8%) 5 (2.9%) 18 (14.4%)
Adequate 41 (13.8%) 29 (16.9%) 12 (9.6%)
Satisfied 170 (57.2%) 96 (55.8%) 74 (59.2%)
Very satisfied 63 (21.2%) 42 (24.4%) 21 (16.8%)
For cases of 2 or more days stay. Do you think you would have
found it better to have a shorter stay and a home visit?
No 38 (30.4%)
Indifferent 33 (26.4%)
Better 54 (43.2%)
For early discharge cases. How did you value the program?
Unsatisfied 5 (2.9%)
Indifferent 12 (7%)
Satisfied 94 (54.6%)
Very satisfied 61 (35.5%)
For early discharge cases. What did you find better?
Intrahospital monitoring 13 (7.6%)
Home visit by health personnel 159 (92.4%)
For early discharge cases. When did you receive most information
about care for yourself and your infant?
The hospital follow-up 13 (7.6%)
Home visit by health personnel 147 (85.5%)
Equal 12 (6.9%)
In the event of another pregnancy, and all went well, would you prefer
Traditional monitoring 59 (19.8%)
Early discharge 193 (65%) 5 (2.9%) 38 (26.4%)
Indifferent 45 (15.2%) 155 (90.1%) 54 (43.2%)
anxiety-depressive pathology in the study groups. (We would be expected, statistically significant differences
did not find differences in the incidence of anxiety- are found (P ! .0005 c2, Pearson) (Table V).
depression between the study groups).
Maternal lactation
Puerperal fatigue
It is interesting to note that there are statistically
All the programs of early discharge recommend that significant differences between the groups with regard
maternal puerperal fatigue be measured, but the re- to the continuation of breastfeeding to 3 months in
quired number of controls is not stated nor which scale favor of the early discharge group (P = .016 !.05
should be used.1,7,19 Smith-Hanrahan and Deblois28 Fisher exact test). At 6 months there is the same
found that early discharge with adequate home follow- statistical trend that is almost significant (P = .053 O
up did not affect maternal fatigue or functional ability to .05 Fisher exact) (Table IV).
any significantly greater extent than traditional care. We Most of the studies on early discharge that give
used the ‘‘Fatigue Severity Scale’’ of Krupp et al,29 information on breastfeeding indicate that this is not
which is widely used for the measurement of fatigue in affected by the short stay in hospital.19,20 Brown et al24
medical pathologies and is easy to apply. There were no conclude that breastfeeding is not affected by the length
differences in puerperal fatigue because of early dis- of hospital stay, but is by other variables. Dershewitz
charge at 1 week or at 1 month postpartum (Table IV). and Marshall30 mention that in cases of early discharge
If puerperal fatigue is measured in patients who present there is a tendency for the discontinuation of breastfeed-
a puerperal pathology and in those who do not, then as ing, and they advise domiciliary aid for these mothers.
722 Sainz et al
Table VII Intervention values

Average Daily case cost in
GDR Description Weighting stay (d) US dollars ($)
373 Vaginal birth with no complications. 0.484 2.78 303.60
376 Postpartum and postabortion diagnoses without surgery 0.559 4.09 235.80
628 Neonate; birth weight O2500 g, with minor problems 0.689 7.17 218.08
but without significant surgery.
627 Neonate; birth weight O2,500 g, with major problems 1.026 8.79 264.90
but without significant surgery.
427 Septicemia, !18 y 1.004 8.17 278.89
423 Other diagnoses of infectious and parasitic illnesses. 1.074 8.33 292.61
422 Virosis and fever of unknown origin, age !18 y 0.590 4.17 321.10
Cost of external consultation 39.27
Cost of emergency consultation 23.36
Cost of home visit 39.15
Adapted from GDR 1999.9
Table VIII Costs in US dollars for ED and CO groups

No. Average Cost (US$) Total
Length of stay ED 202 1.13 (1.13 ! 303.60) C 39.15 77,208.00
CO 202 2.13 2.13 ! 303.60 130,626.90
No. (d) Cost (US$) Total
Maternal readmission ED 20 235.80 4,716.00
CO 39 235.80 9,196.20
Neonatal readmission ED 16 218.08 3,489.20
CO 22 218.08 4,797.70
Maternal consultations ED 17 23.36 397.10
to the health systems CO 22 23.36 513.90
Neonatal consultations ED 18 23.36 420.40
to the health systems CO 13 23.36 303.60
Maternal-neonatal review ED 2 ! 202 39.27 15,865.10
at 1 wk CO 2 ! 202 39.27 15,865.10
10-min interprovincial ED 3 ! 202 0.08 US $ 412.08
telephone call (141) CO 3 ! 202 connection C 0.06 412.08
US $/min = 0.68 US $
Total sum 264,223.73
Cooper et al31 report severe cases of dehydration in for the abandonment of breastfeeding, we have only
breastfed children despite good maternal preparation or made a descriptive valuation without considering the
good post discharge follow-up. In the review made by diverse influencing factors. We found that the reasons
the Cochrane Library, it was concluded that early for abandonment were very similar to those in the
discharge can be of benefit in the continuation of consulted literature,33-35 the most frequent reason
breastfeeding, with a valuation II-2 and recommenda- (57.8%) being that ‘‘the infant is still hungry.’’
tion type C, although more studies are needed to
determine which factors influence the duration of
breastfeeding, and which is the best type of domiciliary Neonatal pathology
aid for those mothers that have had an early discharge.32
In our study, we believe that the difference between Hospital readmission rates are between 0.6% and
the continuation of breastfeeding at 3 and 6 months is 5%.10,11,36 Liu et al,11 in a large, well-designed study,
due to the combination of the recommendations made gives a rate of neonatal rehospitalization of 2%.
by the specialized staff who visited the mothers at home, Significant differences in the rates of rehospitalization
giving information about the technique and the benefits between infants from early discharge groups and con-
of breastfeeding, particularly when doubts and problems trols have not been found.6,10,18,37-41 Some studies
about breastfeeding have started. Regarding the reasons observe that the rate of rehospitalization of neonates
Sainz et al 723
levels of 16-18 mg/dL), the second most frequent being

Table IX Costs in the ED group
fever at 25%. Thus, the data found in our study agree
ED group Cost (US$) with those in the global literature for early dis-
Hospital stay 77,208.03 charge.3,5,11,16,18,30,37-43
Maternal readmission 4,716.00 As far as other data are concerned, it should be
Neonatal readmission 3,489.28 emphasized that the time of readmission depends on the
Maternal consultations to the health systems 397.10 type and severity of the pathology and the response of
Neonatal consultations to the health systems 420.48
the neonate to treatment. Regarding consultations with
Maternal-neonatal review at 1 wk 15,865.10
Telephone calls 412.08
the health services, we did not find any diminution for
Total sum 102,508.07 early discharge.
Evaluation of satisfaction
Table X Costs in the CO group In the satisfaction survey, a favorable valuation of this
CO group Cost (US$) type of practice is observed: ‘‘early puerperal discharge
with home visit’’ (Table VI). The studies of Lieu et al45
Hospital stay 130,626.94
and Escobar et al46 also find high maternal satisfaction
Maternal readmission 9,196.20
Neonatal readmission 4,797.76
in this practice.
Maternal consultations to the health systems 513,90 Our aim is to provide the mother and child with the
Neonatal consultations to the health systems 303,68 best possible health care and not only to achieve a cost
Maternal-neonatal review at 1 wk 15,865.1 saving, for this reason we believe that the home visit by
Telephone calls 412,08 qualified health personnel is fundamental and indispens-
Total sum 161,715.66 able for early discharge programs. We base this affir-
mation on the fact that health personnel:
Monitor puerperal pathology in the 48 to 72 hours
from early discharge groups is lower than that of postpartum;
neonates from traditional discharge groups. However, Monitor neonatal pathology in the 48 to72 hours
these low rates were produced in low-risk groups in postpartum;
which all the neonates were receiving good monitoring Resolve doubts and problems about neonatal and
care.18-21,23,41 puerperal care when they occur, and in the comfort
Hyperbilirubinemia is the most frequent medical of the maternal home environment;
problem. According to Catz et al15 etween 1% and Check and correct the breastfeeding technique, re-
4% of suckling term neonates are rehospitalized during solve problems, and consolidate the recommenda-
the first week of life. Of these readmissions, 45% to 85% tions about breastfeeding.
were due to jaundice.4,6,36,42,43 Many studies do not
Brown et al,24 in their protocol ‘‘Early postnatal
specify what concentration of total bilirubin is sufficient
discharge from hospital for healthy mothers and term
to justify readmission. Seidman et al13 suggest the
infants,’’ proposes a division between the study sub-
adoption of a uniform definition of serious hyperbilir-
groups of those with and without home visits. Although
ubinemia, treatable in the home by phototherapy with
it has been demonstrated that neonatal and puerperal
a reduction in the need for readmission. After the
pathology do not increase in early discharge groups
change of therapeutic approach in these infants, by
without home visits, we believe that the home visit is
deferring phototherapy until the serum bilirubin levels
necessary for improvement in the quality of assistance
increase beyond 18 mg/dL,13,44 jaundice ceased to be an
for the mother. Also, we believe that the increase in
important problem.
cases of neonatal jaundice and dehydration in some
In our study we did not find statistically significant
early discharge protocols4,11-15 are due to patients
differences in the rate of readmission between the study
foolishly abandoning the Health Systems. With moni-
groups. The rate of neonatal rehospitalization in the first
toring by, and contact with, health practitioners we do
28 days’ postpartum in the early discharge group was
not think that this happens and that attention for these
1.4%, and 2.3% in the control group (RR 0.61 with
patients is secured.
95% CI 0.15-2.56). It must be emphasized that this
difference is due to hyperbilirubinemia (33% ED vs 66%
CO). The home visit by staff qualified in the control and Evaluation of cost
assessment of neonatal care seems to be the cause of this
benefit. Among the reasons for readmission, the most Gazmararian and Koplan,47 in their review of the
frequent is hyperbilirubinemia at 50% (with readmission topic, conclude: The frequent supposition is that early
724 Sainz et al
discharge is cost effective. However, the true cost of by important variables such as cultural level, age,
these programs, when compared with traditional hospi- initiation-evolution- completion of the birth, or by
tal stay costs, have not been analyzed with great care. neonatal characteristics.
The few studies that have examined the topic of cost, We made no previous preparation for the early
regularly do so in terms of costs to the hospital, the discharge group. Neither a more rigorous follow-up if
patients, or to third-party payers, and do not consider there was easy access to health services for any type of
the true costs of an early discharge program. The few problem.
attempts to make an analysis of cost-effectiveness do not We did not find differences between the study groups
adjust by the standards formulated for these studies.48,49 in: the rate of maternal or neonatal rehospitalization,
In theory, the cost savings for an early discharge maternal puerperal pathology, maternal fatigue, anxi-
program should be equal to the extra costs of a tradi- ety-depressive pathology, or consultations to the health
tional stay for mother and child together with the systems for maternal or neonatal pathology.
program development costs.50 Lukacs51 notes 4 essential There were statistically significant differences between
points for determining the true costs of these programs: the study groups in the maintenance of breastfeeding at
(1) cost of program development, (2) cost of home visits 3 months in favor of the early discharge group. At 6
(including the carer’s time in the home and in trans- months the trend was maintained but lacked statistical
portation), (3) number of visits per patient, and (4) cost significance.
of legal responsibility. Early obstetric-pediatric discharge gives a verified
Most programs of early discharge, in which costs have improvement in the well-being of mother and family.
been evaluated, report a saving,47,52-55 but other authors, Early obstetric-pediatric discharge carries an impor-
such as Lieu et al,45 Escabar et al,46 and Annas,56 agree; tant saving for Health Services.
others, such as Gazmararian and Koplan,47 criticize on
the grounds that the cost evaluations were not performed
in an adequate manner. Grullon and Grimes6 qualify the References
scientific evidence on the cost evaluations in these
programs as type C or inadequate. Brown et al24 and 1. American Academy of Pediatrics, American College of Obstetri-
Lukacs51 make reference to those parameters that should cians and Gynecologists. Postpartum and follow-up care. In:
be included in cost evaluations. Guidelines for perinatal care. 4 ed. Washington (DC): American
College of Obstetricians and Gynecologists; 1997. p. 147-82.
The economic evaluations in our study of early 2. American Academy of Pediatrics, American College of Obstetri-
discharge included all those parameters recommended cians and Gynecologists. Postpartum and follow-up care. In:
by the reviewers,24,47,48 and we obtained a saving of Guidelines for perinatal care. 3 ed. Washington (DC): American
18.3% (for a program of early discharge with home visit College of Obstetricians and Gynecologists; 1992. p. 91-116.
and telephone follow-up). 3. Margolis LH. A critical review of studies of newborn discharge
timing. Clin Pediatr 1995;34:626-34.
If the early discharge program includes a home visit 4. Braveman P, Egeter S, Pearl M, Marchi K, Miller C. Problems
and a hospital review of mother and child at 1 week, associated with early discharge of newborn infants: early discharge
a saving over traditional discharge is obtained of of newborns and mothers-a critical review of literature. Pediatrics
$43,342.49 US for 202 cases, representing a 13.4% 1995;96:716-26.
saving or $214.57 US per patient. Thus, for the 2,800 5. Renfrew MJ, Lang S. Early vs late discharge postpartum.
Cochrane Database Syst Rev 1998.
births, without complications, that occurred in the 6. Grullon KE, Grimes DA. The safety of early postpartum
Hospital de Valme de Sevilla in 1999,9 a saving might discharge: a review and critique. Obstet Gynecol 1997;90:860-5.
have been obtained in: 7. Conseil d’Evaluation des Technologies de la Sante du Quebec.
Evaluation of the risks and benefits of early postpartum discharge.
A system of early discharge with home visit and Quebec: Minister of Health and Social Services of Quebec; 1997.
telephone follow-up: $820,680 US. 8. Fetus and Newborn Committee, Canadian Paediatric Society;
A system of early discharge with home visit and Maternal Fetal Medicine and Clinical Practice Committees,
hospital review of mother and child at 1 week: Society of Obstetricians and Gynaecologists of Canada. Facilitat-
ing discharge home following a normal term birth. Paediatric
$600,768 US. Child Health 1996;1:165.
Our data indicate that a saving in the range of 18% to 9. Grupos de Diagnóstico Relationado (GDR) 373 Parto vaginal sin
diagnóstico complicado. En: Memoria anual del Área Hospitalaria
20% exists over the cost of traditional discharge with de Valme. Sevilla, Spain; 1999. p. 55.
a minimum stay of 48 hours. 10. Britton JR, Britton HL, Beebe SA. Early discharge of the term
In conclusion, we approached the study of early newborn: a continued dilemma. Pediatrics 1994;94:291.
obstetric-pediatric discharge according to the recom- 11. Liu L, Clemens CI, Shay D, Davis R, Novack AH. The safety of
mendations in the meta-analysis of the topic and newborn early discharge: the Washington State experience. JAMA
1997;278:293-8.
measured the indicated variables. 12. Edmonson MB, Stoddard JJ, Owens LM. Hospital readmission
It is demonstrated that both study groups were with feeding-related problems after early postpartum discharge of
homogeneous and that there was no patient selection normal newborns. JAMA 1997;278:299-303.
Sainz et al 725
13. Seidman DS, Stevenson DK, Ergaz Z, Gale R. Hospital read- 37. Brown LP, Towne SA, York R. Controversial issues surrounding
mission due to neonatal hyperbilirubinemia. Pediatrics 1995; early postpartum discharge. Nursing Clin North Am 1996;31:333-8.
97:727-9. 38. Sutter AL, Leroy V, Dallay D, Verdoux H, Bourgeois M. Post-
14. Maissel MJ, Newman TB. Kernicterus in otherwise healthy, partum blues and mild depressive symptomatology all days three
breast-fed newborns. Pediatrics 1995;96:730-3. and five after delivery. J Affect Dis 1997;44:1-4.
15. Cartz C, Hansin JW, Simpson L, Yaffe S. Summary of worshop: 39. Bragg EJ, Rosenn BM, Khoury JC, Miodovnik M, Siddiqi TA.
early discharge and neonatal hyperbilirubinemia. Pediatrics The effect of early discharge after vaginal delivery on neonatal
1995;96:743-4. readmission rates. Obstet Gynecol 1997;89:930-3.
16. MacDonald MG. Hidden risks: early discharge and bilirubin 40. Kotagal UR, Atherton HD, Bragg E, Lippeert C, Donovan EF,
toxicity due to glucose-6-phosphate dehydrogenase deficiency. Perlstein PH. Use of hospital-based services in the first three
Pediatrics 1995;96:734-8. months of life: impact of an early discharge program. J Pediatrics
17. Hellman LM, Kohl SG, Palmer J. Early hospital discharge in 1997;130:250-6.
obstetrics. Lancet 1962;1:227-32. 41. Rhodes MK. Early discharge of mothers and infants following
18. Yanover MJ, Jones D, Miller MD. Perinatal care of low-risk vaginal childbirth at the United States Air Force Academy:
mothers and infants: early discharge with home care. N Engl J a three-year study. Military Med 1994;159:227-30.
Med 1976;294:702-5. 42. Liu S, Wu Wen S, McMillan D. Increased neonatal readmission
19. Waldenstrom U, Sundelin C, Lindmak G. Early and late discharge rate associated with decreased length of hospital stay at birth in
after hospital birth: hearth of mother and infant in the postpartum Canada. Can J Public Health 2000;91:46-50.
period. Upsala J Med Sci 1987;92:301-14. 43. Kessel W, Kiely M, Nora AH, Sumaya CV. Early discharge: in the
20. Carty EM, Bradley CF. A randomized, controlled evaluation of end, it is judgment. Pediatrics 1995;96:739-42.
early postpartum hospital discharge. Birth 1990;17:199-204. 44. American Academy of Pediatrics. Management of hiperbilirubi-
21. Lemmer C. Early-discharge: outcomes of primiparas and their nemia in the healthy term newborn. Pediatrics 1994;94:558-65.
infants. J Obstet Gynecol Neonatal Nurs 1987;16:230-6. 45. Lieu TA, Braveman PA, Escobar GJ, Fischer AF, Jensvold NG,
22. Beck CT, Reynolds MA, Rutowski P. Maternity blues and Capra AM. A randomized comparison of home and clinic follow-
postpartum depression. J Obstet Gynecol Neonatal Nurs up visits after early postpartum hospital discharge. Pediatrics
1992;21:287-93. 2000;105:1058-65.
23. James ML, Hudson CN, Gebski VJ, Browne LH, Andrews GR, 46. Escobar GJ, Braveman PA, Ackerson L, Odouli R, Coleman-Phox
Crisp SE, et al. An evaluation of planned early postnatal transfer K, Capra AM, et al. A randomized comparison of home visits and
home with nursing support. Med J Aust 1987;147:434-8. hospital-based group follow-up visits after early postpartum
24. Brown S, Lumley J, Small R. Early obstetric discharge: does it discharge. Pediatrics 2001;108:719-27.
make a difference to health outcomes? Paediatr Perinat Epidemiol 47. Gazmararian JA, Koplan JP. Aspectos económicos de estancia
1998;12:49-71. intrahospitalaria perinatal. Egreso perinatal temprano del hospital.
25. Zimond AS, Snaith RP. The hospital anxiety and depression scale. Problemas y Preocupaciones. Clı́nicas Obstétricas y Ginecológica
Acta Psychiatr Scand 1983;67:361-70. 1992:529-45.
26. Tejero A, Guimerá E, Farré JM, Peri JM. Uso clı́nico del HAD 48. Siegel JE, Weinstein MC, Russell LB, Gold MR. Recommenda-
(Hospital Anxiety and Depression Scale) en población psiquiátr- tions for reporting cost-effectiveness analysis. JAMA 1996;
ica: un estudio de su sensibilidad, fiabilidad y validez. Rev Dpto 276:1339-41.
Psiquiatr Fac Med Barc 1986;13:233-8. 49. Weinstein MC, Siegel JE, Gold MR. Recommendations of the
27. Andersson E. The hospital anxiety and depression scale: homoge- panel on cost-effectiveness in hearth and medicine. JAMA
neity of the subscales. Soc Behav Pers 1993;21:197-204. 1996;276:1253-8.
28. Smith-Hanrahan C, Deblois D. Postpartum early discharge: 50. Feldman WE. Evaluation of an early newborn discharge program.
impact on maternal fatigue and functional ability. Clin Nurs Res HMO Pract 1993;7:48-50.
1995;4:50-66. 51. Lukacs A. Issues surrounding early postpartum discharge: effects
29. Krupp LB, LlaRocea NG, Muir-Nash J, Steineberg AD. on the caregiver. J Perinat Neonatal Nurs 1991;5:33.
The fatigue severity scale. application to patients with multiple 52. Elixhauser A, McCarthy E. Clinical classifications for health policy
sclerosis and systemic lupus erythematosus. Arch Neurol 1989; research, version 3: hospital inpatient statistics. Health Care Cost
46:1121-3. and Utilization Project (HCUP-3) Research Note 1. Rockville
30. Dershwitz R, Marshall R. Controversies of early discharge (MD): US Dept of Health and Human Services, Agency for Health
of infants well-newborn nursery. Curr Opin Pediatr 1995;7: Care Policy and Research; 1996. AHCPR publication 96–0017.
494-501. 53. Hickey L, DeRoeck E, Shaw M. En: Maternity day care program
31. Cooper WO, Atherton HD, Kahana M, Kotagal UR. Increased offers economical, family-oriented care. Hospitals 1977;51:85-6.
incidence of severe breastfeeding malnutrition and hypernatremia 54. Cottrell DG, Pittala B, Hey D. One day maternity care: a pediatric
in a metropolitan area. Pediatrics 1995;96:957-60. view. J Am Osteopath Assoc 1986;83:216.
32. Brown S, Small R, Faber B, Krastev A, Davis P. Early postnatal 55. Brumfield CG, Nelson KG, Stotser D, Yarbaugh D, Patterson P,
discharge from hospital for healthy mothers and term infants. Sprayberry NK. 24-hour mother-infant discharge with a follow-up
Cochrane Database Syst Rev 2002:CD002958. home health visit: results in a selected Medicaid population. Obstet
33. Comité de Lactancia Materna de la Asociación Española de Gynecol 1996;88:544-8.
Pediatrı́a. Informe técnico sobre la lactancia materna en España. 56. Annas GJ. Women and children first. N Engl J Med 1995;
An Esp Pediatr 1999;50:333-40. 333:1647-51.
34. Moran Rey J. Lactancia materna en España. Situación actual. An
Esp Pediatr 1992;36:45-50.
35. Garcı́a N, Oria C, Sánchez M, De Miguel JR. Influencia de la
educación sanitaria en la lactancia materna. Toko-Gin Práct
1989;8:63-467. Appendix A
36. Lee CY, Madrazo B, Drukker BH. Ultrasonic evaluation of the
postpartum uterus in the management of postpartum bleeding. We used the ‘‘Hospital Anxiety and Depression scale
Obstet Gynecol 1981;58:227-32. (HAD)’’ at weeks 1 and 4 to measure anxiety and
726 Sainz et al
depression. Designed by Zigmond and Snaith in Appendix B

1983,25 it is an autoevaluation scale with 14 points
of measurement, is easy to apply, and gives rapid We used the ‘‘Fatigue Severity Scale’’ of Krupp et al29 in
results with a high degree of reliability and valid- the first 6 weeks. It is commonly used for the measure-
ity.26,27 A score of 0 to 7 is considered normal, 8 to ment of fatigue in medical pathologies and is easy to use.
10 indicates a probable case of pathology, and A score of 12 to 19 is considered as slight fatigue, and 20
more than 11 a pathologic case of anxiety or de- or more as serious. A score of 5 or more in any 1 item of
pression.27 measurement is also considered a pathologic factor.
www.ajog.org
Impact of maternal-fetal surgery for myelomeningocele

on the progression of ventriculomegaly in utero
Amy Adelberg, MD,a,* Angela Blotzer, BS,b Gary Koch, PhD,b Rachael Moise,a
Nancy Chescheir, MD,a Kenneth J. Moise Jr, MD,a Honor Wolfe, MDa
Division of Maternal Fetal Medicine, Department of Obstetrics and Gynecology, University of North Carolina
School of Medicinea; Department of Biostatistics, University of North Carolina School of Public Health,
Chapel Hill, NCb
Received for publication June 16, 2004; revised January 13, 2005; accepted February 7, 2005
KEY WORDS Objective: Intrauterine myelomeningocele (MMC) repair decreases hindbrain herniation and the
Myelomeningocele need for postnatal ventriculoperitoneal shunting. We examined the impact of intrauterine repair
In utero repair on the progression of ventriculomegaly in utero.
Fetal surgery Study design: Fetuses with MMC were identified through computerized databases from June
Ventriculomegaly 1988 to April 2003. A retrospective cohort design was used to evaluate the impact of intrauterine
repair on ventricular progression with a multivariate linear regression model that included
baseline ventricle measurement, gestational age, level of lesion, and gender.
Results: Fourteen fetuses with intrauterine repair and 39 fetuses with postnatal repair were
identified. The natural history of progression of ventricular diameter increased in a linear fashion
throughout gestation (0.57 mm/week). After adjusting for confounding variables, no transient or
sustained difference was observed in the rate progression of ventriculomegaly between
intrauterine and postnatal repair (0.27 G 0.35 mm/week; P = .45).
Conclusion: Intrauterine MMC repair does not impact the progression of ventriculomegaly.
Neural tube defects are common and serious birth In utero repair of myelomeningocele (MMC), first
defects that contribute to infant morbidity and mortal- reported in 1997, was proposed in an effort to limit
ity. Despite folic acid supplementation, the incidence of damage to exposed fetal neural tissue from mechanical
neural tube defects in the United States is estimated at trauma and the toxic effect of amniotic fluid.3 Subse-
3000 per year.1 The estimated annual medical and quent studies have demonstrated additional benefits,
surgical cost is 200 million dollars,2 underscoring the including an almost universal reversal of hindbrain
importance of research efforts to decrease the significant herniation,4,5 and a decreased need for postnatal ven-
morbidity and mortality associated with this disease. triculoperitoneal shunting.6,7 Although the benefits to in
utero repair have been shown in neonates, there is little
* Reprint requests: Amy M. Adelberg, MD, 214 MacNider, CB
information on its influence on fetal brain development.
7516, University of North Carolina, Chapel Hill, NC 27599-7516. It has been observed that some neonates who do not
E-mail: adelberg@med.unc.edu have overt hydrocephalus at birth will acutely develop
doi:10.1016/j.ajog.2005.02.071
728 Adelberg et al
hydrocephalus in the days or weeks following postnatal for all control fetuses. Fetuses with anomalies were
repair.8 Deleterious effects of hydrocephalus such as excluded from the control group unless a clubbed foot
symptoms of stridor, facial palsy, and poor feeding will was present. Each patient in the cohort had a baseline
resolve in many cases after treatment. It is uncertain if ventricle measurement performed at our institution. For
this acute increase in ventricle size following postnatal the case and control fetuses, this measurement was
repair occurs in fetuses after in utero repair. To date, no obtained at the initial ultrasound performed at our
study has examined changes in ventricular size during in institution. For cases, this measurement was obtained
fetuses with MMC after in utero repair compared with before in utero repair. Both case and control fetuses had
control fetuses. a minimum of 1 subsequent measurement at least 2
The purpose of this study is to evaluate our hypoth- weeks apart.
esis that in the weeks following in utero repair, there
would be an acute increase in the ventricular size when
compared with fetuses with MMC undergoing postnatal Statistical analysis
repair.
The progression of ventricular size of the control fetuses
was first determined using a multivariate linear regres-
Material and methods sion model that controlled for the following variables:
level of spinal cord lesion, gender, and ventricular
Following approval by the University of North Carolina measurement at baseline and at each follow-up visit.
Institutional Review Board, computerized departmental Because there was not a standardized interval for repeat
ultrasound and perinatal hospital discharge databases ultrasound assessments of ventricular size, the general-
from June 1988 to April 2003 were queried to identify ized estimating equation method was applied to the
fetuses and newborns with prenatally diagnosed MMC. analysis to account for the differences in the timing of
Ultrasound reports were reviewed for gestational age these measurements.
and ventricular size at each ultrasound study, as well as The cases of in utero repair were then added and
fetal gender and level of the spinal lesion. Maternal analyzed with the control fetuses in the same linear
medical records were reviewed for maternal age, race, regression model. An additional interaction term of type
and gestational age at delivery. Postnatal imaging stud- of repair was added to evaluate for differences in the rate
ies were reviewed to confirm the diagnosis of MMC. of ventricular progression between case and control
The maximum width of the atria of the lateral fetuses. We considered that an increase in ventricular
ventricle was measured from the medial to the lateral size with advancing gestational age might not occur in a
wall across the posterior aspect of the choroid plexus. linear fashion. In order to detect an acute or transient
This measurement was recorded at baseline and at each change in ventricular progression after in utero repair, a
subsequent ultrasound. Mild, moderate, and severe quadratic equation (the square of gestational age) and
ventriculomegaly were defined as 10 to 14 mm, 15 to its interactions with the other variables were included in
24 mm, and R25 mm, respectively.9 Sonographers the full model.
performed all prenatal ultrasounds at the University of Comparisons of demographic data were made using
North Carolina Prenatal Diagnostic Center, and were chi-square and Fisher exact tests for categoric variables,
directly reviewed by Maternal-Fetal Medicine physi- and the Wilcoxon rank sum test for continuous varia-
cians. Examinations were performed with ATL HDI bles. P values of less than .05 were considered to indicate
5000 (Phillips Medical Systems, Bothell, Wash), Ultra- statistical differences, and 95% CI were used to describe
mark 4 (Advanced Technology Laboratories, Bothell, comparisons.
Wash), General Electric RT3200 Advantage (Milwau-
kee, Wis), or Corometrics 650 (Wallingford, Conn)
machines with 5 MHz or 7 MHz transabdominal Results
probes. All baseline ultrasounds and follow-up ultra-
sounds were performed at UNC, with the exception of A total of 78 fetuses with MMC were identified (Figure).
the follow-up scans in 3 fetuses undergoing in utero Seventeen patients (15 control fetuses and 2 cases) were
repair. These ultrasound reports were obtained from excluded from analysis because they did not have a
their referring institutions and reviewed. Maternal-Fetal minimum of 2 ventricular measurements. Thirty-nine
Medicine specialists supervised each of these referring control fetuses that underwent postnatal repair, and 16
institution’s ultrasounds. cases undergoing in utero repair were included in the
The control group was composed of fetuses with analysis. Ten cases of maternal-fetal surgery were
MMC who underwent postnatal repair. Cases were performed at the University of North Carolina. One of
fetuses that underwent in utero repair. The karyotype these patients delivered preterm and did not have a
was normal for all cases but was not obtained prenatally follow-up ventricular measurement, and was excluded
Adelberg et al 729
Table I Number of ultrasound measurements of ventricular

size
Number of ventricular measurements
Two Three Four Five
Controls (n = 39) 17 8 10 4
Cases (n = 14) 2 2 1 9
Table II Demographic data

Controls Cases
n = 39 n = 14 P value
Baseline GA 24.5 G 5.1 20.2 G 2.7 ! .01*
(wk; mean G SD)
Figure Distribution of patient population by type of repair. GA at delivery 36.6 G 1.8 35.0 G 1.9 ! .01*
(wk; mean G SD)
Fetal gender ! .05y
Males 27 (69%) 5 (36%)
from the analysis. Five patients were diagnosed with Females 12 (31%) 9 (64%)
MMC and followed postoperatively at our institution Level of lesion 1.00y
after intrauterine repair at Vanderbilt University. The Thoracic 4 (10%) 1 (7%)
gestational age of the in utero repair ranged from 21 to Lumbar 28 (72%) 11 (79%)
25 weeks. The number of ventricular measurements Sacral 7 (18%) 2 (14%)
taken throughout the gestation is summarized in Table Ventricular size 14.9 G 6.3 10.3 G 3.3 .02*
I. On average, cases had more frequent prenatal assess- baseline (mm)
ment of ventricular size than control fetuses. Severe 13 (33%) 3 (21%) .51y
ventriculomegaly
Characteristics of the maternal and fetal population
are shown in Table II. The control group who under- * Wilcoxon rank sum test.
y
Fisher exact test.
went postnatal repair presented and delivered at a later
gestational age than fetuses that underwent in utero
repair (P ! .01). Males comprised a greater percentage increasing at a more rapid rate than sacral lesions by
of the control group (P ! .05). There was no difference 0.55 mm/week (0.74 vs 0.19 mm/week; P = .04) (Table
in the distribution of thoracic, lumbar, and sacral lesions III). Females had an increased rate of progression of
between the 2 groups (P = 1.0). However, the mean ventricular size over the males by 0.04 mm/week, but
ventricle size at baseline was 4.6 mm larger in control this difference was not statistically significant (P = .85).
fetuses than cases (P = .02). The overall prevalence of Similarly, the rate of progression was not affected by
mild (100% vs 94%), moderate (36% vs 26%), and baseline ventricle size (P = .94).
severe ventriculomegaly (33% vs 21%), P = .098, was After adjusting for confounders, the fetuses that
similar for in utero and postnatal repair groups. underwent in utero repair did not exhibit a statistically
The overall rate of progression of ventriculomegaly significant increase in ventricular progression over con-
was 0.57 mm/week for fetuses with postnatal repair. trol fetuses (0.27 mm/week; P = .45). The quadratic
However, the rate of progression differed by level of equation using the square of gestational age did not
lesion (thoracic and lumbar lesions 0.70 mm/week; sacral reveal an acute or transient change in ventricular size
lesions 0.08 mm/week [P = .07]). Although this differ- following in utero repair (P = .41).
ence only approaches statistical significance, we felt that
reporting one rate for the entire control group would
overestimate the rate of progression for sacral lesions. Comment
Therefore, for the purpose of the analysis, the level of
lesion was dichotomized into thoracic/lumbar and sacral This study is the first to describe the natural progression
groups. The rate of progression of ventriculomegaly was of ventriculomegaly in fetuses with MMC, and to better
not impacted by gender or baseline ventricular size. define the impact of the baseline degree of ventriculo-
When the case and control fetuses were analyzed megaly, fetal gender, and level of lesion on this progres-
together, the rate of progression of ventriculomegaly sion. When compared with a large group of fetuses with
was 0.62 mm/week. There continued to be a significant MMC undergoing postnatal repair, we found that in
difference in the progression of ventriculomegaly based utero repair did not impact the rate of progression of
on level of lesion, with thoracic and lumbar lesions fetal ventriculomegaly.
730 Adelberg et al
neither study included prenatal ultrasounds of control

Table III Results of multivariate regression for progression
of ventriculomegaly for entire cohort fetuses with MMC for comparison.
Our findings are in direct contrast to a recently
Progression
reported study by Danzer et al, who found an improve-
(mm/wk) 95% CI P value
ment in the cortical index (head circumference/ lateral
Level of lesion 0.55 G 0.22 (0.12, 0.98) .04 ventricular diameter) following in utero surgery for
Gender (females 0.04 G 0.24 (0.42, 0.51) .85 MMC.15 The authors reported an alteration in head
Omales)
growth and a slowing of the progression of ventriculo-
Ventricular size 0.002 G 0.03 (0.06, 0.06) .94
baseline
megaly with a decrease in frequency of severe ventricu-
In utero repair 0.27 G 0.35 (0.42, 0.95) .45 lomegaly. Comparisons were made to a cross-sectional
Gestational age2 0.01 G 0.01 (0.01, 0.02) .41 control group of fetuses without MMC, making con-
clusions difficult to interpret.
Our study found that the ultimate degree of ventri-
culomegaly was not improved with in utero repair. To
In the present study, ventricular size progressed in date, no other study on the effects of in utero repair for
fetuses affected by MMC in a linear fashion with advanc- MMC has compared findings with prenatal information
ing gestational age. This finding confirms that of Babcock gathered from a control group of affected fetuses with-
et al, who noted worsening ventriculomegaly in 17 fetuses out prenatal intervention. An observational study of
with ventricular measurements after 24 weeks’ gesta- serial magnetic resonance imaging (MRI) of fetal brains
tion.10 However, Babcock’s study was limited by the in 10 cases of in utero repair has shown either an arrest
small number of patients with serial ultrasounds through- or slowing of ventricular size as soon as 3 weeks after in
out gestation. Only 4 fetuses had serial scans, with 3 of 4 utero surgery.5 We did not find an acute or transient
showing worsening ventriculomegaly in utero. effect on ventricular size following in utero repair for
We found that thoracic and lumbar lesions were MMC, in contrast to this, and previously reported
associated with a more rapid increase in ventricular size worsening in ventriculomegaly following postnatal
than sacral lesions. Because of the small sample size in MMC repair.
our study, we were unable to perform a more detailed There are several limitations in this study. Although a
analysis of specific vertebral levels of MMC. A previous standardized and accepted method of measuring the
study of 33 fetuses with MMC confirmed our findings of lateral ventricle was used, the results would be strength-
an association between level of lesion and degree of ened if a single observer had performed all of the mea-
hydrocephalus at birth.11 Tulipan and Rintoul indepen- surements. Because of the retrospective study design,
dently found that higher lesions were more likely to there was not a standardized interval for the sono-
require placement of neonatal ventriculoperitoneal graphic assessments of ventricle size. Statistical methods
shunts, suggesting either a greater baseline level of were applied to account for this limitation, yet the
ventriculomegaly and/or more rapid progression irregular timing of these measurements may be respon-
in utero.7,12 In contrast to these findings, previous sible for the difference in findings from previous studies.
investigators reviewing a series of prenatal sonograms The number of cases of in utero repair was small,
in MMC fetuses found that when one controls for limiting our ability to analyze this group alone. Given
gestational age and degree of posterior fossa deformity, the number of confounding factors that were controlled
the level of lesion did not predict ventricular diameter.13 for in the logistic regression analysis, we could not
This disparity with our findings may be explained by the report, with statistical confidence, results of this group
fact that we did not assess or control for the degree of alone. An alternative approach was chosen using cases
hindbrain herniation in the present investigation. and controls in the same model. First, the type of repair
Previous studies have shown that in utero repair of was included as a variable in the logistic regression,
MMC decreases the incidence of shunt-dependent hy- allowing a direct comparison between the groups. Sec-
drocephalus. The initial publication from Bruner et al ond, the rate of ventricular progression was compared
demonstrated a decrease in shunt placement after in between the control group and that of the entire cohort.
utero repair; however, their study did not evaluate the Neither analysis showed an impact of in utero repair on
baseline degree of ventriculomegaly and or the impact of the progression of ventriculomegaly.
level of lesion.6 A subsequent publication from the same Our data suggest that the decrease in the need for
group showed that higher lesions, later gestational age ventriculoperitoneal shunting in the first year of life
at the time of repair, and larger baseline ventricle size after in utero repair of MMC may not be caused by a
are predictive of the need for neonatal ventriculoper- slowing in the progression of ventriculomegaly in utero.
itoneal shunt placement.14 Tulipan also showed that An explanation for decreased rates of ventriculoperito-
fetuses with larger ventricles at baseline and higher level neal shunting requires further exploration. The long-
lesions were more likely to require shunting.7 However, term implications of in utero repair on the developing
Adelberg et al 731
brain are yet to be elucidated. The completion of the the incidence of shunt-dependent hydrocephalus. Pediatr Neuro-
NICHD-sponsored randomized clinical trial will hope- surg 2003;38:27-33.
8. McLone DG, Dias MS. The Chiari II malformation: cause and
fully answer such questions. impact. Childs Nerv Syst 2003;19:540-50.
9. Cardoza JD, Goldstein RB, Filly RA. Exclusion of fetal ventricu-
lomegaly with a single measurement: the width of the lateral
References ventricular atrium. Radiology 1988;169:711-4.
10. Babcock CJ, Goldstein RB, Barth RA, Damato NM, Callen PW,
1. Centers for Disease Control and Prevention. Spina bifida and Filly RA. Prevalence of ventriculomegaly in association with
anencephaly before and after folic acid mandate–United States, myelomeningocele: correlation with gestational age and severity
1995-1996 and 1999-2000. MMWR Morb Mortal Wkly Rep 2004; of posterior fossa deformity. Radiology 1994;190:703-7.
53:362-5. 11. Biggio JR, Owen J, Wenstrom KD, Oakes WJ. Can prenatal
2. Centers for Disease Control and Prevention. Economic burden of ultrasound findings predict ambulatory status in fetuses with open
spina bifida–United States, 1980-1990. MMWR Morb Mortal spina bifida? Am J Obstet Gynecol 2001;185:1016-20.
Wkly Rep 1989;38:264-7. 12. Rintoul NE, Sutton LN, Hubbard AM, Cohen B, Melchionni J,
3. Tulipan N, Bruner JP. Fetal surgery for spina bifida. Lancet 1999; Pasquariello PS, et al. A new look at myelomeningoceles: func-
353:406-7. tional level, vertebral level, shunting, and the implications for fetal
4. Tulipan N, Hernanz-Schulman M, Lowe LH, Bruner JP. Intra- intervention. Pediatrics 2002;109:409-13.
uterine myelomeningocele repair reverses preexisting hindbrain 13. Babcock CJ, Drake CM, Goldstein RB. Spinal level of fetal
herniation. Pediatr Neurosurg 1999;31:137-42. myelomeningocele: does it influence ventricular size? AJR 1997;
5. Sutton LN, Adzick NS, Bilaniuk LT, Johnson MP, Crombleholme 169:207-10.
TM, Flake AW. Improvement in hindbrain herniation demon- 14. Bruner JP, Tulipan N, Reed G, Davis GH, Bennett K, Luker KS,
strated by serial fetal magnetic resonance imaging following fetal et al. Intrauterine repair of spina bifida: preoperative predictors
surgery for myelomeningocele. JAMA 1999;282:1826-31. of shunt-dependent hydrocephalus. Am J Obstet Gynecol 2004;
6. Bruner JP, Tulipan N, Paschall RL, Boehm FH, Walsh WF, Silva 190:1305-12.
SR, et al. Fetal surgery for myelomeningocele and the incidence 15. Danzer E, Johnson MP, Wilson RD, Flake AW, Hedrick HL,
of shunt-dependent hydrocephalus. JAMA 1999;282:1819-25. Sutton LN, et al. Fetal brain growth and head biometry following
7. Tulipan N, Sutton LN, Bruner JP, Cohen BM, Johnson MP, in utero repair of myelomeningocele. Ultrasound Obstet Gynecol
Adzick NS. The effect of intrauterine myelomeningocele repair on 2004;24:606-11.
American Journal of Obstetrics and Gynecology (2005) 193, 732-8
www.ajog.org
Isolated fetal pyelectasis and chromosomal abnormalities

Claudio Coco, MD,a,b Philippe Jeanty, MD, PhDa
Department of Ultrasound, Women’s Health Alliance, Nashville, TN a; Maternal Fetal Medicine Center, Artemisa
Groups, Rome, Italyb
Received for publication July 29, 2003; revised January 19, 2005; accepted February 9, 2005
KEY WORDS Objective: The primary objective of this study was to determine if isolated pyelectasis is a risk
Aneuploidy factor for trisomy 21.
Trisomy Study design: Twelve thousand, six hundred and seventy-two unselected singleton fetuses were
Low-risk population examined by prenatal ultrasound during the second trimester at a single institution. The
Pyelectasis sensitivity, specificity, positive predictive value, negative predictive value, and likelihood ratio of
Dilated renal pelvis pyelectasis (either isolated or in association with other soft markers/structural anomalies) to
Soft markers detect trisomy 21 were calculated.
Prenatal diagnosis Results: Pyelectasis (anteroposterior pelvic diameter R4 mm) was detected in 2.9% (366/12,672)
Fetal ultrasound of the fetuses. Among these, 83.3% (305/366) were isolated, and 16.7% (61/366) were associated
with other markers/structural anomalies. The prevalence of trisomy 21 was 0.087% (11/12,672)
and, among these fetuses, 2 (18.1%) had pyelectasis, 1 isolated, and 1 associated with other
markers/structural anomalies. The presence of isolated pyelectasis had 9.09% sensitivity, 97.6%
specificity, 0.33% positive predictive value, and 99.9% negative predictive value to detect fetuses
with trisomy 21. The likelihood ratio of trisomy 21 in this group of fetuses was 3.79 (95% CI
0.582–24.616). Among fetuses with pyelectasis and other associated markers/structural anom-
alies, the sensitivity, specificity, positive predictive value, negative predictive value, and likelihood
ratio for trisomy 21 were 9.09%, 99.5%, 1.64%, 99.9%, and 19.2 (95% CI 2.91–126.44).
Conclusion: In the absence of other findings, isolated pyelectasis is not a justification for the
performance of an amniocentesis.
Pyelectasis, which is a dilatation of the renal pelvis investigators.4,5 Persutte et al6 reported that repeated
visible with ultrasound, is an anatomic variant that measurements of the renal pelvis performed within
rarely has pathologic significance for fetal and postnatal a period of 2 hours present high variability, and that
renal function. Generally, the renal pelvis is collapsed this observation was independent from the maternal
and, thus, undetectable with ultrasound. hydration status. A possible genetic predisposition to
The renal pelvis diameter may be affected by maternal pyelectasis in consecutive pregnancies has been proposed
hydration,1-3 although this view is not shared by all by Degani et al.7
In 1990, Benacerraf et al8 were first to suggest an
association between pyelectasis and aneuploidy. They
Reprints not available from the authors. Address correspondence to
Claudio Coco, MD, Artemisa Groups, Viale Liegi 45, 00198 Rome,
estimated a 3.3% risk for Down syndrome when pye-
Italy. lectasis was present. In this study, we investigated the
E-mail: cl.coco@vodafone.it prevalence and possible association of pyelectasis with
doi:10.1016/j.ajog.2005.02.074
Coco and Jeanty 733
All patients underwent a thorough ultrasound exam-

ination which, aside from the AIUM-ACR guidelines,
sought as many soft markers as possible, including
nuchal thickening, pyelectasis, echogenic intracardiac
foci, brachymesophalangia of the fifth digit, as well as a
simian crease, whenever possible, for these more subtle
markers were only looked at for a few seconds and, if
not seen, simply passed over. In only a small percentage
of fetuses did we identify this sign in accord with
previous work.13
Hands, feet, and limb segments were included. Outflow
tracts, return of pulmonary veins to the left atrium, and,
when possible, both arches were also part of the routine.
Examinations were performed using either Acuson
XP 128 (Siemens Medical Systems, Mountain View,
Calif), Sequoia (Siemens Medical Systems), or Voluson
730 (General Electric Medical Systems, Kretztechnik,
Figure 1 Measurement of the renal pelvis in the anteropos- Zipf, Austria) scanners.
terior dimension in a fetus with bilateral pyelectasis of 4.6 mm. The results only include data from 2D examinations,
not from 3D.
aneuploidy in an unselected population of patients. We The aim of this study was to determine if isolated
determined whether the finding of fetal pyelectasis in the pyelectasis is a risk factor for trisomy 21. Because of the
second trimester justified alteration in patient manage- compounding risk in the presence of other markers, we
ment, in particular, whether or not patients should be analyzed the population with isolated pyelectasis and the
subjected to a karyotype. population with other markers separately. There is a
Various criteria have been used to define fetal renal general consensus that pyelectasis is only a marker for
pyelectasis. For example, Benacerraf et al defined pye- trisomy 21 and not for other aneuploidy, and thus, our
lectasis as a renal pelvis anteroposterior diameter greater analysis only considered cases of trisomy 21. We also
or equal to 4 mm in fetuses between 15 and 20 weeks; an reported the total cases of aneuploidies recognized in the
anteroposterior renal pelvis diameter greater than or study period and the presence of pyelectasis in these cases.
equal to 5 mm in fetuses between 20 and 30 weeks; and a The sensitivity, specificity, positive predictive value,
renal pelvis anteroposterior diameter greater or equal to negative predictive value, and likelihood ratios with
7 mm in fetuses between 30 and 40 weeks. Other authors 95% CIs of pyelectasis to detect trisomy 21 were
used a cutoff of 4 mm to define pyelectasis.7,9-12 In our calculated. Separate analyses were conducted for fetuses
study, the criterion used to select fetuses with pyelectasis with isolated pyelectasis, and for those with pyelectasis
was an anteroposterior renal pelvis diameter of 4 mm or and other associated markers/structural anomalies. The
more. following markers/structural anomalies were included:
choroid plexus cysts, echogenic heart focus, 2-vessel
cord, nuchal thickening, heart defects, diaphragmatic
Material and methods hernia, esophageal atresia, omphalocele, facial cleft,
micrognathia, myelomeningocele, growth restriction,
In contrast to most published series, our patients came shortening of the limbs, radial aplasia, (for long bones
from a homogeneous base low-risk population from our we routinely measured humerus and femur, and we
practice, which serves the needs of a group of about 30 looked at ulna-radius and tibia-fibula, but only mea-
obstetricians. Patients referred by physicians outside our sured them if they appeared abnormal), overlapping
ultrasound practice were not included in the study fingers, talipes, rocker bottom feet, clinodactyly, and
because they were more likely to have been referred brachymesophalangia of the fifth digit.
for a suspected anomaly and, therefore, may not have Likelihood ratios were also calculated by the for-
represented a low-risk population. mula: Likelihood ratio C = sensitivity /(1 Specificity);
From the 16,272 midgestation patients referred to our likelihood ratio = (1-Sensitivity)/Specificity.
center from January 1998 to December 2002, 12,672 The likelihood ratio expresses the odds that pyelec-
patients between 16 and 23 gestational weeks were tasis occurs in fetuses with Down syndrome versus the
included in the study. The other patients were removed odds that pyelectasis occurs in fetuses without Down
because their initial examinations do not fall within the syndrome (Figure 1).
16 to 23 weeks frame. None of the patients had a first Follow-up information in all the cases was obtained
trimester aneuploidy screening. by amniocentesis, infant postnatal reports, or by
734 Coco and Jeanty
Figure 2 The distribution of pyelectasis with and without trisomy 21 in the patients examined.
Table I The prevalence, sensitivity, specificity, positive Table II The prevalence, sensitivity, specificity, positive
predictive value, negative predictive value, the likelihood predictive value, negative predictive value, likelihood ratio,
ratio, and the diagnostic odds ratio of isolated pyelectasis to and diagnostic odds ratio of pyelectasis in association with
detect trisomy 21 other anomalies to detect trisomy 21
Total cases 12,672 Total cases 12,672
Trisomy 21 Normal Total Trisomy 21 Normal Total
Isolated pyelectasis 1 304 305 Pyelectasis 1 60 61
Absence of isolated 10 12,357 12,367 Absence of pyelectasis 10 12,601 12,611
pyelectasis Total 11 12,661
Total 11 12,661 Prevalence: 0.09%; sensitivity: 9.09% (95% CI 1.62–37.74); specific-
Prevalence: 0.09%; sensitivity: 9.09% (95% CI 1.62–37.74); specificity: 99.53% (95% CI 99.39–99.63); PPV: 1.64%; NPV: 99.92%; CLR:
ity: 97.6% (95% CI 97.32–97.85); PPV: 0.33%; NPV: 99.9%; CLR: 19.1833 (95% CI 2.91–126.44); LR: 0.9134 (95% CI 0.758–1.101);
3.7862 (95% CI 0.582–24,616); LR: 0.9315 (95% CI 0.773–1123); diagnostic odds ratio: 21,002 (CI 2647–166,637).
Diagnostic odds ratio: 4065 (95% CI 0.519–31,853).
contacting the referring physician or pediatrician. If The sex of the affected fetus was identified in 315
these options failed, the mother was interviewed. The cases (in the other cases, the parents requested not to
follow-up was obtained in every case of pyelectasis. know the sex of the fetuses), with 206 (65%) fetuses
being male and 109 (35%) being female. This indicated
Results an increased prevalence of pyelectasis among males,
with a male to female ratio of 1.9:1.
The mean maternal age was 27.2 G (standard deviation The prevalence of aneuploidy in the study population
5.5), ranging from 15 to 42 years old. The prevalence of was 0.19% (24/12,672). Eleven fetuses had Down syn-
pyelectasis in our population was 2.9% (366/12,672). In drome, including a 14-21 translocation, 6 had trisomy 18,
57% (208/366) of cases, the anteroposterior pelvis diam- 2 trisomy 13, and 5 had miscellaneous karyotypic anom-
eter measured 4 to 5 mm, in 23% (85/366) the renal pelvis alies including monosomy X, trisomy X, triploidy, and
measured 5 to 6 mm, in 13% (48/366) the dilatation was 6 ring 14 chromosomes. Among cases of aneuploidy, pye-
to 7 mm, in 5% (19/366) it was 7 to 8 mm, and in 1% (4/ lectasis was present in only 2 fetuses, both with trisomy
366) the renal pelvis measured 8 mm, 1 fetus had 12 mm 21: in 1 case pyelectasis was associated with other anom-
dilatation, and 1 had 16 mm dilatation. alies and, in the other, pyelectasis was an isolated finding.
Coco and Jeanty 735
Table III Details of the 11 cases of trisomy 21 examined in our patient population and their association with pyelectasis or other
anomalies
Karyotype Pyelectasis Other findings Maternal age
1 Trisomy 21 Bilateral Heart echogenic focus, sandal gap on the right foot 23
2 Trisomy 21 None Heart echogenic focus, hydrocephalus, brachycephaly, 30
hypoplasia of the fifth digit, 2-vessel cord
3 Trisomy 21 None 2-vessel cord 32
4 Trisomy 21 None None 39
5 Trisomy 21 None Omphalocele, mild ventriculomegaly, microphthalmia, 33
micrognathia, short femur
6 Trisomy 21 None Abdominal calcification 34
9 Trisomy 21 None Echogenic heart focus 35
11 Translocation 14-21 Bilateral None 23
In the other aneuploidy encountered (the 6 cases of 366, or 0.546%) compared with fetuses without pyelec-
trisomy 18, the 2 trisomy 13, and 5 miscellaneous tasis (9 of 12,306, or 0.0731%).
karyotypic anomalies), none had pyelectasis and were
considered ‘‘normal’’ for statistical analysis. Comment
Table I shows the prevalence, sensitivity, specificity,
positive predictive value, negative predictive value, the This is one of the first large studies examining the
likelihood ratio, and the diagnostic odds ratio of prevalence of pyelectasis in a low-risk unselected pop-
isolated pyelectasis to detect trisomy 21. ulation. We found that 2.9% of the 12,672 fetuses
Table II shows the prevalence, sensitivity, specificity, examined had pyelectasis: 83.3% of these were isolated
positive predictive value, negative predictive value, like- cases, whereas in 16.7% other anomalies were present.
lihood ratio, and diagnostic odds ratio of pyelectasis in Only 2 cases of pyelectasis were associated with aneu-
association with other anomalies to detect trisomy 21. ploidies (0.55%), one of which had isolated pyelectasis
Of the total patients with pyelectasis, 83.3% (305/ and Down syndrome, while the other had pyelectasis
366) were isolated cases with no other associated with associated anomalies and trisomy 21.
anomalies, which was 2.41% (305/12,672) of the total In the present study, only 1 case out of 305 fetuses
population. In one case of isolated pyelectasis with a with isolated pyelectasis had trisomy 21; this prevalence
renal pelvis dilatation of 4 mm in both kidneys, the was higher than the expected 1 in 1100 in the general
female fetus was affected with translocation 14-21. This population reported from the Centers for Disease Con-
patient was 23 years old at the examination, 24 years old trol and Prevention.14 However, although the likelihood
at the time of delivery, and the ultrasound was ratio for fetuses trisomy 21 with isolated pyelectasis was
performed at 18 weeks gestation. The prevalence of 3.79, which appeared to be increased, if the 95% CI is
trisomy 21 in the group with isolated pyelectasis was calculated (0.582–24.616), the estimated risk was actu-
1:305, or 0.33%. In the other patients, no other chro- ally not significantly increased.
mosomal anomalies were found. The male prevalence found in our work (male:female
Of the total patients with pyelectasis, 16.7% (61/366) 1.9:1) confirmed previous reports.15-18 Comparing our
had other associated anomalies, which was 0.48% (61/ work with previous studies was rather difficult because
12,672) of the total population. Most of these cases were of the various standards and parameters used. Different
minor anomalies, such as heart echogenic focus, choroid studies have defined pyelectasis with a range of dilata-
plexus cyst, 2-vessel cords, and short humerus. Trisomy tion sizes, and used populations of different composi-
21 was identified in a case of bilateral pyelectasis with a tions (high-risk, low-risk, or a mix of both). Gestational
right dilatation of 4 mm and left dilatation of 6 mm. The age and patient age also varied from one study to
other minor signs present in this fetus included a cardiac another. In some cases, authors did not differentiate
echogenic focus in the left ventricle and a sandal gap on between isolated pyelectasis and pyelectasis with other
the right foot. The prevalence of trisomy 21 in fetuses associated anomalies.
with pyelectasis and associated anomalies was 1:61, or The prevalence of pyelectasis in our patients was
1.64% (Figure 2 and Table III). 2.9%, which was similar to values calculated in other
In the present study, trisomy 21 was found nearly 8 studies (see Table IV), (range between 0.72% and
times more often in association with pyelectasis (2 of 2.84%). Although the true prevalence of pyelectasis in
736 Coco and Jeanty
Table IV Comparison of several studies on pylelectasis
Prevalence
of Gestational Pyelec.
Author pyelectasis Population age Cut Off N %
Benacerraf (8) 2.84% 7400 O16 R4 mm between 15-20 weeks 210 2.84%
R5 mm between 20-30 weeks
Corteville (12) 2.10% 5944 O14 R4 mm before 33 weeks 127 2.14%
R7 mm after 33 weeks
Nicolaides (18) NA Selected 15-38 R5 mm 258 d
Wickstrom (10) 0.80% 11340 17-39 R4 mm 82 0.72%

Wickstrom (25) 0.72% 7481 R15 R4 mm !33 weeks 121 1.62%
R7 mm O33 weeks
Ouzounian et al. (28) d d R4 mm 84* d
Chudleigh et al. (19) 0.73% 101600 16-26 R5 mm 737 0.73%
Havutcu et al. (16) 1.25% 25586 18-24 R5 mm 320 1.25%

Persutte et al. (15) 5.50% 5529 16-38 R4 mm 306 5.53%
Nyberg (23) NA Selected 14-20 R3 mm 186

Rotmensch (29) NA Selected 14-28 R4 mm between 15-20 weeks
Sairam (30) 2.34%
Current study 2.90% 12672 16-23 R4 mm 366 2.89%
* No distintion there were among isolated and not isolated pyelectasis.
y
The distinction was made only for major anomaly: 12 had concomitant anomaly: 2 T21, 1 recombinant 8 Syndrome, 3 multiple congenital anomaly,
1 IUGR, 5 fetal or infant losses.
z
Analisys of 155 fetuses with Down syndrome.
x
Report of 187 cases of trisomy 21.
the general population was difficult to ascertain, this were in association with ultrasound abnormalities and,
study provided evidence that is as close to the baseline as in the remaining case of trisomy 21, the abnormality was
possible because none of the patients were preselected or suspected. Previous studies did not find any aneuploidy
referred from other centers. A summary of the previous for isolated fetal pyelectasis.16 Others20,21 concluded
study is provided in Table IV. that the presence of pyelectasis increased the risk of
Chudleigh19 used the percentage of pyelectasis to the aneuploidy only if it was in association with other
number of births, not among the total ultrasound exam, anomalies, such as nuchal fold thickening and a short
and different parameter inclusions (anteroposterior di- humerus.
ameter of the renal pelvis of 5 mm up to and including Although Persutte et al15 did not specify if his cases
10 mm, while we used 4 mm). were associated with other structural anomalies or other
Though there was variability in the prevalence of soft signs, he found a high incidence of chromosomal
pyelectasis in past studies (Table IV), there was a relative anomalies (2 trisomy 21 and 1 recombinant 8 syndrome)
uniformity of result for the prevalence of trisomy 21 in his 306 cases and, in addition, a high incidence of
among fetuses with pyelectasis. Because of the small concomitant anomalies (9 cases). In the study of
prevalence of trisomy 21, 1 or 2 cases could have strongly Nicolaides et al,18 the suggested risk for isolated pye-
affected the statistical analysis in most studies. lectasis was an increased risk for all aneuploidy.
In our study the prevalence of trisomy 21 in isolated Different likehood ratios were reported in another
pyelectasis was 1:305; Chudleigh19 reported similar study. In his ‘‘age adjusted ultrasound risk assessment’’
results (1:217, and 1:324 in women under 36 years) for (AAURA) for Down syndrome, Nyberg22 used a like-
isolated pyelectasis, but suggested a higher incidence for lihood ratio for renal pyelectasis of 1.6, and in his other
pyelectasis in association with other anomalies. In the study,23 an isolated pyelectasis within 14 and 20 weeks
study of Corteville,12 among 4 cases of trisomy 21, 3 had a likelihood ratio of 1.5 (95% CI 0.6–3.6) as an
Coco and Jeanty 737
Table IV (Continued)
Non isolated
Pyelectasis pyelectasis Isolated pyelectasis Pyelec. whit other anomalies
N % N % Normal Anormal Normal Anormal
203* 7 Trisomy 21
d d d d d 1 Trisomy 21 6 (3 Trisomy 21; 3 Other)
163 63.18% 95 36.82% 158 1 Trisomy 21 65 30 Trisomies 21, 13, 18

4 other other
d d d d 79 3 Trisomy 21 d d
All isolated 119 1 Trisomy 21
1 mosaic 46, XY/XYY
d 1 Trisomy 21
651 88.33% 98 13.30% 648 3 Trisomy 21 89 3 Trisomy 21
1 Trisomy 13
1 Trisomy 8
2 45, X0
2 Other
301 94.06% 19 5.94% 301 None 19 None
294y 96.08% d d 294y None dy 2 Trisomy 21
1 recombinant 8 Syndrome
z z
5 Trisomy 21 16 trisomy 21
x
d None 13 Trisomy 21
305 83.33% 61 16.67% 304 1 Traslocation 14-21 60 1 Trisomy 21

* No distintion there were among isolated and not isolated pyelectasis.
y
The distinction was made only for major anomaly: 12 had concomitant anomaly: 2 T21, 1 recombinant 8 Syndrome, 3 multiple congenital anomaly,
1 IUGR, 5 fetal or infant losses.
z
Analisys of 155 fetuses with Down syndrome.
x
Report of 187 cases of trisomy 21.
isolated finding and 5.2 overall if in combination with Although the observed 3.79-fold appeared as an
other markers. The reported risk was higher in the study increase, the CIs were so large (to the point of including
of Whitlow24 in the first trimester by a likelihood ratio a decreased risk) that this study does not support an
of 8 (95% CI 2.0–31.7) for all aneuploidy and 9.6 (95% increased risk. We concluded that the recognition of an
CI 1.4–64.7) for trisomy 21, although it was not clearly isolated pyelectasis is not an indication for amniocente-
specified if this risk was only for isolated pyelectasis or sis, but an indication for a detailed exam to assess the
in association with other soft markers or structural presence of other ultrasound markers for aneuploidy.
anomalies. The analysis made on 99 women by Wick- For fetuses with multiple markers (only 1 in our study,
strom et al25 indicated an increased risk for isolated fetal with an echogenic heart focus and a right-sided sandal
pyelectasis over that related to age for both Down gap) our likelihood ratio demonstrated an increased risk
syndrome and all chromosomal abnormalities: 3.9-fold for aneuploidy.
increase in Down syndrome risk, and a 3.3-fold increase
in risk for all chromosomal abnormalities.
Two different studies showed results very similar to
ours among the total cases of aneuploidy and associa-
References
tion with pyelectasis.26,27 1. Babcook CJ, Silvera M, Drake C, Levine D. Effect of maternal
Table IV summarizes the comparison with other hydration on mild fetal pyelectasis Department of Radiology, The
previous study for number of population exanimate University of Utah Medical Center, Salt Lake City. J Ultrasound
and different inclusion criteria. Med 1998;17:539-44;quiz 545-6.
2. Robinson JN, Tice K, Kolm P, Abuhamad AZ. Effect of maternal
Our data do not suggest an increased risk of aneu- hydration on fetal renal pyelectasis. Department of Obstetrics and
ploidy for isolated pyelectasis compared with the general Gynecology, Eastern Virginia Medical School, Norfolk, VA.
population. Obstet Gynecol 1998 Jul;92:137-41.
738 Coco and Jeanty
3. Graif M, Kessler A, Hart S, Daitzchman M, Mashiach S, Boichis 17. Petrikovsky BM, Nardi DA, Rodis JF, Hoegsberg B. Elevated
H, et al. Renal pyelectasis in pregnancy: correlative evaluation of maternal serum alpha-fetoprotein and mild fetal uropathy. Obstet
fetal and maternal collecting systems. Department of Diagnostic Gynecol 1991;78:262-4.
Imaging, Chaim Sheba Medical Center, Tel Hashomer, Israel. Am 18. Nicolaides KH, Cheng HH, Abbas A, Snijders RJ, Gosden C. Fetal
J Obstet Gynecol 1992;167:1304-6. renal defects: associated malformations and chromosomal defects.
4. Hoddick WK, Filly RA, Mahony BS, Callen PW. Minimal fetal Harris Birthright Research Centre for Fetal Medicine, King’s
renal pyelectasis. J Ultrasound Med 1985;4:85-9. College Hospital, London, UK. Fetal Diagn Ther 1992;7:1-11.
5. Allen KS, Arger PH, Mennuti M, Coleman BG, Mintz MC, 19. Chudleigh PM, Chitty LS, Pembrey M, Campbell S. The associ-
Fishman M. Effects of maternal hydration on fetal renal pyelec- ation of aneuploidy and mild fetal pyelectasis in an unselected
tasis. Radiology 1987;163:807-9. population: the results of a multicenter study. Ultrasound Obstet
6. Persutte WH, Hussey M, Chyu J, Hobbins JC. Striking findings Gynecol 2001;17:197-202.
concerning the variability in the measurement of the fetal renal 20. Vintzileos AM, Egan JF. Adjusting the risk for trisomy 21 on the
collecting system. Ultrasound Obstet Gynecol 2000;15:186-90. basis of second-trimester ultrasonography. Am J Obstet Gynecol
7. Degani S, Leibovitz Z, Shapiro I, Gonen R, Ohel G. Department 1995;172:837-44.
of Obstetrics and Gynecology, Bnai-Zion Medical Center, Haifa, 21. Vintzileos AM, Campbell WA, Guzman ER, Smulian JC, McLean
Israel. Fetal pyelectasis in consecutive pregnancies: a possible DA, Ananth CV. Second-trimester ultrasound markers for detec-
genetic predisposition. Ultrasound Obstet Gynecol 1997;10:19-21. tion of trisomy 21: which markers are best? Obstet Gynecol
8. Benacerraf BR, Mandell J, Estroff JA, Harlow BL, Frigoletto FD 1997;89:941-4.
Jr. Fetal pyelectasis: a possible association with Down syndrome. 22. Nyberg DA, Luthy DA, Resta RG, Nyberg BC, Williams MA.
Department of Obstetrics and Gynecology, Brigham & Women’s Age-adjusted ultrasound risk assessment for fetal Down’s syn-
Hospital, Boston, Massachusetts. Obstet Gynecol 1990;76:58-60. drome during the second trimester: description of the method
9. Gotoh H, Masuzaki H, Fukuda H, Yoshimura S, Ishimaru T. and analysis of 142 cases. Ultrasound Obstet Gynecol 1998;
Detection and assessment of pyelectasis in the fetus: relationship to 12:8-14.
postnatal renal function Department of Obstetrics and Gynecol- 23. Nyberg DA, Souter VL, El-Bastawissi A, Young S, Luthhardt F,
ogy, Nagasaki University School of Medicine, Japan. Obstet Luthy DA. Isolated sonographic markers for detection of fetal
Gynecol 1998;92:226-31. Down syndrome in the second trimester of pregnancy. J Ultra-
10. Wickstrom E, Maizels M, Sabbagha RE, Tamura RK, Cohen LC, sound Med 2001;20:1053-63.
Pergament E. Isolated fetal pyelectasis: assessment of risk for 24. Whitlow BJ, Lazanakis ML, Kadir RA, Chatzipapas I, Econo-
postnatal uropathy and Down syndrome. Ultrasound Obstet mides DL. The significance of choroid plexus cysts, echogenic
Gynecol 1996;8:236-40. heart foci and renal pyelectasis in the first trimester. Ultrasound
11. Adra AM, Mejides AA, Dennaoui MS, Beydoun SN. Fetal Obstet Gynecol 1998;12:385-90.
pyelectasis: is it always ‘‘physiologic?’’ Department of Obstetrics 25. Wickstrom EA, Thangavelu M, Parilla BV, Tamura RK, Sabba-
and Gynecology, University of Miami School of Medicine, gha RE. A prospective study of the association between isolated
Florida. Am J Obstet Gynecol 1995;173:1263-6. fetal pyelectasis and chromosomal abnormality. Obstet Gynecol
12. Corteville JE, Dicke JM, Crane JP. Fetal pyelectasis and Down 1996;88:379-82.
syndrome: is genetic amniocentesis warranted? Obstet Gynecol 26. Salihu HM, Boos R, Schmidt W. Antenatally detectable markers
1992;79:770-2. for the diagnosis of autosomally trisomic fetuses in at-risk preg-
13. Jeanty P. Prenatal detection of simian crease. J Ultrasound Med nancies. Am J Perinatol 1997;14:257-61.
1990;9:131-6. 27. Vergani P, Locatelli A, Piccoli MG, Ceruti P, Mariani E, Pezzullo
14. Centers for Disease Control and Prevention. Down syndrome JC, et al. Best second trimester sonographic markers for the
prevalence at birthdUnited States, 1983-1990. Accessed August detection of trisomy 21. J Ultrasound Med 1999;18:469-73.
26, 1994. Available at: http://www.cdc.gov/mmwr/preview/ 28. Ouzounian JG, Castro MA, Fresquez M, al-Sulyman OM, Kovacs
mmwrhtml/00032401.htm. BW. Prognostic significance of antenatally detected fetal pyelecta-
15. Persutte WH, Koyle M, Lenke RR, Klas J, Ryan C, Hobbins JC. sis. Ultrasound Obstet Gynecol 1996;7:424-8.
Mild pyelectasis ascertained with prenatal ultrasonography is 29. Rotmensch S, Liberati M, Bronshtein M, Schoenfeld-Dimaio M,
pediatrically significant. Ultrasound Obstet Gynecol 1997;10:12-8. Shalev J, Ben-Rafael Z, et al. Prenatal sonographic findings in
16. Havutcu AE, Nikolopoulos G, Adinkra P, Lamont RF. The 187 fetuses with Down syndrome. Prenat Diagn 1997;17:1001-9.
association between fetal pyelectasis on second trimester ultra- 30. Sairam S, Al-Habib A, Sasson S, Thilaganathan B. Natural history
sound scan and aneuploidy among 25,586 low risk unselected of fetal hydronephrosis diagnosed on mid-trimester ultrasound.
women. Prenat Diagn 2002;22:1201-6. Ultrasound Obstet Gynecol 2001;17:191-6.
www.ajog.org
Endometrial microbial colonization and plasma cell

endometritis after spontaneous or indicated preterm
versus term delivery
William W. Andrews, PhD, MD,a Robert L. Goldenberg, MD,a
John C. Hauth, MD,a Suzanne P. Cliver, BA,a Michael Conner, MD,b
Alice R. Goepfert, MDa
Departments of Obstetrics and Gynecologya and Pathology,b Center for Research in Women’s Health,
University of Alabama at Birmingham, Birmingham, AL
Received for publication October 19, 2004; accepted February 7, 2005
KEY WORDS Objective: This study was undertaken to determine whether endometrial microbial colonization
Preterm birth or plasma cell endometritis is increased after spontaneous versus indicated preterm delivery or a
Prematurity spontaneous term delivery.
Endometritis Study design: Postpartum, endometrial specimens were obtained after a spontaneous (mean 83,
Infection G 17.6 days) or indicated (mean 83, G 16.7 days) preterm delivery before 34 weeks’ gestation and
after a spontaneous term delivery (mean 82, G 15.7 days; P = .980). Cultures for aerobic and
anaerobic bacteria, Trichomonas vaginalis, and genital mycoplasmas were performed. Histologic
endometritis was defined as the presence of plasma cells.
Results: The study population (n = 820) was 71% black, 29% white, 69% unmarried, and 31%
had less than 12 years of education. Endometrial cultures were positive for at least
1 microorganism in 82% of the women. No significant difference in positive endometrial
cultures were observed among women after a spontaneous versus an indicated preterm delivery
(85% vs 79%, P = .102), or a spontaneous preterm versus a spontaneous term delivery (85% vs
81%, P = .123). Plasma cell endometritis was present in 39% of 506 specimens sufficient for
histologic examination and was also similar in the three groups (P = .160).
Conclusion: Microbial colonization of the endometrium and plasma cell endometritis are similar
3 months after spontaneous or indicated preterm or term births. Therefore, chronic infection and
inflammation of the endometrium (documented at 3 months postpartum) do not appear to be risk
factors for subsequent delivery in women with a prior spontaneous delivery less than 34 weeks’
gestation.
Supported by a grant from the National Institute of Child Health and Human Development (HD33883).
Reprints not available from the authors.
doi:10.1016/j.ajog.2005.02.128
740 Andrews et al
Preterm birth complicates up to 12% of all pregnan- with a recent indicated preterm or spontaneous term
cies and remains the primary cause of perinatal mortality delivery. The objective of the current study was to test this
and approximately one half of long-term neurologic hypothesis.
morbidity.1,2 The majority of this mortality and mor-
bidity is concentrated among very early preterm births.1,2 Material and methods
Considerable data implicate a clinically silent upper
genital tract infection and histologic chorioamnionitis as The study was conducted at the University of Alabama
key components of the pathophysiology of a majority of at Birmingham between August 1995 and August 2001.
very early spontaneous preterm births. The bacteria iso- Women with singleton pregnancies who presented with
lated from the upper genital tract among women who spontaneous labor or preterm premature rupture of
have preterm births are all microorganisms that are membranes that ultimately resulted in a preterm birth or
generally believed to be of low virulence and are pregnancy loss between 16 weeks 0 days’ and 33 weeks 6
typically representative of the normal microbial flora days’ gestation were recruited and offered enrollment by
of the cervix and vagina.3 research personnel. These women represented the spon-
The prevailing hypothesis is that microbial coloniza- taneous preterm birth group. After discharge from the
tion of the chorioamnion and amniotic fluid results from hospital, these women received follow-up care at a clinic
ascending infection by bacteria initially localized in the specifically dedicated to this study. Also recruited were
cervix and vagina.3,4 These bacteria may reside for some women delivered between 16 weeks 0 days’ and 33 weeks
time in the upper genital tract producing a clinically silent 6 days’ gestation because of maternal medical or ob-
inflammatory response before ultimately resulting in a stetric indications. These women represented the indi-
spontaneous preterm birth.3 The timing of the upper cated preterm delivery group. Research personnel also
genital tract infection is not completely understood. recruited a control group of women that included
However, available evidence indicates that bacterial as- women with the spontaneous onset of labor who deliv-
cension from the lower to the upper genital tract occurs ered a singleton infant at or beyond 37 weeks’ gesta-
early in gestation.3 This is supported by evidence that tional age. To eliminate bias in the selection process for
there is an inverse relationship between upper genital tract the term control group, this group was selected from
infection or inflammation with both birth weight and among all women delivered at term at our institution.
gestational age at delivery among women with a sponta- After recruitment of women with a preterm delivery, the
neous, but not indicated preterm delivery.3,5-7 In addition, next woman with a spontaneous term delivery who met
reports indicate that an activated host immune response, inclusion criteria and who agreed to participate was
as evidenced by elevated concentrations of interleukin-6 recruited. The goal of this process was to allow the term
in amniotic fluid, can be demonstrated as early as the control group to reflect the overall population delivered
midtrimester and, when present, is associated with an at our institution. All women were subsequently mon-
increased risk of early pregnancy loss and spontaneous itored in the Center for Research in Women’s Health at
preterm delivery.8-10 the University of Alabama at Birmingham where written
Very early preterm delivery in an index pregnancy has informed consent was obtained before initiation of the
been significantly associated not only with preterm deliv- study. The study was approved by the Institutional
ery in the subsequent pregnancy, but also with delivery at Review Board.
earlier gestational ages.11 The mechanism(s) that results The women enrolled in all 3 groups received an initial
in the high rate of repeat preterm birth in these women evaluation at approximately 3 months postpartum dur-
remains uncertain. However, the studies linking sponta- ing a scheduled visit to the Center for Research in
neous preterm birth to intrauterine inflammation that can Women’s Health. We chose to perform this initial
be demonstrated as early as the midtrimester and data evaluation at 3 months postpartum: (1) because we
revealing the potential for asymptomatic plasma cell believed there would be resolution of the normal phys-
endometritis to be present in nonpregnant women led us iologic changes from the previous pregnancy and (2)
to speculate that microbial colonization of the endome- there would likely be an opportunity to obtain this
trial cavity may actually precede conception. If this were information before a subsequent pregnancy. Presence of
true, then chronic interpregnancy interval microbial col- pregnancy was excluded before the interval evaluation
onization or inflammation of the endometrium in non- by menstrual history, physical examination, and preg-
pregnant women may partially explain the high incidence nancy test when necessary. The initial evaluation con-
of repeat spontaneous preterm birth in women with a sisted of a history and physical examination, including a
previous spontaneous preterm delivery. Therefore, we speculum examination (described later). Specimens
hypothesized that nonpregnant women with a recent obtained included a vaginal smear to be Gram stained
spontaneous preterm delivery would have higher frequen- for determination of bacterial vaginosis and cervical and
cies of endometrial microbial colonization and plasma endometrial cultures. Specific cervical cultures included
cell endometritis compared with nonpregnant women those for Ureaplasma urealyticum, Mycoplasma species,
Andrews et al 741
Neisseria gonorrhoeae, group B beta Streptococcus, collection of the endometrial specimen was designed to
Trichomonas vaginalis, and a ligase chain reaction test minimize such contamination. We selected several com-
for Chlamydia trachomatis. Endometrial specimens were mon lower tract microbial markers to confirm that the
obtained (described later) and cultures for these same method of sampling alone did not result in positive
microorganisms were performed as well as for aerobic endometrial cultures. These microbial markers included
and anaerobic bacteria. Histopathology was also per- group B beta Streptococcus, Ureaplasma and Myco-
formed on the endometrial specimens as described later. plasma species, and data validating the use of these
microorganisms to document a low incidence of microbial
contamination of the endometrial specimen are reported
Microbiologic sampling of the cervix and in the Results section.
endometrium
Isolation and identification of mycoplasmas
After confirming the absence of pregnancy by menstrual
history and a pregnancy test (if indicated), a sterile Media (10B, A8, and SP4 broth and agar) were prepared
speculum was placed in the vagina to expose the cervix. and quality controlled as described previously12,13 for
Appropriate swabs for the individual cervical cultures cultivation of mycoplasmas. All swab-inoculated speci-
were then obtained as were specimens from the poste- mens and tissue homogenates were thoroughly mixed on
rior vaginal fornix to perform the assessment for bac- a vortex mixer. Five 10-fold dilutions were made from a
terial vaginosis. After all vaginal and cervical specimens 100 mL aliquot of each specimen type in 10B and SP4
were obtained, the vagina and cervix were liberally broths. An aliquot (20 mL) of the undiluted specimen and
prepped with a povidone-iodine 10% solution. A sterile each dilution was inoculated onto A8 and SP4 agar. All
plastic endometrial suction device (Pipelle, Cooper plated media were incubated at 37(C under 5% CO2 in a
Surgical, Wilton, Conn) was then used to obtain an humidified incubator. A8 plates were held for 2 weeks and
endometrial sample for culture by using a technique SP4 plates for 6 weeks before being designated negative.
designed to minimize the possibility of microbial con- Plates were evaluated microscopically for growth every
tamination of the endometrial specimen with microor- 1 to 3 days. All broths were incubated at 37(C under
ganisms from the cervix and vagina. The technique used appropriate atmospheric conditions. Ten B broths were
careful attention to the placement of the Pipelle into the held for 2 weeks and SP4 broths were held for 8 weeks and
cervix and endometrial cavity so that it did not come in blind passaged between days 14 and 21. Any broth tube
contact with the vaginal walls. This technique protected showing color change suggestive of mycoplasmal growth
against contamination with vaginal bacteria until the tip was subcultured to solid media and incubated further.
of the Pipelle was passed beyond the internal cervical os. Colonies of U urealyticum were identified on A8 agar
Once within the endometrial cavity, the Pipelle plunger by urease production in the presence of CaCl2 indicator.
was then completely withdrawn to the hub end and the Mycoplasma species were identified as M hominis by
endometrial specimen was obtained. After obtaining polymerase chain reaction.14 Positive and negative con-
the specimen, and before removing the Pipelle from the trols were run at the same time as the clinical isolates to
endometrial cavity, the hub end of the Pipelle was cut ensure that there were no cross reactions or false-positive
with a pair of sterile scissors. This released the vacuum or false-negative results.
within the Pipelle lumen and prevented aspiration of
cervical microorganisms (and contamination of the Isolation and identification of aerobic and
endometrial specimen) as the Pipelle tip was withdrawn anaerobic bacteria
through the internal os. A sterile syringe and connector
was then used to aspirate the specimen. The endometrial All endometrial samples were placed into appropriate
specimen was then aseptically removed from the syringe transport media used for the recovery of both aerobic and
into a sterile petri dish and portions of the specimen anaerobic bacteria, and the samples were processed
were placed in appropriate media for transport to the within 4 hours of collection. The following media were
Laboratory for Research in Women’s Health at inoculated for aerobic bacterial isolation: thioglycolate
the University of Alabama Birmingham. On arrival in broth, chocolate agar, trypticase soy agar containing 5%
the laboratory, all specimens were processed for culture sheep blood, and human bilayer with Tween (HBT) agar.
by using appropriate culture media and atmospheric The thioglycolate broth was held for 5 days and the plated
conditions. media for 48 hours before being designated as negative.
Although the microbiologic focus of this investigation All plated media were incubated at 35(C in 5% CO2.
involved an evaluation of microbial colonization of the Colonies were identified on the basis of Gram stain
upper genital tract, specimen contamination by microor- reaction and standard biochemical procedures. Identifi-
ganisms residing in the lower genital tract was an impor- cation procedures were those described in the Manual of
tant potential confounding factor. The technique used for Clinical Microbiology,15 and Bailey and Scott’s Diagnostic
742 Andrews et al
Microbiology.16 In addition, anaerobic media used for all plates were placed into an incubator containing 5%
primary isolation included Brucella blood agar, and CO2 at 35(C. The plates were held for 48 hours before
chopped meat carbohydrate broth. All media were pre- being designated as negative for N gonorrhoeae. Growth
reduced before inoculation. Additional prereduced media exhibiting typical colonial and microscopic morphology
used when needed were kanamycin-vancomycin laked was identified by using standard biochemical and enzy-
blood agar and Columbia CNA agar. Cultures were matic reactions as described in the Manual of Clinical
examined for growth after 48 to 72 hours of incubation Microbiology.15
in an anaerobic chamber atmosphere of 90% N2, 5% CO2,
and 5% H2 at 35(C. Negative primary isolation plates Group B streptococcal cultures
were discarded after 5 days. Liquid media were observed Endocervical and endometrial swabs were each used to
for up to 7 days, and any broth showing evidence of inoculate a Todd-Hewitt broth containing gentamicin
growth at any time was Gram stained and subcultured to (8 mg/mL) and nalidixic acid (15 mg/mL). After a
solid media. In addition, any anaerobic growth was minimum 24-hour incubation at 35(C, the broth was
subcultured and incubated aerobically for determination subcultured onto a 5% sheep blood agar plate. The plate
of the organism’s atmospheric requirements. Organisms was held for 48 hours and was examined for the typical
that were confirmed to be obligate anaerobes were iden- colonial morphology and hemolysis of group B beta
tified by standard laboratory procedures as outlined in the Streptococcus (S agalactiae) before being designated as
Wadsworth Anaerobic Bacteriology Manual,17 and Prin- negative. Positive cultures were confirmed with Gram
ciples and Practices of Clinical Anaerobic Bacteriology.18 stain and standard Lancefield grouping techniques.
These techniques included rapid enzymatic biochemical
reactions, antibiotic disc susceptibility testing, and addi- Endometrial histopathology
tional testing as required.
Histopathology was performed by a single pathologist.
Endometrial tissue sections for light microscopic exam-
Trichomonas cultures ination were fixed in 10% neutral buffered formalin,
paraffin-embedded, and 5 mm sections stained with
InPouch TV (Biomed Diagnostics) trichomonas culture
hematoxylin and eosin. Histologic endometritis was
pouches were used for the cultivation of T vaginalis. The
characterized by identification of plasma cells under
pouches were incubated at 37(C for 5 days and were
40! magnification.
examined microscopically each day for the presence of
characteristic motile trichomonads before being desig-
nated as negative. Statistical analyses
All statistical analyses were performed with SAS version
Chlamydia ligase chain reaction assay 8.2 software (Cary, NC). Categorical variables were
compared with the use of c2 tests. Means for the 3 groups
Swab specimens of the cervix and of the endometrial were compared by using analysis of variance with a
tissue were stored in buffer at ÿ70(C for batched Duncan multiple range test for multiple comparisons.
analysis. Detection of C trachomatis was accomplished Comparisons of means between 2 groups were performed
with the ligase chain reaction (LCR) for DNA amplifi- with a 2-tailed unpaired Student t test or Wilcoxon rank
cation (Abbott Laboratories, Chicago, Ill). The tech- sum test where indicated. We calculated adjusted odds
nique for LCR Chlamydia assay in both cervical and ratios for endometrial colonization for previous sponta-
endometrial specimens was a modification of the proce- neous and indicated deliveries compared with term by
dure as previously described.19 We have reported that using logistic regression analysis adjusting for possible
this procedure is superior to culture for the identification confounding variables. To compare matched cervical and
of this organism.20,21 endometrial culture results, McNemar’s test for paired
data was performed. An alpha level of .05 was used to
Gonorrhea cultures determine statistical significance. The sample size allows
for a statistical power of 99% for an effect size of 50%
All sample sites were directly inoculated at the time of ranging down to a power of 82% for an effect size of 15%
collection (endocervical) or processing (endometrium) (with alpha = .05).
onto a selective plate containing GC-Lect Agar (Becton-
Dickinson, Franklin Lakes, NJ) and which used the Results
Jembec System. This product contains a CO2-generating
system that provides an appropriate environment of A total of 820 women were enrolled in this study
carbon dioxide required for the growth of Neisseria including 375 (45.7%) in the prior spontaneous preterm
gonorrhoeae. Within 1 hour of collection or processing, birth group, 142 (17.3%) in the prior indicated preterm
Andrews et al 743
Table I Selected characteristics of the study population Table II Most common endometrial bacterial isolates
Preterm delivery Term Percent Positive cultures
Sponta- Microorganisms Overall SPTD IPTD STD P value*
Characteristic Overall neous Indicated Delivery P-value Gardnerella vaginalis 46 45 42 47 .539
Black 71% 70% 58% 79% !.0001 Lactobacillus spp. 37 40 39 32 .084
White 32% 34% 49% 21% !.0001 S viridans 15 17 8 16 .043
Unmarried 69% 66% 57% 77% !.0001 Peptostreptococcus 14 16 13 11 .162
%12 y of 31% 28% 24% 37% .002 species
education M hominis 8 8 7 7 .775
Smoking* 23% 27% 24% 18% .024 U urealyticum 7 8 1 8 .023
* Self reported. S agalactiae 6 6 7 5 .480
Staphylococcus spp. 5 5 4 6 .663
(coagulase negative)
birth group, and 303 (37.0%) in the prior term delivery Diphtheroid spp. 4 4 4 5 .804
group. The mean gestational age at delivery in the Gemella morbillorum 4 4 4 4 .941
preceding pregnancy was 26.4 G 4.7 weeks in the Enterococcus faecalis 3 2 5 2 .130
spontaneous preterm delivery group, 27.2 G 4.8 weeks Propionibacterium acnes 3 5 3 1 .001
in the indicated preterm birth group, and 39.6 G 1.2 Prevotella bivius 3 4 4 2 .196
weeks in the term delivery group (P ! .001). Mean Mobiluncus species 2 2 4 2 .213
delivery gestational ages in the 2 prior preterm birth C trachomatis 2 2 3 2 .883
groups were not significantly different. The interval from S aureus 2 2 1 3 .551
delivery to sampling for cultures and histology was SPTD, Spontaneous preterm delivery; IPTD, indicated preterm delivery;
similar in the spontaneous preterm birth group (83 days STD, spontaneous term delivery.
* P value for comparison of the 3 study groups (SPTD, IPTD, and
G 17.6 days; mean G SD), the indicated preterm birth STD).
group (83 days G 16.7 days), and the spontaneous term
group (82 days G 15.7 days; P = .980). Characteristics
of the study population are provided in the Table I. In metrial cultures were positive for 1 or more microorgan-
this study population, women with a term delivery were isms in 83% of the women who had plasma cell
more likely to be black, were less educated, and were endometritis. Plasma cells were present in 41% of cases
more likely to be single (Table I). in which the corresponding endometrial culture was
Endometrial cultures were positive for at least 1 micro- negative. Combining the results of the cultures and
organism in 82% of the women. The most common histology, 86% of the women had evidence of either
endometrial isolates are listed in Table II. The median endometrial microbial colonization or histologic inflam-
number of different bacterial species isolated from the mation. Plasma cell endometritis was higher in the spon-
endometrium in the spontaneous preterm delivery group taneous preterm delivery (43%) and the indicated preterm
was not significantly greater than the indicated preterm delivery (40%) groups compared with the spontaneous
delivery group (2, range 0-9 vs 2, range 0-6 microorgan- term delivery group (33%) but this difference was not
isms, P = .132) but was significantly greater than the statistically significant (P = .160).
term delivery group (2, range 0-9 vs 1, range 0-7 micro- To address the issue of potential microbial contam-
organisms, P = .015). However, no significant difference ination of the endometrial specimen during its acquisi-
in endometrial cultures positive for any microorganism tion by bacteria from the cervix and vagina, cervical
were observed among women after a spontaneous versus cultures were obtained in all women for 3 microorgan-
an indicated preterm delivery (85% vs 79%, P = .102), or isms that are commonly present in the cervix (U
a spontaneous preterm versus a spontaneous term deliv- urealyticum, M hominis, and S agalactiae). Although
ery (85% vs 81%, P = .123). The distribution of the frequently isolated from the cervix, isolation of these 3
commonly isolated microorganisms from the endome- microorganisms was significantly less common from the
trium was largely similar among the 3 groups (Table II). endometrium (Table III).
In a logistic regression analysis controlling for race,
marital status, maternal education, smoking, and mater-
nal body mass index, the odds ratio for having a positive Comment
endometrial culture for any organism was 1.49 (95% CI
0.96-2.31) in the previous spontaneous preterm birth These results do not support the original hypothesis that
group and 1.06 (95% CI 0.61-1.84) in the previous microbial colonization of the endometrium and plasma
indicated preterm birth group. cell endometritis would be higher among women with a
Plasma cell endometritis was present in 39% of the 506 recent spontaneous preterm delivery compared with
specimens adequate for histologic evaluation and endo- women with a recent indicated preterm or spontaneous
744 Andrews et al
these 3 microorganisms were the result of contamination

Table III Culture results for selected microorganisms iso-
lated from the cervix and endometrium of the endometrial specimen with cervical bacteria when
the concomitant cervical culture was positive for the same
Cervix/endometrium Microorganism microorganism, the estimated rate of contamination
culture (%) U urealyticum M hominis S agalactiae would range only between 4.1% and 7.5%. Therefore,
Negative/negative 43.7 70.0 78.5 we believe that the extremely high rate of endometrial
Negative/positive 1.1 0.1 1.9 microbial colonization observed in this study largely
Positive/negative 49.6 22.4 15.6 represents a true picture of the microbiologic status of
Positive/positive 5.5 7.5 4.1 the endometrium in these women.
P ! .001 in each case (McNemar’s test). Another potential explanation for the high rate of
positive endometrial cultures and plasma cell endometri-
tis observed in this study is ascertainment bias caused by
term birth. Instead, the frequencies of microbial coloni- the limitation of the upper genital tract assessment to
zation and plasma cell endometritis assessed at 3 months’ 1 time point at 3 months’ postpartum. This time frame
postpartum were similar in all 3 groups of women. was chosen for several reasons. First, genital tract
Therefore, these results do not explain the increased risk changes resulting from hormonal, mechanical, and other
of subsequent preterm birth among women with a prior influences of pregnancy are believed to resolve by 3
early spontaneous preterm birth before 34 weeks’ gesta- months’ postpartum. Second, recruitment and retention
tional age. A striking and unexpected finding in this study of subjects for this study, because of the necessity of
was the extraordinarily high number of positive endome- voluntary submission to an endometrial biopsy, was
trial cultures and the high frequency of plasma cell anticipated to be challenging. We believed that our
endometritis in all 3 groups. Plasma cell endometritis success in this regard would be improved if the interval
was present in nearly 40% of the specimens that were between first contact immediately after delivery and
adequate for histologic evaluation and four fifths of all ultimate enrollment in the study were shorter rather
women in the study had endometrial bacterial coloniza- than longer. Third, we believed that serial endometrial
tion. Combining the results of the cultures and histology, biopsies would be unacceptable to women and would
86% of the women had culture or histologic evidence of potentially have untoward effects. Therefore, we agreed
endometrial microbial colonization or inflammation. to limit the number of endometrial biopsies performed on
One potential explanation for the high frequency of a given subject. Performing a single biopsy on groups of
endometrial microbial colonization observed in this study women at different time intervals also would have mark-
is the possibility of microbial contamination of the edly increased the expense of the study and the number of
endometrial tissue during acquisition of the specimen women required for meaningful analyses. Nevertheless,
with bacteria from the cervix or vagina. Such microbial we acknowledge that our endometrial culture and histo-
contamination could occur despite the precautions taken logic observations are limited to a single investigation at
to limit this complication (see Methods). However, we do approximately 3 months’ postpartum. Lower frequencies
not believe that microbial contamination of the endome- of endometrial colonization or plasma cell endometritis
trial specimen is the explanation for the observed results might have been observed at later postpartum time
for several reasons. First, such contamination would not intervals in any or all of the 3 study groups as has been
explain the high frequency of plasma cell endometritis reported for plasma cell endometritis.22 Indeed, plasma
that was observed in all 3 groups. Additional compelling cell endometritis is a common occurrence for several
support for the conclusion that the high observed rate of months after birth.22
microbial endometrial colonization is not due to gross In summary, the results of this study indicate that
contamination is derived from our data of matched differences in endometrial microbial colonization or
cervical and endometrial cultures for specific microor- inflammation do not explain the high rate of repeat
ganisms (Table III). If the high rate of positive endome- preterm delivery among women with a prior spontaneous
trial cultures resulted from frequent contamination of the preterm birth. However, the high frequency of positive
endometrial specimen with cervical bacteria, then we endometrial cultures and histologic endometritis ob-
would have expected to find a high rate of positive served in otherwise asymptomatic nonpregnant women
endometrial cultures for these three organisms (U ure- (without signs or symptoms of endometrial infection) do
alyticum, M hominis, and S agalactiae) when the corre- call for a reevaluation of current assumptions about the
sponding cervical culture was positive. In fact, this was microbiologic status of the endometrium in nonpregnant
not the case. Although frequently isolated from the cervix, women. A prevailing assumption is that the endometrium
isolation of U urealyticum, M hominis, and S agalactiae in nonpregnant women is largely devoid of bacterial
was much less common from the endometrium. On the colonization. The previously described results argue oth-
basis of results depicted in Table III, even in the unlikely erwise, at least 3 months after delivery. If, in fact,
event that 100% of the positive endometrial cultures for endometrial colonization is a common condition in
Andrews et al 745
nonpregnant women, why then are some women who 9. Wenstrom KD, Andrews WW, Tamura T, DuBard MB, Johnston
have such colonization destined to have subsequent KE, Hemstreet GP. Elevated amniotic fluid interleukin-6 levels at
genetic amniocentesis predict subsequent pregnancy loss. Am J
infection-related spontaneous preterm births while most Obstet Gynecol 1996;175:830-3.
women are not? The answer may be that there are 10. Wenstrom KD, Hauth JC, DuBard MB, Andrews WW, Golden-
differences in the host response to the presence of these berg RL, Cliver SP. Elevated second trimester amniotic fluid
bacteria rather than the mere presence of the bacteria interleukin-6 levels predict preterm delivery. Am J Obstet Gynecol
themselves. One example might be the presence of certain 1998;178:546-50.
11. Goldenberg RL, Mayberry SK, Copper RL, Dubard MB, Hauth
proinflammatory cytokine gene polymorphisms that ge- JC. Pregnancy outcome following a second-trimester loss. Obstet
netically program for an enhanced immune response in Gynecol 1993;81:444-6.
the presence of bacteria resulting in an increased release of 12. Cassell GH, Blanchard A, Duffy L, Crabb D, Waites KB.
bioactive factors that unfortunately lead to a spontaneous Mycoplasmas. In: Howard BJ, editor. Clinical and pathogenic
preterm birth. In any case, elucidation of this question microbiology. St Louis: Mosby-Year Book; 1993. p. 491-502.
13. Cassell GH, Blanchard A, Duffy L, Crabb D, Waites KB.
remains an unanswered challenge and an important Mycoplasmas. In: Howard BJ, Klaas J, Rubin SJ, Weissfeld AS,
subject for future research. Tilton RC, editors. Clinical and pathogenic microbiology.
St Louis: Mosby-Year Book; 1994. p. 491-502.
14. Blanchard A, Yanez A, Dybvig K, Watson HL, Griffiths G,
References Cassell GH. Evaluation of intraspecies genetic variation within the
16S rRna gene of Mycoplasma hominis and detection by polymer-
1. Hack M, Fanaroff AA. Outcomes of children of extremely low ase chain reaction. J Clin Micro 1993;31:1358-61.
birthweight and gestational age. Sem Neonatol 2000;5:89-106. 15. Murray PR, Baron EJ, Pfaller MA, Tenover FC, Yolken RH,
2. Martin JA, Hamilton BE, Ventura SJ, Menacker F, Park MM, editors. Manual of clinical microbiology. 7th ed. Washington
Sutton PD. Births: final data for 2001. Natl Vital Stat Rep (DC): American Society for Microbiology Press; 1999.
2002;51:1-102. 16. Baron EJ, Peterson LR, Finegold SM, editors. Bailey and Scott’s
3. Goldenberg RL, Hauth JC, Andrews WW. Intrauterine infection diagnostic microbiology. 9th ed. St Louis: Mosby-Year Book;
and preterm delivery. N Engl J Med 2000;342:1500-7. 1994.
4. Romero R, Mazor M. Infection and preterm labor. Clin Obstet 17. Sutter VL, Citron DM, Edelstein MAC, Finegold SM. Wades-
Gynecol 1988;31:553-84. worth anaerobic bacteriology manual. 4th ed. Belmont (CA): Star
5. Andrews WW, Hauth JC, Goldenberg RL. Infection and preterm Publishing Company; 1985.
birth. Am J Perinatol 2000;17:357-65. 18. Engelkirk PG, Duben-Engelkirk J, Dowell VR Jr. Principles and
6. Hauth JC, Andrews WW, Goldenberg RL. Infection-related risk practices of clinical anaerobic bacteriology. Belmont (CA): Star
factors predictive of spontaneous preterm birth. Prenat Neonat Publishing Company; 1992.
Med 1998;3:86-90. 19. Dille BJ, Butzen CC, Birkenmeyer LG. Amplification of Chla-
7. Andrews W, Hauth J, Goldenberg R, Mazur M, Gomez R, mydia trachomatis DNA by ligase chain reaction. J Clin Microbiol
Romero R, et al. Amniotic fluid IL-6: correlation with chorioamn- 1993;31:729-31, 1993.
ion colonization and gestational age in women delivered following 20. Schachter J, Stamm WE, Quinn TC, Andrews WW, Burzak JD,
spontaneous labor versus indicated delivery. Am J Obstet Gynecol Lee HH. Ligase chain reaction (LCR) to detect Chlamydia
1994;170:277. trachomatis infection of the cervix. J Clin Micribiol 1994;32:2540-3.
8. Romero R, Munoz H, Gomez R, Sherer F, Ghidini A, DeVore G, 21. Andrews WW, Lee HH, Roden WJ, Mott CW. Detection of
et al. Two thirds of spontaneous abortions/fetal deaths after genitourinary tract Chlamydia trachomatis infection in pregnant
midtrimester genetic amniocentesis are the result of a pre-existing women by ligase chain reaction. Obstet Gynecol 1997;89:556-60.
subclinical inflammatory process of the amniotic cavity. Am J 22. Sharman A. Post-partum regeneration of the human endome-
Obstet Gynecol 1995;172:261. trium. J Anat 1953;87:1-14.
www.ajog.org
Variation in microbiologic profiles among pregnant

women with bacterial vaginosis
Leonardo Pereira, MD,a,* Jennifer Culhane, PhD, MPH,b Kelly McCollum, MPH,b
Kathy Agnew, BS,c Paul Nyirjesy, MDb
Division of Maternal-Fetal Medicine, Oregon Health and Science University, Portland, ORa; Departments of
Obstetrics and Gynecology, Drexel University, Philadelphia, PAb; and University of Washington, Seattle, WAc
Received for publication September 27, 2004; revised January 26, 2005; accepted January 26, 2005
KEY WORDS Objective: The purpose of this study was to determine if clinical findings and sociodemographic
Bacterial vaginosis variables among bacterial vaginosis (BV)-positive pregnant women are associated with different
Mobiluncus microbiologic profiles.
Pregnancy Study design: Pregnant women were assessed for BV by Nugent criteria. BVC women were
separated into 6 mutually exclusive microbiologic groups. In unadjusted analyses, we compared
(1) sociodemographic and behavioral characteristics, and (2) 3 clinical characteristics among
BVC women with and without Mobiluncus (MC vs M). Unadjusted data were analyzed using
the chi-square test. Multiple logistic regression was used to assess the likelihood of having clinical
signs of BV in women with and without Mobiluncus spp while controlling for confounders.
Results: A total of 1756 BVC pregnant women were followed. The MC group (n = 702) was
significantly more likely than the M group (n = 1054) to be non-Hispanic black (80.9% vs
66.2%; P ! .0001), older than 21 years (61.7% vs 48.7%; P ! .0001), and to have had more than
3 lifetime sexual partners (66.4% vs 54.9%; P ! .0001). The MC group was also more likely to
have clue cells on wet mount (63.9% vs 47.2%; P ! .0001) and a positive amine odor after
addition of KOH (57.2% vs 45.0%; P = .001). There was no difference in other demographic
variables or physician diagnosis of abnormal vaginal discharge. In the adjusted analyses for each
clinical outcome, all findings were consistent with the unadjusted analyses.
Conclusion: BVC pregnant women with Mobiluncus spp are more likely to have clue cells present
on wet mount, a positive amine odor after KOH preparation, and to be older, non-Hispanic
black, and have had more lifetime sexual partners compared to BVC women without any
Mobiluncus species.
Bacterial vaginosis (BV) is a condition in which the

Supported by Center for Disease Control and Prevention/ATPM, normal lactobacillary flora are replaced by an over-
grant nos. TS-0561/TS-0626. growth of Gardnerella vaginalis, Prevotella/Bacteroides
* Reprint requests: Leonardo Pereira, MD, Division of Maternal- species, Mobiluncus species, and other organisms.1-3 The
Fetal Medicine, Mail Code L-458, Oregon Health and Science
University, 3181 SW Sam Jackson Park Road, Portland, OR 97239-
association between BV and preterm birth (PTB) has
3098. been reported by several authors.4-8 However, studies
E-mail: pereiral@ohsu.edu aimed at prevention of PTB by treatment of BV have
doi:10.1016/j.ajog.2005.01.069
Pereira et al 747
yielded differing results.9-10 This discrepancy has led the

Table I Gram-stain scoring system
American College of Obstetricians to take the position
that ‘‘current data do not support the use of BV Score No. of morphotypes/100! oil immersion field
screening as a strategy to identify or prevent preterm 0 None
birth.’’11 1C Less than 1 per field
As our clinical recognition of BV has become stan- 2C 1 to 4 per field
dardized by adopting Nugent criteria,12 we can now 3C 5 to 30 per field
4C More than 30 per field
appreciate that BV is not the same disease in all women.
Microbiologic variation exists among BV-positive indi-
viduals and can be characterized by Gram-stain analysis.
All women with a Nugent score of 7 to 10 are diagnosed as
BV positive, but among these individuals, there is a wide Table II Points* awarded by Gram-stain score
variety of microflora. The purpose of this study was to Morphotype Points
determine if the microbiologic variation among BV
No Lactobacillus spp. 4
positive individuals could explain variations in clinical
1C Lactobacillus spp. 3
signs of BV and differences in the prevalence of BV among 2C Lactobacillus spp. 2
different race/ethnic groups. 3C Lactobacillus spp. 1
4C Lactobacillus spp. 0
No G vaginalis or Prevotella/Bacteroides spp. 0
Material and methods 1C G vaginalis or Prevotella/Bacteroides spp. 1
2C G vaginalis or Prevotella/Bacteroides spp. 2
This study is part of a larger, ongoing clinical prevalence 3C G vaginalis or Prevotella/Bacteroides spp. 3
study assessing the association between maternal stress, 4C G vaginalis or Prevotella/Bacteroides spp. 4
BV, and risk of preterm birth. Pregnant women were 1 to 2C Mobiluncus spp. 1
3 to 4C Mobiluncus spp. 2
recruited from public health centers in Philadelphia,
Pennsylvania, between February 1, 1999 and April 20, * Points were totaled using the following equation: (Lactobacil-
lus C G vaginalis or Prevotella/Bacteroides C Mobiluncus = total
2003. Women were recruited for participation in this
score). Total scores: 0 to 3 normal, 4 to 6 intermediate flora, 7 to
study at the time of their first prenatal care visit, at 10 BV positive.
a mean gestational age of 14 G 6 weeks. All English or
Spanish speaking women with singleton pregnancies
were eligible.
Practitioners collected 2 vaginal smears from all intermediate BV status, and a score of 7 to 10 was defined
study participants, one for Gram-stain and one for wet as positive BV status.
mount. Practitioners also assessed the following clinical Based on the individual components of this classifi-
signs: the presence of clue cells on wet mount, amine cation scheme, all smears were categorized into 1 of 14
odor after addition of potassium hydroxide (KOH), and mutually exclusive microbiologic groups. Based on the
vaginal discharge. microbiologic distribution, it was impossible for 8 of the
Air-dried vaginal smears were assessed for BV accord- groups to achieve a Nugent score above 6. Therefore, all
ing to Nugent criteria.12 Each Gram stain was evaluated BV-positive women (Nugent score 7-10) were classified
for the following morphotypes under oil immersion into one of 6 mutually exclusive microbiologic groups.
(100!) magnification: large uniform gram-positive rods Three microbiologic groups contained Mobiluncus spe-
(corresponding to Lactobacillus spp), small pleomorphic cies (MC), and 3 did not (M).
Gram-variable rods (G vaginalis morphotypes), small Gram-stain scoring was performed by a microbiolo-
Gram-negative rods (Prevotella/Bacteroides morpho- gist (KA) who was blinded to the clinical findings and
types), and curved Gram-variable rods (Mobiluncus demographics of subjects. The investigator (LP) who
morphotypes). After scanning the slide for representative designated individuals into respective microbiologic
areas, each morphotype was quantified from 1C to 4C groups was similarly blinded. The following outcomes
with regard to the average number of morphotypes seen were compared between BV C, MC women and BV C,
on 5 nonadjacent fields, using the scoring system listed in M women: clue cells identified on wet mount, positive
Tables I and II. Each morphotype was given a number amine odor after addition of KOH, and physician
and added to yield a score of 0 to 10 using the following diagnosis of vaginal discharge. The same comparisons
equation: Lactobacillus C G vaginalis/Prevotella/Bacter- were carried out for BV intermediate women.
oides C Mobiluncus = total score. In this scoring system, The statistical significance of all categorical data was
Lactobacillus is inversely scored, compared to G vaginalis/ determined using the chi-square test. To adjust for
Prevotella/Bacteroides and Mobiluncus. A score of 0 to 3 potential confounding variables, and to derive maximum
was considered normal, a score of 4 to 6 corresponded to likelihood estimates of adjusted relative odds with 95%
748 Pereira et al
Figure 1 *No Lactobacillus species were detected in all 6 groups.
CI, multivariate logistic regression analyses were per- G vaginalis and Prevotella/Bacteroides on Gram stain,
formed. Potential confounders were identified a priori by and an absence of Lactobacillus and Mobiluncus species.
reviewing the literature and based on theoretical consid- The second largest subgroup (39.7%) had the same
erations. The following confounders were included in all profile, except that Mobiluncus species were present on
of the multivariate models: race, maternal age, age at Gram stain. The other 4 possible classifications were
sexual debut, number of lifetime sexual partners, recent rarely identified, accounting for just 2.5% of subjects.
smoking, douching, parity, marital status, education, Because of this distribution, women with Mobiluncus
maternal annual income, and foreign-born status. Type spp (40.0% combined 3 groups, MC) were compared
I error was set at 0.05 (two-sided). All analyses were to women without Mobiluncus spp (60.0% combined
conducted using STATA 8.0 (College Station, Tex). 3 groups, M) (Figure 1).
In unadjusted analyses, BVC women with Mobilun-
cus were significantly more likely to be non-Hispanic
Results black (80.9% vs 66.2%; c(df)2 = 45.2(1); P ! .0001), to
be above the median age of 21 (61.7% vs 48.7%;
A total of 4361 women met inclusion criteria and were c(df)2 = 28.7(1); P ! .0001), and to have had more
eligible to participate in this study. Of these, 3881 than 3 lifetime sexual partners (66.4% vs 54.9%;
(89.0%) consented to participate, and 3765 (86.3%) c(df)2 = 22.9(1); P ! .0001) compared with BVC
completed the study. There were 1756 (46.6%) BVC women without Mobiluncus. However, the presence of
women, and 553 (14.7%) BV-intermediate women. The Mobiluncus spp was not associated with age at sexual debut,
largest subgroup of BVC women (57.8%) demonstrated recent smoking, or douching (Figure 2). Furthermore,
Pereira et al 749
Figure 2 BV, bacterial vaginosis; MC, Mobiluncus present; M, Mobiluncus absent; yo, years of age.
in adjusted analyses, women with Mobiluncus spp were outcomes. Nonetheless, we did not find an association
more likely to have clue cells on wet mount (63.9% vs between the presence of Mobiluncus and physician
47.2%; c(df)2 = 23.2(1); P ! .0001) and a positive amine diagnosis of abnormal vaginal discharge (Table III),
odor after addition of potassium hydroxide (57.2% vs suggesting that this null finding is not likely to be the
45.0%; c(df)2 = 11.6(1); P = .002). However, there was result of a type II error.
no difference in the physician finding of an abnormal
vaginal discharge (64.6% vs 61.5%; c(df)2 = 1.18(1); Comment
P = .28). In the adjusted analysis for each clinical
outcome, all findings were consistent with those from In the past 20 years of obstetric research, few topics have
the unadjusted analyses. MC women were 1.7 times been as widely studied as BV. A MEDLINE search
more likely to have clue cells present on wet mount using the keywords ‘‘bacterial vaginosis,’’ and ‘‘preg-
(95%CI: 1.3–2.3), and 1.5 times more likely to have nancy’’ produced more than 500 corresponding articles
amine odor after the addition of KOH to the wet mount since 1984. In fact, the association between BV and
compared with M women. In addition, there was no preterm birth (PTB) has been so consistently reported
difference between the 2 groups with regard to physician that there is little doubt in today’s medical community
finding of abnormal vaginal discharge (OR 1.1; 95% CI that a true relationship exists between them.4-8 Despite
0.8–1.4) (Table III). this body of evidence, however, an understanding of the
When we conducted this same analysis in BV-interme- exact nature of the relationship between BV and PTB
diate women, all of the demographic distributions and remains elusive.
clinical outcomes followed identical trends when compar- Strong association does not equal causation, and
isons were made between MC (n = 35) and M perhaps this explains, at least in part, why treatment
(n = 518) BV-intermediate women (data not shown). trials aimed at eradication of BV to decrease the rate of
However, none of the findings were statistically signifi- PTB have yielded differing results.9-10 Another possibil-
cant. ity is that although the relationship with BV and PTB
This study had sufficient power (O80%) to detect may in fact be causative, it may be causative only for
a minimum odds ratio of 1.5 for the association between certain individuals. This may be due to differences in
the presence of Mobiluncus spp and each of the 3 clinical host response to BV, confounding risk factors for PTB,
750 Pereira et al
Table III Multivariate logistic regression of the association between the presence of Mobiluncus species in BV-positive women and
3 clinical signs*
Mobiluncus species Mobiluncus species Mobiluncus species
Dependent variable presenty OR (95% CI) present (%) absent (%) P value
Clue cells on wet mount 1.7 (1.3, 2.3) 63.9 47.2 ! .00001
Amine odor after KOH addition 1.5 (1.1, 2.0) 57.2 45.0 .002
Physician diagnosis of abnormal vaginal discharge 1.1 (0.8, 1.4) 64.6 61.5 .28
* All models adjusted for the following variables: race, maternal age, number of lifetime sexual partners, recent smoking, douching, parity, marital
status, education, maternal annual income, and foreign-born status.
y
Mobiluncus species negative women are the reference group for all models.
or variations in the profile of pathogens constituting BV women is consistent with the findings of previous
among different women. authors.
The importance of the Nugent scoring system in Royce et al previously reported significant race/ethnic
providing standardized diagnostic criteria for BV cannot variation in bacterial morphology among BV-positive
be overstated. However, the diagnosis of BV by Nugent pregnant women, with black women 9.26 times as likely
score alone is limited because it clusters women with to have Mobiluncus compared with their white counter-
different vaginal flora into one group of BV-positive parts.17 In a nonpregnant population of women seeking
individuals. The limitations of Nugent criteria have treatment at an STD clinic, Hillier et al also reported
already been recognized by some authors who have race/ethnic differences in the presence of Mobiluncus spp
reported subsets of women with intermediate BV scores identified by Gram stain, with black women 2.5 times as
who appear to be at similar risk of preterm birth as their likely to be MC compared with white patients.18 Why
BV-positive counterparts.13-15 By further refining our black women exhibit more colonization of Mobiluncus is
diagnostic criteria for BV, we may be able to better not known. To date, sexual transmission does not seem
identify which individuals are at highest risk for preterm to be the primary method of acquisition of Mobiluncus
birth. spp, and a gastrointestinal reservoir has been sug-
We were able to identify 6 different morphotypes gested.19 It is possible that race/ethnic differences in
among women with a Nugent score of 7 to 10 by Gram- host factors that modulate bacterial binding to vaginal
stain analysis. We observed that BV-positive pregnant and gastrointestinal cells may account for the observed
women with any Mobiluncus spp were significantly more race/ethnic differences in Mobiluncus carriage rates. For
likely to be non-Hispanic black, above age 21, to have example, Lewis antigens are thought to inhibit the
had more than 3 lifetime sexual partners, to have clue binding of infectious organisms to human cells. Lewis
cells on wet mount, and to have a positive amine odor antigen deficiency has been associated with increased
after KOH preparation compared to those without rate of urinary tract infections in women,20 recurrent
Mobiluncus spp (Figure 2 and Table III). The associa- vaginal candidiasis,21,22 and Helicobacter pylori infec-
tion between Mobiluncus spp and a positive amine odor tions.23 Some data suggest that blacks are more likely to
after KOH preparation (‘‘whiff test’’) has been pre- be Lewis antigen-negative, and differences in Lewis
viously reported,16 and our findings in this study antigen status or a similar host factor could potentially
confirmed the hypothesis that variations in the micro- explain the race/ethnic variations in Mobiluncus coloni-
biologic profile of BV are associated with different zation rates.24
demographic variables and clinical signs. In conclusion, different BV morphotypes exist among
An important limitation of our study is the reliance individuals. Demographic variations (age, race/ethnic-
on Gram-stain analysis to differentiate the different BV ity, etc) are observed when comparing BVC women by
morphotypes. Ideally, this should be done on the basis Mobiluncus spp status. Microscopy also demonstrates
of culture results because Gram-stain interpretation, that Mobiluncus spp is associated with cellular variation
specifically differentiating Gram-variable, pleiomorphic (more clue cells) and diagnostic variation (a positive
bacillus (G vaginalis) from more uniform Gram-negative amine odor after KOH preparation). However, this
bacillus (Prevotella/Bacteroides spp) can be difficult and distribution of BV morphotypes does not seem to
introduces a subjective element.12 However, since our explain why some BV-positive individuals are diagnosed
results focus primarily on the presence or absence of with abnormal vaginal discharge while others are not.
Mobiluncus spp, which can be identified with excellent Differential expression of Lewis antigens may provide
reliability on Gram stain,12 this should not affect the a genetic mechanism for the increased prevalence of
validity of our observations. Furthermore, our observa- Mobiluncus among BV-positive non-Hispanic blacks
tion that a significantly higher proportion of BVC black and warrants further study. In the future, we plan to
women have Mobiluncus spp compared with BVC white investigate whether the BV morphotypes characterized
Pereira et al 751
in this study are associated with different vaginal levels 11. American College of Obstetrics and Gynecologists. Assesment of
of hydrolytic enzymes and proinflammatory cytokines risk factors for preterm birth. ACOG Practice Bulletin no. 31.
and, ultimately, with different risks of preterm birth. As 12. Nugent RP, Krohn MA, Hillier SL. Reliability of diagnosing
our understanding of BV evolves, it is important to bacterial vaginosis is improved by a standardized method of gram
recognize that microbiologic and clinical variation exists stain interpretation. J Clin Microbiol 1991;29:297-301.
among BV-positive women, and our thinking of BV- 13. Ugwumadu A, Manyonda I, Reid F, Hay P. Effect of early oral
positive women as a homogeneous population harbor- clindamycin on late miscarriage and preterm delivery in asymp-
tomatic women with abnormal vaginal flora and bacterial vagi-
ing the same disease must evolve as well. nosis: a randomised controlled trial. Lancet 2003;361:983-8.
14. Goffinet F, Maillard F, Mihoubi N, Kayem G, Papiernik E,
Cabrol D, et al. Bacterial vaginosis: prevalence and predictive
value for premature delivery and neonatal infection in women with
References preterm labour and intact membranes. Eur J Obstet Gynecol
Repro Biol 2003;108:146-51.
1. Eschenbach DA, Gravett MG, Chen KC, Hoyme UB, Holmes 15. Cauci S, Hitti J, Noonan C, Agnew K, Quadrifoglio F, Hillier SL,
KK. Bacterial vaginosis during pregnancy. An association with et al. Vaginal hydrolytic enzymes, immunoglobulin A against
prematurity and postpartum complications. Scand J Nephrol Gardnerella vaginalis toxin, and risk of early preterm birth among
(Suppl) 1984;86:213-22. women in preterm labor with bacterial vaginosis or intermediate
2. Hillier SL, Krohn MA, Rabe LK, Klebanoff SJ, Eschenbach DA. flora. Am J Obstet Gynecol 2002;187:877-81.
The normal vaginal flora, H2O2-producing lactobacilli, and 16. Hallen A, Pahlson C, Forsum U. Bacterial vaginosis in women
bacterial vaginosis in pregnant women. Clin Infect Dis 1993; attending STD clinic: diagnostic criteria and prevalence of
16(Suppl 4):S273-81. Mobiluncus spp. Gentiourin Med 1987;63:386-9.
3. Holst E, Goffeng AR, Andersch B. Bacterial vaginosis and vaginal 17. Royce RA, Jackson TP, Thorp JM Jr, Hillier SL, Rabe LK,
microorganisms in idiopathic premature labor and association Pastore LM, et al. Race/ethnicity, vaginal flora patterns, and pH
with pregnancy outcome. J Clin Microbiol 1994;32:176-86. during pregnancy. Sex Transm Dis 1999;26:96-102.
4. Hillier SL, Nugent RP, Eschenbach DA, Krohn MA, Gibbs RS, 18. Hillier SL, Critchlow CW, Stevens CE, Roberts MC, Wolner-
Martin DH, et al. Association between bacterial vaginosis and Hanssen P, Eschenbach DA, et al. Microbiological, epidemiolog-
preterm delivery of a low-birth-weight infant. The Vaginal ical and clinical correlates of vaginal colonisation by Mobiluncus
Infections and Prematurity Study Group. N Engl J Med 1995; species. Genitourin Med 1991;67:26-31.
333:1737-42. 19. Holst E. Reservoir of four organisms associated with bacterial
5. Gravett MG, Hummel D, Eschenbach DA, Holmes KK. Preterm vaginosis suggests lack of sexual transmission. J Clin Microbiol
labor associated with subclinical amniotic infection and with 1990;28:2035-9.
bacterial vaginosis. Obstet Gynecol 1986;67:229-37. 20. Sheinfeld J, Schaeffer AJ, Cordon-Cardo C, Rogatko A, Fair WR.
6. Goldenberg RL, Hauth JC, Andrews WW. Intrauterine infection The association of the Lewis antigen blood group phenotype with
and preterm delivery. N Eng J Med 2000;342:1500-7. recurrent urinary tract infections in women. N Engl J Med
7. McGregor JA, French JI, Seo K. Premature rupture of membranes 1989;320:773-7.
and bacterial vaginosis. Am J Obstet Gynecol 1993;169:463-6. 21. Hilton E, Chandrasekaran V, Rindos P, Isenberg HD. Association
8. Gibbs RS. Chorioamnionitis and bacterial vaginosis. Am J Obstet of recurrent candidal vaginitis with inheritance of Lewis blood
Gynecol 1993;169:460-2. group antigens. J Infect Dis 1995;172:1616-9.
9. Carey JC, Klebanoff MA, Hauth JC, Hillier SL, Thom EA, Ernest 22. Chaim W, Foxman B, Sobel JD. Association of recurrent vaginal
JM, et al. Metronidazole to prevent preterm delivery in pregnant candidiasis and secretory ABO and Lewis phenotype. J Infect Dis
women with asymptomatic bacterial vaginosis. N Engl J Med 1997;176:828-30.
2000;342:534-40. 23. Rothenbacher D, Weyermann M, Bode G, Kulaksiz M, Stahi B,
10. McDonald HM, O’Loughlin JA, Vigneswaran R, Jolley PT, Harvey Brenner H. Role of Lewis A and Lewis B blood group antigens in
JA, Bof A, et al. Impact of metronidazole therapy on preterm birth in Helicobacter pylori infection. Helicobacter 2004;9:324-9.
women with bacterial vaginosis flora (Gardnerella Vaginalis): 24. Molthan L. Lewis phenotypes of American Caucasians, American
a randomized, placebo-controlled trial. BJOG 1997;104:1391-7. Negroes and their children. Vox Sang 1980;39:327-30.
www.ajog.org
Use of DNA hybridization to detect vaginal pathogens

associated with bacterial vaginosis among asymptomatic
pregnant women
Kim A. Boggess, MD,a,b Thomas N. Trevett, MD,a Phoebus N. Madianos, DDS, PhD,b
Lorna Rabe, BS,c Sharon L. Hillier, PhD,c James Beck, PhD,b Steven
Offenbacher, DDS, PhDb
Department of Obstetrics and Gynecology,a and Center for Oral and Systemic Diseases,b University of North
Carolina at Chapel Hill, Chapel Hill, NC; Department of Obstetrics, Gynecology and Reproductive
Sciences, Magee-Women’s Hospital of the University of Pittsburgh Medical Center, Pittsburgh, PAc
KEY WORDS Objective: The purpose of this study was to determine whether microbial DNA hybridization is
Bacterial vaginosis a useful method to study bacterial vaginosis in asymptomatic pregnant women.
Diagnosis Study design: Vaginal specimens were collected at !26 weeks’ gestation from 230 women, and
Pregnancy analyzed for bacterial vaginosis by both Gram stain using Nugent criteria and DNA
hybridization for Gardnerella vaginalis, Prevotella bivia, Bacteroides ureolyticus, and Mobiluncus
curtisii. Results were analyzed using McNemar’s paired test and chi-square test for trend, with
significance set at P ! .05.
Results: By Gram stain, 60 (26.1%) of 230 were positive for bacterial vaginosis, and 134 (58.3%)
were negative. By DNA hybridization, 99 (43%) were positive for at least 1 pathogen, and DNA
results were significantly associated with Gram-stain results (P ! .01). As the Nugent score
progressed from normal to abnormal flora, the proportion with O1 pathogen detected by DNA
hybridization increased significantly (P ! .001).
Conclusion: DNA hybridization may be a useful method to study shifts in vaginal flora during
pregnancy.
Bacterial vaginosis has been reported in 16% to 50% post-cesarean section endometritis, spontaneous mis-
of pregnant women.1,2 Bacterial vaginosis has been carriage, prematurity, and fetal growth restriction.1,3,4
associated with numerous adverse outcomes, including Despite its prevalence and impact, some important
biologic questions remain unanswered. The etiology of
this complex microbiologic syndrome characterized by
Supported by NIDCR grants R01-DE-12453, P60-DE-13079, and a shift from normal vaginal flora to other organisms
RR 00046.
Presented at the Annual Meeting of the Infectious Diseases Society
(Prevotella spp and Bacteroides spp, Gardnerella vaginalis,
for Obstetrics and Gynecology. Mobiluncus spp, anaerobic Gram-positive cocci, and
Reprints not available from the authors. genital mycoplasmas) is unknown.4-7 Studies of treatment
doi:10.1016/j.ajog.2005.01.068
Boggess et al 753
Table Gram stain and DNA hybridization results

Gram-stain interpretation
Normal flora Intermediate flora Bacterial vaginosis
(Nugent score 0–3) (Nugent score 4–6) (Nugent score 7–10) Total
DNA hybridization results* Negative 108 18 5 131
Positive 26 18 55 99
Total 134 36 60
* Positive defined as detection of at least 1 vaginal pathogen at 104 CFU/mL.
of bacterial vaginosis to prevent preterm birth have re- Material and methods
ported conflicting results,3,8-13 but a recent meta-analysis
affirmed the benefit of 1 week of oral antimicrobial As part of the Oral Conditions and Pregnancy Study,
therapy in women with asymptomatic bacterial vaginosis 1115 pregnant women were prospectively enrolled to
who have previously delivered preterm.14 These unre- characterize the inflammatory and infectious compo-
solved issues emphasize that further research on the nents of periodontal disease.21 Institutional Review
biology of bacterial vaginosis is needed. Board approval was granted to conduct the study, and
The diagnosis of bacterial vaginosis is usually based written informed consent obtained from subjects. All
on clinical criteria, including homogeneous vaginal women enrolled had vaginal specimens collected. A
discharge, an elevated vaginal pH, the presence of clue Dacron swab was placed in the posterior vaginal fornix,
cells, and an amine odor,15,16 but these clinical criteria smeared on a glass slide, and allowed to dry. A second
are imprecise for various reasons. Recent intercourse, swab of vaginal fluid was placed in a microcentrifuge
douching, menstruation, or contamination by cervical tube containing 0.15 mL TE buffer (10 mmol/L Tris-
mucus can affect vaginal pH.17 In addition, both amine HCl, 1 mmol/L EDTA, pH 7.6) and 0.15 mL of 0.5 mol/L
odor and characteristics of the vaginal discharge are NaOH, and frozen at ÿ80(C for microbial DNA
observer dependent and subjective criteria. Determina- analysis. Checkerboard DNA hybridization was com-
tion of clue cells on wet mount is also prone to pared to Gram stain using Nugent criteria for the
variability depending on the skill and training of the diagnosis of bacterial vaginosis among a random subset
reader, the quality of the microscope, and the adequacy of 230 (20%) enrolled women.
of the specimen.15,18 DNA hybridization was performed using a modifica-
To address these difficulties, Nugent et al19 proposed tion of the ‘‘checkerboard’’ DNA-DNA hybridization
a scoring system for the diagnosis of bacterial vaginosis method developed by Socransky.22 This is a quantitative
based on a Gram-stained vaginal smear. Nugent’s method for identifying microorganisms based on the
criteria are based on visual recognition of bacterial hybridization of whole genomic DNA-probes with
morphotype. A score is assigned to the vaginal flora as homologous DNA from bacterial samples. The sensitiv-
a weighted continuum from 0 to 10 based on large ity of the technique is 104 CFU/mL. Digoxigenin-
Gram-positive rods, small Gram-negative, or variable labeled, whole genomic DNA-probes were developed
rods, or curved rod morphotypes.19 The sensitivity of for Gardnerella vaginalis, Prevotella bivia, Bacteroides
Gram stain is significantly higher than clinical assess- ureolyticus, and Mobiluncus curtisii by extracting chro-
ment,17,20 and is a reliable means to diagnose bacterial mosomal DNA from pure cultures grown from Ameri-
vaginosis. However, microscopic evaluation by Gram can Type Culture Collection strains 14018, 29303, 33387,
stain requires special skills that might not be available to and 35241. Probes were labeled using the High-Prime
all practitioners or researchers, and does not yield labeling kit (Boehringer Mannheim, Indianapolis, Ind).
information about specific pathogens. Checkerboard Vaginal samples were thawed, boiled for 5 minutes,
DNA hybridization identifies presence of microbial and neutralized with 0.8 mL 5 mol/L ammonium
DNA by using hybridizing labeled probes. One advan- acetate. The samples were then deposited in parallel
tage over microscopy is the identification of specific lanes onto nylon membranes (Boehringer Mannheim)
microbes based on nucleic acids, with less subjectivity using a Minislot device (Immunetics, Cambridge, Mass),
than relying on bacterial morphotype recognition. Also, and immobilized by exposure to ultraviolet light. After 2
more than 1 pathogen can be studied on a single hours of prehybridization, the DNA probes were placed
specimen. Our objective was to determine whether on the membrane at right angles to the sample lanes
DNA hybridization of target vaginal pathogens is by means of a Miniblotter device (Immunetics), and
a useful tool to study shifts in vaginal flora typical of allowed to hybridize with the vaginal sample DNA over-
bacterial vaginosis among asymptomatic pregnant night at 42(C. After a series of stringency washes, hybrids
women. were detected by incubation with an anti-digoxigenin
754 Boggess et al
Figure Graph demonstrating proportion of vaginal specimens with 0, 1, or O2 target vaginal pathogens detected by DNA
hybridization stratified by Nugent score category.
antibody conjugated with alkaline phosphatase and the pathogens, and 2 for all 4 vaginal pathogens tested. The
addition of a chemiluminescent substrate (Boehringer most common pathogen detected by DNA hybridization
Mannheim). Signals were detected and quantitated was G vaginalis (35.1%), followed by P bivia (22.6%),
with the help of a Lumi-ImagerÔ workstation (Boeh- then B ureolyticus, and M curtisii (3% each). Specimens
ringer-Mannheim) by comparing the unknown signals with a Nugent score of 7 to 10 were more likely to have
to the ones generated by pooled bacterial standards G vaginalis detected by DNA hybridization than those
containing 106 of each of the 4 species. In this analysis, with a score of 0 to 3 (90% vs 9%, P ! .001). The
the signal of the high standard (106) was used as proportion of specimens demonstrating 0, 1, or O2
reference to calculate the bacterial load of the vaginal vaginal pathogens by DNA hybridization was signifi-
samples. Any sample generating a signal equivalent to cantly associated with Nugent score (Figure, P ! .001).
104 CFU/mL or greater was considered positive for Defining bacterial vaginosis as the presence of G vaginalis
the species. and 1 other anaerobe by DNA hybridization and com-
Gram-stain methodology and interpretation using paring with Nugent score detection of bacterial vaginosis
Nugent’s criteria19 was applied to stored vaginal smears (score 7–10), the sensitivity and specificity of the DNA test
by a single trained reviewer who was blinded to the were 60% and 89%, respectively.
DNA hybridization results. Results were analyzed using
McNemar’s paired test and chi-square test for trend,
with statistical significance set at P ! .05. Comment
Based on our findings, checkerboard DNA hybridiza-
Results tion is a useful method to detect multiple vaginal
pathogen targets. This technique allows detection of
Of the 230 women studied, 55% were black, 53% bacterial vaginosis by specific morphotype, which may
unmarried, 23% reported tobacco, and 24% alcohol assist in documenting the pathogen most often associ-
use during pregnancy, and 62% were nulliparous. ated with perinatal complications. The sensitivity of
Bacterial vaginosis was detected by Gram stain in 60 Gram stain is 105 organisms.23 A commercial DNA
(26.1%) of women. Vaginal pathogens were detected by hybridization test (AffirmÔ VP III, Microbial Identifi-
DNA hybridization in 99 (43.0%) of women. Gram- cation Test, Becton-Dickinson, Sparks, Md) has been
stain results were significantly associated with DNA developed and compared to Gram stain for the di-
hybridization detection of vaginal pathogens (Table, agnosis of bacterial vaginosis among symptomatic
chi-square test for trend, P ! .001). pregnant women. The automated probe system detects
Of the 99 specimens positive for vaginal pathogens by O2 ! 105 CFU/mL G vaginalis.24 However, the DNA
DNA hybridization, 53 (53.5%) were positive for at least probe was positive in 43% of women with lower
1 vaginal pathogen, 40 (40.4%) for 2 pathogens, 4 for 3 organism counts. If presence of microbial DNA at low
Boggess et al 755
levels proves to be associated with clinical symptoms or generalizable to all pregnant women. However, the
adverse reproductive outcomes, use of DNA hybridiza- detection rate of bacterial vaginosis by Gram stain was
tion may be useful in studying treatment response and comparable with other reports in pregnant women.1
understanding treatment failures. Furthermore, in our There is no follow-up information regarding develop-
population of asymptomatic women, a significant pro- ment of symptoms suggestive of bacterial vaginosis later
portion of women with intermediate flora by Nugent’s in the pregnancy, to determine if detection by DNA
criteria did not have 1 of the pathogens detected, and as hybridization predicts transition from asymptomatic to
the Nugent score increased the proportion of specimens symptomatic infection. We have not examined preg-
with 2 or more pathogens also increased, emphasizing nancy outcome data stratified by method of detection,
the intricate microbiology of asymptomatic bacterial and it is possible that detection by DNA hybridization
vaginosis in pregnancy and the potential utility of DNA does not correlate with clinically important infection.
testing to study shifts in vaginal flora. We recognize that these issues warrant further study
We believe that this method may prove useful to on a larger sample of women, and conclude that DNA
study vaginal flora shifts typical of bacterial vaginosis hybridization may be a useful technology to apply to the
but caution interpretation for clinical utility. This study study of bacterial vaginosis and its complications during
was not designed to test the DNA hybridization test as pregnancy.
a clinical diagnostic test. Reporting the sensitivity and
specificity of the DNA test implies that it was compared
with a ‘gold standard’ test. The only results that we have References
on the study women are Gram-stain results because
women were asymptomatic, and vaginal pH was not 1. Hillier SL, Nugent RP, Eschenbach DA, Krohn MA, Gibbs RS,
determined. There are many different criteria used and Martin DH, et al. Association between bacterial vaginosis and
accepted for the diagnosis of bacterial vaginosis, and in preterm delivery of a low- birth-weight infant. The Vaginal
Infections and Prematurity Study Group. N Engl J Med 1995;
the absence of a gold standard to compare the DNA 333:1737-42.
test, the sensitivity and specificity may be misinter- 2. McGregor JA, French JI. Bacterial vaginosis in pregnancy. Obstet
preted. It would be preliminary to suggest that the Gynecol Surv 2000;55:S1-19.
DNA test is appropriate for clinical uses as the clinical 3. Hauth JC, Goldenberg RL, Andrews WW, DuBard MB, Copper
implications of presence of DNA of vaginal pathogens RL. Reduced incidence of preterm delivery with metronidazole
and erythromycin in women with bacterial vaginosis. N Engl J
in absence of symptoms remains to be determined. The Med 1995;333:1732-6.
intent of our study was to demonstrate that vaginal 4. Kimberlin DF, Andrews WW. Bacterial vaginosis: association
pathogens typical of bacterial vaginosis are present at with adverse pregnancy outcome. Semin Perinatol 1998;22:242-50.
low levels, which may be underappreciated by the 5. Eschenbach DA, Davick PR, Williams BL, Klebanoff SJ, Young-
Nugent score. Smith K, Critchlow CM, et al. Prevalence of hydrogen peroxide-
producing Lactobacillus species in normal women and women with
Socransky et al developed the checkerboard method bacterial vaginosis. J Clin Microbiol 1989;27:251-6.
in 1994 to simultaneously hybridize large numbers of 6. Hillier SL, Krohn MA, Rabe LK, Klebanoff SJ, Eschenbach DA.
DNA samples.22 DNA segments are fixed in separate The normal vaginal flora, H2O2-producing lactobacilli, and
lanes on a membrane, which are then exposed at right bacterial vaginosis in pregnant women. Clin Infect Dis 1993;
angles to multiple labeled whole genomic DNA probes 16(Suppl 4):S273-81.
7. Hawes SE, Hillier SL, Benedetti J, Stevens CE, Koutsky LA,
against specific bacterial species. Hence, the multiplex Wolner-Hanssen P, et al. Hydrogen peroxide-producing lactoba-
assay allows for simultaneous detection of multiple cilli and acquisition of vaginal infections. J Infect Dis 1996;174:
organisms, which is the greatest advantage. However, 1058-63.
checkerboard DNA-DNA hybridization results take 3 8. Hillier SL, Lipinski C, Briselden AM, Eschenbach DA. Efficacy of
days to return, and an experienced laboratory is re- intravaginal 0.75% metronidazole gel for the treatment of bacterial
vaginosis. Obstet Gynecol 1993;81:963-7.
quired to perform this test and, thus, is more costly than 9. Klebanoff MA, Regan JA, Rao AV, Nugent RP, Blackwelder WC,
Gram stain. With these limitations the use of DNA Eschenbach DA, et al. Outcome of the Vaginal Infections and
hybridization may be best applied to research protocols Prematurity Study: results of a clinical trial of erythromycin
aimed at determining the natural history of bacterial among pregnant women colonized with group B streptococci.
vaginosis during pregnancy, response to therapy, or Am J Obstet Gynecol 1995;172:1540-5.
10. McGregor JA, French JI, Parker R, Draper D, Patterson E, Jones
reasons for conflicting results with treatment to prevent W, et al. Prevention of premature birth by screening and treatment
adverse pregnancy complications. for common genital tract infections: results of a prospective
While our findings are encouraging and suggest controlled evaluation. Am J Obstet Gynecol 1995;173:157-67.
usefulness of this DNA hybridization test for studying 11. McDonald HM, O’Loughlin JA, Vigneswaran R, Jolley PT, Harvey
asymptomatic bacterial vaginosis during pregnancy, our JA, Bof A, et al. Impact of metronidazole therapy on preterm birth
in women with bacterial vaginosis flora (Gardnerella vaginalis): a
study has limitations. Vaginal smear specimens were randomised, placebo controlled trial. BJOG 1997;104:1391-7.
randomly selected, and it is possible that our sample is 12. Carey JC, Klebanoff MA, Hauth JC, Hillier SL, Thom EA, Ernest
not representative of the entire study population, or not JM, et al. Metronidazole to prevent preterm delivery in pregnant
756 Boggess et al
women with asymptomatic bacterial vaginosis. National Institute 19. Nugent RP, Krohn MA, Hillier SL. Reliability of diagnosing
of Child Health and Human Development Network of Maternal- bacterial vaginosis is improved by a standardized method of gram
Fetal Medicine Units. N Engl J Med 2000;342:534-40. stain interpretation. J Clin Microbiol 1991;29:297-301.
13. Leitich H, Brunbauer M, Bodner-Adler B, Kaider A, Egarter C, 20. Tam MT, Yungbluth M, Myles T. Gram stain method shows
Husslein P. Antibiotic treatment of bacterial vaginosis in preg- better sensitivity than clinical criteria for detection of bacterial
nancy: a meta-analysis. Am J Obstet Gynecol 2003;188:752-8. vaginosis in surveillance of pregnant, low-income women in
14. Leitich H, Bodner-Adler B, Brunbauer M, Kaider A, Egarter C, a clinical setting. Infect Dis Obstet Gynecol 1998;6:204-8.
Husslein P. Bacterial vaginosis as a risk factor for preterm delivery: 21. Lieff S, Boggess K, Murtha A, Jared H, Moss K, Beck J, et al. The
a meta-analysis. Am J Obstet Gynecol 2003;189:139-47. Oral Conditions and Pregnancy study: peridontal status of a cohort
15. Amsel R, Totten PA, Spiegel CA, Chen KC, Eschenbach D, of pregnant women. J Periodontol 2004;75:116-26.
Holmes KK. Nonspecific vaginitis. Diagnostic criteria and micro- 22. Socransky SS, Smith C, Martin L, Paster BJ, Dewhirst FE, Levin
bial and epidemiologic associations. Am J Med 1983;74:14-22. AE. ‘‘Checkerboard’’ DNA-DNA hybridization. Biotechniques
16. Eschenbach DA, Hillier S, Critchlow C, Stevens C, DeRouen T, 1994;17:788-92.
Holmes KK. Diagnosis and clinical manifestations of bacterial 23. Krohn MA, Hillier SL, Eschenbach DA. Comparison of methods
vaginosis. Am J Obstet Gynecol 1988;158:819-28. for diagnosing bacterial vaginosis among pregnant women. J Clin
17. Mastrobattista JM, Bishop KD, Newton ER. Wet smear com- Microbiol 1989;27:1266-71.
pared with gram stain diagnosis of bacterial vaginosis in asymp- 24. Witt A, Petricevic L, Kaufmann U, Gregor H, Kiss H. DNA
tomatic pregnant women. Obstet Gynecol 2000;96:504-6. hybridization test: rapid diagnostic tool for excluding bacterial
18. Hillier SL. Diagnostic microbiology of bacterial vaginosis. Am J vaginosis in pregnant women with symptoms suggestive of
Obstet Gynecol 1993;169:455-9. infection. J Clin Microbiol 2002;40:3057-9.
www.ajog.org
Is zygosity or chorionicity the main determinant

of fetal outcome in twin pregnancies?
Stephen G. M. Carroll, MD,* Linda Tyfield, PhD, Louise Reeve, PhD,
Helen Porter, MD, Peter Soothill, MD, Phillipa Mm Kyle, MD
St Michael’s Hospital, Bristol, United Kingdom
Received for publication August 23, 2004; revised December 20, 2004; accepted January 11, 2005
KEY WORDS Objective: The purpose of this study was to examine whether fetal outcome in twin pregnancies is
Twin pregnancy dependent on zygosity or chorionicity.
Zygosity Study design: This was a prospective observational study comprised of women with twin
Chorionicity pregnancies who attended the fetal medicine unit at St Michael’s Hospital, Bristol, Ireland, during
the years 1998 to 2000 and who were delivered in hospitals in south west England. After delivery,
zygosity was determined with umbilical cord blood with the use of microsatellite markers that
were amplified by polymerase chain reaction. Placentae were examined histologically for
chorionic type. The perinatal outcomes of 3 groups of monozygotic monochorionic, monozygotic
dichorionic, and dizygotic pregnancies were compared with the use of the Mann-Whitney U test
and the Fisher’s exact test.
Results: All 92 dizygotic and 15 monozygotic dichorionic pregnancies resulted in live births. In 7
of the 39 cases in the monozygotic monochorionic group, either both twins were not live born or
delivery occurred !24 weeks of gestation. The gestational age at delivery and birth weight were
significantly lower, and there were a greater number of cases with birth weight discordancy of
O25% in the monochorionic pregnancies compared with the other 2 groups (P ! .05). There were
no significant differences in any of the study parameters between the monozygotic dichorionic
and dizygotic groups.
Conclusion: Fetal outcome in twin pregnancies is related to chorionicity rather than zygosity.
Two thirds of twin pregnancies are dizygotic, and one dichorionic placentation, depending on the timing of
third is monozygotic. All dizygotic twins are dichor- mitotic division after fertilization of the single egg.
ionic. Monozygotic twins may have monochorionic or Approximately 30% of monozygotic twins are dichor-
ionic; all twins with monochorionic and 10% with
dichorionic placentae are monozygotic.1,2 It has been
Supported by the Medical Research Committee of the Special
shown that fetal outcome is worse for monochorionic
Trustees for United Bristol Hospitals.
* Reprint requests: Stephen Carroll, MD, National Maternity
than dichorionic twins, which has been attributed
Hospital, Holles St, Dublin 2, Ireland. mainly to complications that are caused by placental
E-mail: scarroll@nmh.ie vascular anastomoses.3,4 However, it is unclear whether
doi:10.1016/j.ajog.2005.01.024
758 Carroll et al
absolute outcome is related to monochorionicity or monozygous. The placentae were examined macroscop-
monozygosity. The very few studies that have examined ically, and microscopic sections of the membranes were
outcome in relation to zygosity by comparing the 3 analyzed for the determination of the chorionic and
groups of dizygotic, monozygotic monochorionic, and amniotic type. Pathologic placental examination was
monozygotic dichorionic pregnancies were retrospective not necessary in twins who were discordant for sex
and have produced conflicting results.5-7 Furthermore because these were dizygotic. The 3 groups of mono-
only one of the studies determined zygosity by neonatal zygotic monochorionic, monozygotic dichorionic, and
DNA examination,6 which is a method that is more dizygotic pregnancies were compared.
accurate than testing by infant blood group.5,7 The In pregnancies in which delivery occurred after 24
purpose of this prospective study was to examine weeks of gestation and both twins were live born, the
whether outcome is dependent on zygosity or chorio- gestational age at delivery and birth weight distributions
nicity with zygosity being determined by DNA analysis were compared by the Mann-Whitney U test. Birth
with the use of highly polymorphic markers. weights were also expressed as a multiple of the SD of
birth weight for a given gestational age and sex.9 The
Fisher’s exact test was used to compare proportions (ie,
Material and methods percentages) between the groups. A probability value of
!.05 was considered significant.
This was a prospective study that examined fetal
outcome in relation to zygosity and chorionicity in
women with twin pregnancies who attended the fetal Results
medicine unit during the years 1998 to 2000. The women
were referred from St. Michael’s Hospital, Bristol, and There were 146 sets of twins of a total of 205 twins for
others throughout the south west of England for whom complete delivery details on outcome, gestation
ultrasound assessment of chorionicity in the first tri- at delivery, birth weight, zygosity, and chorionicity
mester, which was the subject of a previous publication were available; these twins were used for final analysis.
in this journal.8 The patients gave informed consent to In 44 of the 59 excluded cases, zygosity was not
enter the study that had been approved by the Hospital performed. In another 2 cases, although cord blood
Ethics Committee. After delivery, umbilical cord was examined, it was not possible to ascertain zygosity
blood and placentae were collected and sent from the because of maternal contamination of the cord blood
place of delivery to the analytic centers: Cord blood DNA. In 9 cases for which zygosity testing in same sex
went to the Department of Molecular Genetics, South- twins showed all to be monozygotic, placental histo-
mead Hospital, Bristol, for DNA extraction and zygos- logic condition was not determined; in 5 of these cases,
ity determination; and the placentae went to the histologic examination was impaired because of placen-
Department of Pathology, St Michael’s Hospital, for tal autolysis or damage to the membranes. In 4 cases of
diagnosis of chorionicity. twins with different sex, delivery details were not
Zygosity was determined with microsatellite markers available. There were 5 cases with congenital malfor-
that were amplified with the polymerase chain reaction. mations; 4 of these cases were in monozygotic preg-
Microsatellites are di-, tri-, or tetranucleotide repeat nancies, which included 2 cases of gastroschisis and
sequences that show variability in their repeat copy 1 case each with ventriculomegaly and anencephaly.
number. These sequences are scattered throughout the Three of these 4 cases had a monochorionic placenta.
genome and are found in non-coding regions. In this The fifth case was of a fetus with hydrocephalus in
study, 12 markers from 12 different chromosomes were a dichorionic pregnancy with undetermined zygosity.
amplified in 4 multiplex reactions. Individual markers There were 92 dizygotic twin pregnancies. In 51 of
were chosen because of their high degree of heterozy- these cases in which the twins were the same sex,
gosity and were combined in multiplex reactions classification was based on DNA microsatellite analysis;
based on similar reaction conditions for amplification, in all cases the placentae were classified histologically as
non-complementarity of primer sequences and non- dichorionic. In 41 cases, the twins were different sex and
overlapping repeat lengths in amplified products. Am- therefore dizygotic. All 39 monozygotic monochorionic
plified maternal and twin DNAs were run in consecutive pregnancies had histologic placental examination. In 23
lanes. Dizygosity was established when twins had cases, histologic results were concordant with DNA
inherited opposite maternal (or paternal) alleles at microsatellite results. In a further 16 cases in which
a minimum of 2 independent markers. Monozygosity microsatellite analysis was not performed, the diagnosis
was established with Bayes calculation to a likelihood of of monozygosity was based on the histologic diagnosis
95% with 5 informative markers and to O99% with R8 of monochorionicity. There were 15 monozygotic di-
informative markers. Calculations were based on a pre- chorionic sets that were diagnosed after microsatellite
vious probability of 0.3 that single sex twins are analysis and histologic placental examination.
Carroll et al 759
Table I Outcome in monochorionic twin pregnancies in cases in which both twins were not live born or in which delivery occurred at
!24 weeks of gestation
Gestational age
Cases Twin 1 Twin 2 at delivery (wk) Comment
1 Live birth Intrauterine death 34 Intrauterine death at 33 wk
2 Live birth Intrauterine death 27 Twin-to-twin transfusion in second trimester
5 Miscarriage Miscarriage 20 Twin-to-twin transfusion in second trimester
6 Intrauterine death Intrauterine death 32 Monoamniotic; intrauterine death at 32 wk
7 Neonatal death Neonatal death 23 Spontaneous labor
Table II Outcome in live infants who were born after 24 weeks of gestation in monozygotic monochorionic (MZMC), monozygotic
dichorionic (MZDC), and dizygotic (DZ) pregnancies
Outcome MZMC (n=32) MZDC (n=15) DZ (n=92) P value between groups
Median week of gestation at delivery (range) 35 (27-38) 37 (32-40) 37 (28-40) .02* (MZMC vs MZDC)
.02* (MZMC vs DZ)
.38 (MZDC vs DZ)
Delivery at !37 weeks of gestation (%) 59.4 46.7 48.9 .53 (MZMC vs MZDC)
.41 (MZMC vs DZ)
1.0 (MZDC vs DZ)
Delivery at !32 weeks of gestation (%) 18.7 0 7.6 .16 (MZMC vs MZDC)
.09 (MZMC vs DZ)
.59 (MZDC vs DZ)
Birth weight discordance (%)y 9.8 (0-42) 7.5 (3.3-20) 9.9 (0-38) .50 (MZMC vs MZDC)
.62 (MZMC vs DZ)
.59 (MZDC vs DZ)
Proportion with discordancy O25% (%) 15.6 0 3.3 .16 (MZMC vs MZDC)
.03* (MZMC vs DZ)
1.0 (MZDC vs DZ)
One fetus birth weight !3% (%) 18.7 0 14.1 .16 (MZMC vs MZDC)
.57 (MZMC vs DZ)
.21 (MZDC vs DZ)
Both fetuses birth weight !3% (%) 6.2 0 5.4 1.0 (MZMC vs MZDC)
1.0 (MZMC vs DZ)
1.0 (MZDC vs DZ)
* Statistically significant.
y
Range given in parentheses.
In the study population, all 92 dizygotic and 15 ionic group compared with the monozygotic dichorionic
monozygotic dichorionic pregnancies resulted in live and dizygotic groups. Additionally, the gestational age
births. In the monozygotic monochorionic group, 7 of at delivery and birth weight were significantly lower, and
the 39 cases were excluded from statistical analysis of there were a greater number of cases with birth weight
gestation at delivery and birth weight distributions discordancy of O25% in the monochorionic pregnan-
because either both twins were not live born or delivery cies. There were no significant differences in any of the
occurred at !24 weeks of gestation; outcome data in study parameters between the monozygotic dichorionic
these cases are summarized in Table I. Statistical and dizygotic groups.
analysis of outcome in the remaining 139 pregnancies The hypothesis that chorionicity rather than zygosity
is detailed in Table II. determines fetal outcome is supported by the finding in
our study that all 10 fetal losses were in the monochor-
Comment ionic group; 6 of these losses were related to twin-twin
transfusion. In another study that involved 300 preg-
These data show that fetal outcome in twin pregnancies nancies, the perinatal mortality rate was found to be
is related to chorionicity rather than zygosity because 16% in the monochorionic group, compared with 11%
there was a greater number of deaths in the monochor- and 1% in the dizygotic and monozygotic dichorionic
760 Carroll et al
groups, respectively; 88% of the deaths in the mono- Our method demonstrates a reliable way of assessing
chorionic group were related to twin-twin transfusion, zygotic type. Examination by blood group analysis has
and 3 of the 16 deaths in the dizygotic group were been used for many years. However, this method is less
related to congenital anomalies.6 In a much earlier study inaccurate because the segregation of numbers into the
of 293 sets of twins, the perinatal mortality rate was also monozygotic dichorionic group is based on population
higher in the monochorionic group at 11.2%, compared probabilities of same sex twins with the same blood
with 3.5% in the dizygotic and 7.7% in the monozygotic groups, which may lead to incorrect classification in
dichorionic groups; only 1 of the 15 deaths in the some cases.5,7,12 Additionally, some blood group anti-
monochorionic group was attributed to vascular anas- gens may be expressed weakly on some fetal cells, which
tomoses in the placenta.5 However, it is difficult to make could lead to errors in interpretation.13 Indeed, the most
definitive statements relating to outcome in this series recent and largest study to assess outcome in relation to
because it was published in the 1960s and 6 of the 9 zygosity and chorionicity used infant blood group
deaths in the monozygotic dichorionic group were analysis.7 In another study that analyzed outcome
related to low birth weight. The finding in our study according to zygosity with extracted DNA from umbil-
that the perinatal outcome is similar in dizygotic and ical cord or membranes and testing by restriction
monozygotic dichorionic pregnancies is supported by fragment length polymorphism analysis, it was not clear
a study that involved 1008 twin pregnancies.7 In that whether chorionicity had been determined by histologic
series, the perinatal mortality rate was 6.5% in the analysis.6
monochorionic group, compared with rates of 2.2% and In monozygotic pregnancies, heterokaryotypia may
2.6% in the dizygotic and monozygotic dichorionic preg- very rarely occur where a chromosomal abnormality
nancies, respectively; the causes of deaths were not given.7 occurs in 1 of the fetuses because of a post-zygotic non-
The relation between chorionicity and outcome is disjunction.2,14,15 Therefore, the finding of 2 different
strengthened by the finding that the monochorionic karyotypes does not necessarily indicate dizygosity, and
group had a higher proportion of twin pairs with birth DNA studies are recommended whenever there is doubt
weight discordancy of O25% compared with the other 2 about the chorionic status of the twins. Although
groups. This is in agreement with another study that ultrasound examination in the first trimester can reliably
showed that the incidence of weight discordancy was assess chorionicity, there may be doubt in a few cases,
8.6% in the monochorionic group and 7.3% and 5.4% in particularly if the diagnosis of chorionicity is based on
the dizygotic and monozygotic dichorionic groups, re- a second-trimester scan.8 This is important because, if
spectively.7 In our study, there was also a trend for an there is doubt, DNA or microsatellite analysis can be
increased rate of fetal growth restriction below the third used to indicate a dizygotic pregnancy and thus exclude
percentile in the monochorionic group, compared with monochorionicity. Selective termination of the affected
the other 2 groups; however, this did not reach statistical fetus can be carried out more safely if the twins are
significance. This finding was also reported in a previous dizygotic and, by definition, have 2 separate placentae.
study that examined the differences among the 3 groups.7 Indeed, molecular genetic prenatal determination of
Another study that compared monochorionic with twin zygosity with the use of DNA from chorionic villi
dichorionic pregnancies found that the incidence of fetal and amniocytes has been reported previously.16
growth restriction was significantly higher in monochor-
ionic pregnancies.3 Therefore, it is reasonable to assume
that significant fetal growth discordance results either Acknowledgments
from interfetal transfusion or placental insufficiency. The
finding that the median birth weight in the monochor- We thank all the medical and laboratory staff in the
ionic group was lower also reflects the earlier gestational Department of Paediatric Pathology for their assistance
age at delivery in this group; this trend is shown in Table in pathologic examination of the placentae.
II. Indeed, earlier gestation at delivery in monochorionic
pregnancies has been reported in other studies.3,6 The References
number of cases in our report is not large enough to
determine whether the incidence of congenital abnor- 1. Burn J, Corney G. Zygosity determination and the types of
twinning. In: MacGillivray I, Campbell D, Thompson B, editors.
mality is related to zygosity or chorionicity. The in- Twinning and twins. Chicester, UK: John Wiley; 1988. p. 7-25.
cidence of structural malformations is reported to be 2. Machin GA. Twins and their disorders. In: Reed GB,
higher in monozygotic than dizygotic twins, and some Claireaux AE, Cockburn F, editors. Diseases of the fetus and
anomalies may be explained partly by the process of newborn. 2nd ed. London: Chapman & Hall; 1995. p. 201-25.
embryo cleavage itself or secondary to an hypoxic insult 3. Sebire NJ, Snijders RJM, Hughes K, Sepulveda W, Nicolaides
KH. The hidden mortality of monochorionic twin pregnancies.
that is related to the vascular anastomes.10,11 Further BJOG 1997;104:1203-7.
research to examine the association between fetal defects 4. Snijder MJ, Wladimiroff JW. Fetal biometry and outcome in
and zygotic and chorionic type is required. monochorionic vs dichorionic twin pregnancies: a retrospective
Carroll et al 761
cross-sectional matched-control study. Ultrasound Med Biol 11. Jones JL. Monozygotic twinning and structural defects. In: Smith’s
1998;24:197-201. recognizable patterns of human malformation. 4th ed. Philadel-
5. Potter EL. Twin zygosity and placental form in relation to the phia: Saunders; 1988. p. 591-601.
outcome of pregnancy. Am J Obstet Gynecol 1963;87:566-77. 12. Fujikura T, Froehlich LA. Twin placentation and zygosity. Obstet
6. Machin G, Bamforth F, Innes M, McNichol K. Some perinatal Gynecol 1971;37:34-42.
characteristics of monozygotic twins who are dichorionic. Am J 13. Pollack MS, Heagney SD, Braaun D, O’Neill GJ. Technical and
Med Genet 1995;55:71-6. theoretical considerations in the HLA typing of amniotic fluid cells
7. Dube J, Dodds L, Armson A. Does chorionicity or zygosity for prenatal diagnosis and paternity testing. Prenat Diagn
predict adverse perinatal outcomes in twins? Am J Obstet Gynecol 1981;1:183-95.
2002;186:579-83. 14. Kaplowitz PB, Bodurtha J, Brown J, Speence JE. Monozygotic
8. Carroll SG, Abdel-Fattah S, Monteague I, Porter H, Soothill PW, twins discordant for Ullrich-Turner syndrome. Am J Med Genet
Kyle PM. Prediction of chorionicity in twin pregnancies at 10-14 1991;41:78-82.
weeks of gestation. BJOG 2002;109:182-6. 15. Nieuwint A, Van Zalen-Sprock R, Hummel P, Pals G, Van Vugt J,
9. Yudkin PL, Aboualfa M, Eyre JA, Redman CWG, Wilkinson AR. Van Der Harter H, et al. ‘‘Identical’’ twins with discordant
New birth weight and head circumference centiles for gestational karyotypes. Prenat Diagn 1999;19:72-6.
ages 24 to 42 weeks. Early Hum Dev 1987;15:45-52. 16. Kovacs B, Shahbahrami B, Platt LD, Comings DE. Molecular
10. Livingston JE, Poland BJ. A study of spontaneously aborted genetic prenatal determination of twin zygosity. Obstet Gynecol
twins. Teratology 1980;21:139-48. 1988;72:854-5.
www.ajog.org
Sonographic myometrial thickness predicts the latency

interval of women with preterm premature rupture of
the membranes and oligohydramnios
Catalin S. Buhimschi, MD,a,* Irina A. Buhimschi, MD,a Errol R. Norwitz, MD, PhD,a
Anna K. Sfakianaki, MD,a Benjamin Hamar, MD,a Joshua A. Copel, MD,a
George R. Saade, MD,b Carl P. Weiner, MDc
Yale University, New Haven, CT a; University of Texas Medical Branch, Galveston, TXb; and University of Maryland
School of Medicine, Baltimore, MD c
KEY WORDS Objective: Term labor is associated with global thinning of the myometrium. We hypothesized
PPROM that a thickened myometrium at the time of preterm premature rupture of membranes (PPROM)
Myometrium predicts less myometrial wall stress and, consequently, a longer latency interval.
Oligohydramnios Study design: Myometrial thickness was measured prospectively in 76 pregnant women enrolled
Ultrasound preterm in the following groups: PPROM (n = 28, mean [range], gestational age [GA]: 29.5 weeks [w]
birth [21.0 w-33.0 w]), preterm nonlabor control group (P-CTR), (n = 21, GA: 27.5 w [23.0 w-32.0 w])
and term nonlabor control (T-CTR) (n = 27, GA: 38.6 w [37.0 w-41.6 w]). All PPROM women
had oligohydramnios (AFI: 1.4 cm [0.0 cm-5.1 cm]). MT was measured ultrasonographically at
the midanterior, fundal, posterior, and lower uterine segment wall in cases and controls with an
intraoperator variability !10%.
Results: Women in the PPROM group displayed uniform thickness of the uterine body (mean G
SEM, anterior: 10.6 G 0.6 mm, fundal: 10.7 G 0.7 mm, posterior: 8.9 G 0.5 mm, P = .078).
At midanterior site the myometrium of the PPROM group was thicker compared to both P-CTR
(P ! .001) and T-CTR (P = .025) groups. This difference was preserved at the fundus (PPROM
vs P-CTR, P ! .001; PPROM vs T-CTR, P = .015). There was a positive correlation between
fundal MT and latency period (r = 0.43, P = 0.02) that persisted after adjusting for GA
(P = .04). A fundal MT less than 12.1 mm was 93.7% sensitive and 63.6% specific for the
identification of women whose latency period was less than 120 hours.
Conclusion: Significant thickening of the anterior and fundal walls of the uterus follows PPROM.
A thick myometrium in nonlaboring patients with PPROM is associated with longer latency
interval. Sonographic evaluation of MT may represent an alternative clinical tool for the
prediction of a short latency interval in women with PPROM.
Supported in part by a grant from National Institutes of Health IRO1 HD 047321-01 (IAB).
* Reprint requests: Dr Catalin Buhimschi, Yale University, Department of Obstetrics, Gynecology & Reprod. Sci., 333 Cedar Street, LLCI 804,
New Haven, CT 06520.
E-mail: catalin.buhimschi@yale.edu
doi:10.1016/j.ajog.2005.01.053
Buhimschi et al 763
Table I Maternal and sonographic characteristics of the study population

PPROM (n = 28) P-CTR (n = 20) T-CTR (n = 27) P value
Maternal variables
Age (y), mean G SEM 29.2 G 1.2 25.3 G 1.2 25.1 G 1.2 .028y
Gravidity, median (range) 3 (0-11) 3 (1-8) 3.5 (1-7) .912z
Parity: median (range) 2 (0-5) 1 (0-4) 1 (0-6) .887z
Maternal weight (kg), mean G SEM 90.8 G 5.9 82.1 G 3.9 92.3 G 4.3 .347y
Gestational age (wk), median (range) 29.5 (21.0-33.0) 27.5 (23.0-32.0) 38.6 (37.0-41.6) ! .001z
Sonographic variables
SEFW (kg), mean G SEM 1287.6 G 99.0 1192.7 G 110.2 3270.7 G 81.4 ! .001y
Placental thickness (cm), mean G SEM 4.1 G 0.3 3.7 G 0.2 3.7 G 0.2 .454y
AFI (cm), median (range) 1.7 (0.0-5.1) 11.8 (7.0-17.3) 9.8 (2.8-25.0) ! .001z
Abdominal thickness (mm), mean G SEM 21.8 G 2.5 17.6 G 1.9 14.0 G 2.2 .058y
Data was analyzed by one-way ANOVA (y), Kruskal-Wallis ANOVA (z). SEFW, Sonographically estimated fetal weight; AFI, amniotic fluid index.
Spontaneous rupture of the membranes is a normal that uterine wall stress (defined as applied force per unit
component of labor and delivery. Membrane rupture cross-sectional area of material) is directly proportional
before the onset of labor is considered premature to both the intracavitary pressure and the radius of the
(PROM), and induction of labor is common if the curvature, but inversely proportional to the thickness of
patient is at or close to term. Patient management the myometrium.16 Thus, the thicker the myometrium,
becomes more challenging when membrane rupture the lower the uterine wall stress. We hypothesized that
occurs preterm (PPROM), and in the absence of labor. a thick myometrium at the time of PPROM would be
The incidence of PPROM ranges from 2% to 20%, and associated with less myometrial wall stress and, conse-
is associated with 18% to 20% of perinatal deaths.1-3 quently, longer latency interval. We tested this hypothesis
Most women with PPROM deliver within 48 hours of by measuring MT by ultrasound scanning in patients with
rupture, but the neonatal impact and general outcome PPROM immediately following rupture.
depend largely on the gestational age (GA) at rupture.4,5
Though the physiologic explanation is obscure, the
interval from PPROM to delivery varies inversely with
GA at rupture.4 At less than 25 weeks’ (w) gestational Patients and protocol
age (GA), the average interval from rupture to delivery
(latency) is 11 days (d).6 Myometrial thickness (MT) was measured prospectively
Although scientists have long investigated ‘‘the timing in 76 pregnant women: PPROM (n = 28), preterm non-
of birth,’’ our understanding of the biological mecha- labor control (P-CTR, n = 21), and term nonlabor
nisms regulating the events that prevent and initiate labor control (T-CTR, n = 27). The Institutional Review
remains limited.7 Not surprisingly, any prediction of the Boards of Yale University and Wayne State University
latency interval for women with PPROM is imprecise. approved the study. We approached women admitted to
Amniotic fluid volume, GA, cervical length, and presence the Labor and Delivery ward and to the antepartum
of intra-amniotic markers of inflammation have various inpatient High Risk service at both institutions. All
prognostic values.8-10 Indeed, oligohydramnios is a risk women solicited for enrollment agreed to participate
factor for earlier delivery because abruption and in- and provided written informed consent. All women in
fection are each more common when amniotic fluid P-CTR group were recruited from the Ultrasound Unit at
volume is diminished.11,12 Women with PPROM and Yale University. Women were enrolled based on the
oligohydramnios at less than 25 w deliver earlier com- availability of one of the investigators (CSB), and all
pared to those with adequate amniotic fluid volume enrolled women were included in the final analysis. For
(pocket O2 cm).8 It is thus not surprising to find that the PPROM group, inclusion required PPROM with
85% of women with adequate amniotic fluid deliver singleton from 22 to 34 w GA. Exclusion criteria included:
beyond 25 w, and have much lower neonatal morbidity fetal anomalies, suspected fetal growth restriction
and mortality rates.8 Nevertheless, prophylactic therapy (IUGR) (sonographic fetal weight !10% percentile for
with broad-spectrum antimicrobial treatment (but no GA), abnormalities of placentation (low lying placenta,
tocolytic therapy) is also associated with longer latency abruptio placenta), uterine structural abnormalities,
interval than placebo.13 cervical cerclage, previous uterine scar. Management of
Similar to the myocardium, the force of labor is uterine the patients was left up to the treating team. All patients
wall tension opposed to the resistance of the cervix, except one (22 w GA) received corticosteroids for
perineum, and pelvis.14,15 Mathematical modeling reveals lung maturity and antibiotics per PPROM protocol
764 Buhimschi et al
Figure 1 Representative ultrasonographic image of the anterior wall of the uterus with myometrial thickness measurement marked
between calipers in a woman with intact membranes (A) (GA = 27.2 weeks) and preterm premature rupture of membranes
(GA = 27.4 weeks) (B). The anterior wall of the uterus (arrows) was thicker in PPROM woman; cm, Centimeters; mm, millimeters.
(ampicillin/erythromycin). In the absence of signs or identified before evaluation of fundal MT and used as
symptoms of chorioamnionitis (fever over 100.4(F, a reference for all subsequent measurements. Measure-
abdominal tenderness, fetal tachycardia), and/or abnor- ment of the posterior uterine wall was technically the
malities of fetal heart rate (variable or late decelerations), most challenging. We demarcated the posterior wall
and/or abruption, PPROM was managed expectantly. using pulsations of the maternal abdominal aorta as
The diagnosis of PPROM was confirmed by visualiza- anatomic marker. Each measurement was made from
tion of amniotic fluid ‘‘pooling’’ through the cervical os separate scan images. At least 3 measurements were
during speculum examination, ‘‘nitrazine,’’ ‘‘ferning,’’ or obtained at each site and averaged. The investigator
amniocentesis-dye positive tests. Tocolysis and/or digital (CSB) was not aware of the previous numeric evaluation
exams were not permitted. Patients received corticoste- of MT in between measurements. Previous experience
roids for lung maturity if less than 32 w GA, and antibiotic demonstrated no differences in MT among uterine wall
therapy (ampicillin/erythromycin or clindamycin in the sites.14 One investigator (CSB) conducted all ultrasound
event of allergy to penicillin). Women were monitored examinations and made all measurements. The intra-
by cardiotocography at least twice daily for the presence observer coefficient of variation for assessment of MT
of fetal heart abnormalities and/or uterine contractions. varied from 8% to 10% at each uterine site. The
The ultrasound examination was performed within 12 placental thickness was measured at the site of umbilical
hours of PPROM. An abdominal ultrasound survey cord insertion. Abdominal wall thickness was estimated
(Acuson, Sequoia 512, Acuson Corporation, Mountain at the same site used to evaluate thickness of the
View, Calif [WSU] and/or Voluson 730 Expert, General midanterior uterine wall.
Electric, Milwaukee, Wis [Yale]) was performed using PPROM women were managed expectantly, and
a 5.0 or 7.5 MHz transabdominal probe. The amniotic underwent serial evaluations of fetal well-being up to
fluid index (AFI) was measured using the 4-quadrant delivery (spontaneously or for clinical indications con-
technique.17 Oligohydramnios was defined as an AFI sistent with chorioamnionitis or abruption). The latency
less than 5 cm.17 The myometrium was sonographically interval was defined as the time period (days or hours)
identified as the echo homogeneous layer between the from the time of membrane rupture reported by the
serosa and the decidua. The MT was measured at 4 patient to delivery. None of the PPROM women were
different sites: lower segment (LUS) (approximately 2 delivered for topics unrelated to PPROM (elective
cm above reflection of the urinary bladder), midanterior induction at 34-35 w, preeclampsia, or other medical
wall (with the scan probe 1 cm above the maternal complications of pregnancy).
umbilicus), fundus, and posterior walls of the uterus.14
Thickness of the fundus was measured by placing the Statistical analysis
scan probe above the uterine fundus so that the entire
curvature of the uterus was visualized. To assure All data sets were subjected to normality testing using
consistency in the anatomic site, aortic pulsations were the Kolmogorov-Smirnov method. The data are reported
Buhimschi et al 765
Figure 2 Mean G SEM of MT as measured at the level of the anterior (A), fundus (F), posterior (P), and lower uterine segment
(L) wall of the uterus of nonlaboring women with intact membranes evaluated at term (T-CTR) (n = 27) and preterm (P-CTR)
(n = 21), as well as those whose pregnancies were complicated by preterm premature rupture of membranes (PPROM) (n = 28).
Values are represented as log (mm).
as mean and standard error of the mean (SEM) (for A P ! .05 was considered to indicate statistically sig-
normally distributed data), or as median and range (for nificant difference.
skewed data). Continuous normally distributed data
were compared using one-way analysis of variance
(ANOVA) or one-way repeated measures ANOVA fol-
lowed by post-hoc Student-Newman–Keuls tests as ap-
Results
propriate. Categorical data sets or data without normal Characteristics of women at enrollment
distribution were compared using Kruskal-Wallis
ANOVA on ranks followed by Dunn’s tests. Statistical Table I presents a series of demographic and ultraso-
analysis of all MT data sets was completed after nographic variables assessed at enrollment. Women with
logarithmic transformation to obtain a normal distribu- PPROM were significantly older compared with those in
tion (one-way ANOVA). The effect of PPROM on MT the P-CTR and T-CTR groups (Student-Newman–
at different uterine sites was determined using two-way Keuls P ! .05). There were no significant differences
repeated measures ANOVA. Multivariate analysis with among groups in terms of gravidity (Kruskal-Wallis
linear regression model was applied to identify any ANOVA, P = .912), parity (P = .887), or maternal
significant associations between maternal, fetal, or labor body weight (P = .347).
characteristics as independent variables and MT as the Women with PPROM and P-CTR had similar GA
dependent variable. A Pearson product moment corre- (median [range] PPROM: 29.5 w [21.0 w-33.0 w] vs
lation was used to measure colinearity between the P-CTR: 27.5 w [23.0 w-32.0 w], Dunn’s P O .05) and
selected independent variables, as well as other relevant sonographic estimated fetal weights (SEFW) (mean G
correlations between dependent and independent varia- SEM, PPROM: 1287.6 G 524 g vs P-CTR: 1192.7 G
bles. Receiver operating characteristics (ROC) curve 505 g, Student-Newman–Keuls, P = .500). The GA and
analysis was performed using MedCalc (Broekstraat, SEFW were significantly different from T-CTR group
Belgium) statistical software. Survival analysis was (GA: 38.6 w [37.0-41.6], Dunn’s P ! .001; SEFW: 3270.0
performed using GraphPad Software (San Diego, Calif). G 81.4 g; Student-Newman–Keuls P ! .001). PPROM
766 Buhimschi et al
Figure 3 Graphic representation of myometrial thickness of the anterior, fundus, posterior, and lower uterine segment wall of the
uterus of nonlaboring women with intact membranes evaluated at term (T-CTR) and preterm (P-CTR) (n = 21), as well as those whose
pregnancies were complicated by preterm premature rupture of membranes (PPROM). MT values are represented as mean G SEM.
Significance between sites and groups is presented in the legend. P ! .05 was considered to indicate statistically significant difference.
women had significantly lower AFIs compared to both significantly thicker than LUS (4.7 G 0.5 mm, Student-
control groups (one-way ANOVA P ! .001). There was Newman–Keuls, P ! .001). Similarly, MT assessment in
no significant difference in the AFI between P-CTR and the PPROM group revealed uniform thickness at each site
T-CTR groups (Dunn’s P O .05). Oligohydramnios was of the uterine body (anterior: 10.6 G 0.6 mm, fundal: 10.7
present in 96.4% (27/28) of PPROM patients vs 0% of G 0.7 mm, posterior: 9.0 G 0.6 mm, Student-Newman–
P-CRL (0/21) and 22.2% of T-CTR (6/27) (chi-square Keuls, P = .078), although the LUS was thinner in
P ! .001). The sonographic estimated abdominal wall PPROM women compared to the other sites (7.7 G 0.7
thickness did not differ among groups (one-way ANOVA, mm, Student-Newman–Keuls, P ! .001) (Figures 2 and
P = .06), and correlated directly with maternal body 3). In the P-CTR group the differences in MT between
weight (Pearson r = 0.371, P = .002). sites including LUS did not reach statistical significance
(anterior: 7.2 G 0.4 mm, fundal: 7.7 G 0.4 mm, posterior:
Sonographic estimated myometrial 7.4 G 0.4 mm, LUS: 6.2 G 0.5 mm, one-way ANOVA,
thickness (MT) one-way repeated measurements ANOVA, P = .06).
Among groups, MT was significantly thicker at
Figure 1 illustrates representative ultrasound images of midanterior site in the PPROM group compared with
the anterior uterine wall of a woman in the P-CTR both P-CTR (one-way ANOVA, P ! .001) and T-CTR
(Figure 1A) and PPROM (Figure 1B) groups. Both (P = .02) (Figure 3). This difference was maintained at
women had similar GA at MT assessment (27 w). the fundal site (PPROM vs P-CTR, P ! .001; PPROM
Sonographic evaluation of the myometrial wall at term vs T-CTR, P = .01). The posterior wall site was only
(T-CTR) demonstrated that MT for each patient was marginally thicker in PPROM women compared with
uniform between uterine body sites (mean G SEM, both control groups (P = .05). LUS was significantly
anterior: 8.8 G 0.5 mm, fundal: 8.6 G 0.4 mm, posterior: thicker in PPROM compared with P-CTR (P = .04)
8.2 G 0.3 mm, one-way ANOVA, Student-Newman– and T-CTR (P = .003) women. MT of the LUS at term
Keuls, P = .557). At term all uterine body sites were was not different from P-CTR (P = .07).
Buhimschi et al 767
Figure 4 (A) Scattergram of the myometrial thickness versus latency period in women whose pregnancies were complicated by
preterm premature rupture of membranes (PPROM). R, regression line; PI, prediction interval (confidence interval for the
population). (B) Survival analysis of labor duration in women whose pregnancies were complicated by PPROM according to a cut-
off in fundal MT above 12.1 mm.
To evaluate global differences of MT, we performed significantly related (r = 0.533, P = .001) to maternal
a two-way repeated measures ANOVA and found weight (P = .01) and membrane status (P = .02), while
a significant difference in MT among groups (P ! .001, the anterior and posterior MT were best determined using
power = 0.975) and sites (P ! .001, power = 0.999). a combination of membrane status and GA (anterior:
There was a significant interaction among groups and r = 0.597, P ! .001, posterior: r = 0.418, P = .02). There
sites (P = .049, power = 0.445). was no significant colinearity among the final parameters
in the model.
Outcome characteristics Table II lists the sensitivities and specificities of fundal
MT measurement for the prediction of the latency
Women with PPROM had a median GA at delivery of interval with the optimum cut-offs (value corresponding
30.6 w (range: 24.3 w-34.4 w) with a median latency of 4 with the highest accuracy, ie, minimal false-negative and
days (d) (range: 0.16 d-31.2 d). Twenty-nine percent of false-positive results as reported by MedCalc) in predict-
PPROM women were delivered by cesarean section. One ing delivery from 48 to 168 hours. A MT less than 12.1
of the PPROM women was induced with misoprostol at mm was 93.7% sensitive for the identification of women
22 w following neonatology consult, and was excluded whose latency period would be less than 120 hours (h),
from the latency analysis. There was a direct correlation and 63.6% specific for the identification of women whose
between fundal MT and latency period (r = 0.52, P ! latency period would be higher than 120 h. A survival
.005, n = 27 (Figure 4A). The median GA at delivery for analysis of the delivery interval for PPROM group
the P-CTR group was 39 w (range: 36.2 w-41.1 w). The demonstrated that women with MT above 12.1 mm
median time between MT assessment and delivery for the had longer latency intervals than those with MT less than
P-CTR group was 68 d (range 41 d-103 d) and for the T- 12.1 mm at enrollment (MT !12.1 mm: latency 72 h vs
CTR group was 8 d (range: 1 d-23 d). MT O12.1 mm: latency 228 h) (log rank chi-
We modeled MT as dependent variable against AFI, square = 8.412, P = .003) (Figure 4B).
gestational age, placental thickness, SEFW, gravidity,
parity, maternal weight, maternal abdominal wall thick- Comment
ness, and membrane status (ruptured = 1, intact = 0) as
independent variables to determine the possible impact of We demonstrate in the present investigation that uterine
external factors on the MT measurement. This analysis was wall thickness is altered in women with PPROM, and
limited to the patients in the PPROM and P-CTR groups correlates with latency interval. This finding has both
(n = 49). Regression analysis suggested fundal MT is clinical and physiologic implications.
768 Buhimschi et al
biochemical tests.28 Cervical length measurement after

Table II Sensitivity and specificity of myometrial thickness
in women with preterm premature rupture of membranes PPROM may also be useful for predicting preterm birth,
stratified by duration of latency period as the risk of ascending infection remains low.9 Un-
fortunately, the modest sensitivity with high specificity
Sensitivity Specificity MT cut-off
of cervical length evaluation may reflect the fact there
Latency % % (mm)
are several different patterns of ‘‘normal’’ change in
!48 h 55.6 88.9 !8.1 cervical length. These patterns may vary from a gradual
!72 h 80 58 !9.7
to an accelerated change or even a precipitous decrease
!96 h 91.7 53.3 !10.3
!120 h 93.7 63.6 !12.1
in cervical length at term.29
!144 h 93.7 63.6 !12.1 The clinical management after PPROM is complicated
!168 h 89.5 62.5 !12.6 by the absence of a gold standard method to predict
pathogenic processes leading to parturition.26 Our un-
derstanding of the mechanisms that determine the length
of the latency interval after PPROM is hindered by the
PPROM and associated preterm delivery are consid- fact that the human myometrium and cervix appear to
ered the leading causes of perinatal morbidity and have redundant and parallel mechanisms to ensure
mortality in the US.18,19 Clinically, the GA at PPROM, adequate length of gestation.30 Furthermore, the impact
SEFW, fetal presentation, fetal lung maturity, absence of pregnancy and labor on the uterus and cervix differ
of intra-amniotic inflammation, degree of cervical di- greatly.31,32 The prevailing theories surrounding PPROM
latation, and state of myometrial contractility are latency interval may well overestimate the importance of
carefully evaluated before deciding on a course of the cervix, leaving the role played by myometrial activa-
management. In the absence of clinical symptoms or tion largely unexplored. Markers with prognostic value in
laboratory signs of chorioamnionitis, the management predicting the latency interval (chorionic-decidual and
of a pregnancy with PPROM is usually expectant, based myometrial cell activation) would be beneficial as aides to
on the assumption that even a minor delay in the clinical management, as well as to enhance our under-
interval to delivery will be beneficial to the fetus.20 standing of the mechanisms triggering preterm labor
Even though investigators have searched for factors contractions and PPROM.26
that predict the onset of preterm labor, thickness of the Our previous sonographic observation14 that the
myometrium following PPROM has never been tested. myometrium thins symmetrically during active labor
Digital cervical examination, home uterine monitoring with the least amount of thinning at the uterine fundus
of uterine contractility, and thickness of LUS have each stimulated us to rethink the mechanisms responsible for
been studied.21-24 The digital cervical examination and the uniform dispersion of the contractile forces that
frequency of uterine contractions have weak prognostic ensure efficient fetal expulsion. Consistent with our
values.24 Not only are digital cervical examination of previous report, we now demonstrate that women with
women with PPROM and frequency of uterine contrac- spontaneous PPROM and in the absence of myometrial
tions poorly predictive, but a digital exam may actually activation have a thicker anterior and fundal wall
increase the risk of ascending infection.25 Conversely, compared with women who have intact membranes.
vaginal bleeding, risk scoring schemes, and fetal breath- Sudden decompression of the uterine sac, which had
ing activity are also predictive of the onset of labor, but been filled with a minimally compressible fluid that
either have poor sensitivity and specificity, or are normally opposed thickening, is the most likely physi-
accurate only at late stages in the pathogenic process.26 ologic explanation.33 We assume that women with
Despite being ineffective, many of the previously listed a long latency interval after spontaneous PPROM are
prediction strategies are widely used in the clinical in a state of myometrial quiescence or incomplete
practice. The most recent efforts to estimate the pre- myometrial activation, and demonstrate that the long
dictive value of LUS thickness in women with intact latency and presumed myometrial quiescence are asso-
membranes also proved to be unsuccessful.23 ciated with a greater thickness of the anterior and fundal
There has been much attention focused on the sono- wall myometrium. These observations are consistent
graphic assessment of cervical length since shortening is with previous interpretations that the mechanisms un-
associated with an increased risk of preterm delivery in derlying physical disruption of amniochorion integrity
both nulliparous and multiparous women.21,22,27 The are complex, and collagenolytic activation of matrix
preterm delivery prediction study conducted by the metalloproteinases can occur in the absence of uterine
NICHD Maternal Fetal Medicine Unit Network con- contractility (myometrial activation).34 It is possible that
cluded that the most powerful factors associated with those women with PPROM and thin myometrium
preterm birth before 32 w are a positive fetal fibronectin already experienced functional complete myometrial
test and a cervical length less than 10th percentile either activation that allows for coordinated tone, contrac-
alone or in combination with other maternal serum tions, and shorter latency interval.
Buhimschi et al 769
Sonographic evaluation of cervical length in women 4. Johnson JW, Daikoku NH, Niebyl JR, Johnson TRB Jr, Khou-
with PPROM is reported to have maximum sensitivities zami VA, Witter FR. Premature rupture of the membranes and
prolonged latency. Obstet Gynecol 1981;57:547-56.
and specificities of 63% and 81%, respectively.29 We 5. Carroll SG, Papaionnou S, Nicolaides KH. Preterm labor amnior-
find that the sonographic measurement of fundal MT rhexis: outcome of live births. Obstet Gynecol 1995;86:18-25.
less than 8.1 mm has a similar sensitivity and specificity 6. Morales WJ, Talley T. Premature rupture of membranes less than
(55.6%, respectively, 88.9%). However, we further deter- 25 weeks. A management dilemma. Am J Obstet Gynecol
mined that a MT 12.1 mm or more is 93.7% sensi- 1993;168:503-7.
7. Norwitz ER, Robinson JN, Challis JR. The control of labor.
tive and 63.6% specific for the prediction of a latency N Engl J Med 1999;34:2098-9.
period longer than 120 h. Unfortunately, no cervical 8. Hadi HA, Hodson CA, Strickland D. Premature rupture of the
length data are currently available for comparison at membranes between 20 and 25 weeks of gestation. Role of
120 h. As a corollary to these findings, survival analysis amniotic fluid volume in perinatal outcome. Am J Obstet Gynecol
revealed that a thickened myometrium in nonlaboring 1994;170:1139-44.
9. Gire C, Faggianelli P, Nicaise C, Shojai R, Fiori A, Chau C, et al.
women with PPROM was associated with latency Ultrasonographic evaluation of cervical length in pregnancies
longer than 120 h. This is consistent with our previous complicated by preterm premature rupture of membranes. Ultra-
report demonstrating that only active myometrial con- sound Obstet Gynecol 2002;19:565-9.
tractility is associated with widespread thinning of the 10. Buhimschi IA, Christner R, Buhimschi CS. Proteomic biomarker
myometrium independent of ROM,14 and explains analysis of amniotic fluid for identification of intra-amniotic
inflammation. BJOG 2005;112:173-81.
why nonactive laboring women (thick myometrium) 11. Vintzileos AM, Campbell WA, Nochimson DJ, Weinbaum PJ.
have longer latency periods than those with MT less Preterm premature rupture of the membranes: a risk factor for the
than 12.1 mm. development of abruptio placentae. Am J Obstet Gynecol
Given the likely heterogeneity in the causes of 1987;156:1235-8.
preterm labor, our present and previous reports raise 12. Vermillion ST, Kooba AM, Soper DE. Amniotic fluid index values
after preterm premature rupture of the membrane and perinatal
more questions. We have insufficient data at this time to infection. Am J Obstet Gynecol 2000;183:271-6.
determine how MT changes longitudinally over the 13. Mercer BM, Miodovnik M, Thurnau GR, Goldenberg RL, Das
course of the latency period in women who will undergo AF, Ramsey RD, et al. Antibiotic therapy for reduction of infant
spontaneous onset of uterine contractions. Further, we morbidity after preterm premature rupture of the membranes. A
still do not know the appropriate method to predict randomized control trial. National Institute of Child and Human
Development Maternal–Fetal Medicine Units Network. JAMA
latency in women with PPROM. Studies combining 1997;278:989-95.
cervical length and MT sonography, fetal fibronectin, 14. Buhimschi CS, Buhimschi IA, Malinow AM, Weiner CP.
proteomic analysis of the amniotic fluid at the time of Myometrial thickness during human labor and immediately
PPROM, and development of highly sensitive noninva- postpartum. Am J Obstet Gynecol 2003;188:553-9.
sive uterine contraction monitoring methods are war- 15. Veille JC, Hosenpud JD, Morton MJ, Welch JE. Cardiac size and
function in pregnancy induced hypertension. Am J Obstet Gynecol
ranted. Transabdominal ultrasound evaluation of MT 1984;150:443-9.
and surface electromyographic analysis of uterine con- 16. Deyer TW, Ashton-Miller JA, Van Baren PM, Pearlman MD.
tractions remain the only noninvasive methods to Myometrial contractile strain at the uteroplacental separation
evaluate choriodecidual myometrial activation.26,31 during parturition. Am J Obstet Gynecol 2000;183:156-9.
While transabdominal sonography is unsatisfactory for 17. Phelan JP, Ahn MO, Smith CV, Rutherford SE, Anderson E.
Amniotic fluid index measurements during pregnancy. J Reprod
cervical evaluation, it is well accepted by the patients for Med 1987;32:601-4.
MT evaluation.35 Sensitive biochemical assays for b- 18. Garite TJ. Premature rupture of the membranes: the enigma of the
human chorionic gonadotropin (b-hCG) hormone, cy- obstetrician. Am J Obstet Gynecol 1985;151:1001-5.
tokines, and corticotrophin releasing hormone (CRH), 19. Berkowitz GS, Papiernik E. Epidemiology of preterm birth.
as well as serial evaluation of vaginal amniotic fluid Epidemiol Rev 1993;15:414-43.
20. Daikoku NH, Kaltreider DF, Johnson TR Jr, Johnson JW,
combined with cervical length and MT sonography, may Simmons MA. Premature rupture of membranes and preterm
provide the context required for a reassessment of the labor: neonatal infection and perinatal mortality risks. Obstet
mechanisms responsible for early or delayed delivery of Gynecol 1981;58:417-25.
the fetus. 21. Iams JD, Goldenberg RL, Meis PJ, Mercer BM, Moawad A, Das
A, et al. The length of the cervix and the risk of spontaneous
premature delivery. National Institute of Child Health and Human
Development Maternal Fetal Medicine Unit Network. N Engl J
References Med 1996;334:567-72.
22. Owen J, Iams JD, Hauth JC. Vaginal sonography and cervical
1. Gunn GC, Mishell DR Jr, Morton DG. Premature rupture of the incompetence. Am J Obstet Gynecol 2003;188:586-96.
membranes: a review. Am J Obstet Gynecol 1970;106:469-83. 23. Yost NP, Owen J, Berghella V, MacPherson C, Swain M, Dildy
2. Cox SM, Williams ML, Leveno KJ. The natural history of preterm GA, et al. Second trimester cervical sonography: features other
rupture of the membranes. What to expect of expectant manage- than cervical length to predict spontaneous preterm birth. Obstet
ment. Obstet Gynecol 1988;71:558-62. Gynecol 2004;103:457-62.
3. Naeye RL. Causes of perinatal mortality in the US Collaborative 24. Iams JD. Prediction and early detection of preterm labor. Obstet
Perinatal Project. JAMA 1977;238:228-9. Gynecol 2003;101:402-12.
770 Buhimschi et al
25. Lewis DF, Major CA, Towers CV, Asrat T, Harding JA, Garite 31. Pritchard JA, MacDonald PC. Maternal adaptation to pregnancy.
TJ. Effects of digital examinations on latency period in preterm In: Williams obstetrics. 16th ed. New York: Appleton-Century-
premature rupture of membrane. Obstet Gynecol 1992;80:630-4. Crofts; 1980. p. 221-59.
26. Lockwood CJ. The diagnosis of preterm labor and the prediction 32. Buhimschi IA, Ali M, Jain V, Chwalisz K, Garfield RE.
of preterm delivery. Clin Obstet Gynecol 1995;38:675-87. Differential regulation of nitric oxide in the rat uterus and
27. Welsh A, Nicolaides K. Cervical screening for preterm delivery. cervix during pregnancy and labor. Human Reprod 1996;11:
Curr Opin Obstet Gynecol 2002;14:195-202. 1755-66.
28. Goldenberg RL, Iams JD, Mercer BM, Meis P, Moawad A, Das 33. Halliday D, Resnick R, Walker J. Fluids. In: Extended funda-
A, et al. What we have learned about the predictors of preterm mentals of physics. 6th ed. New York: John Wiley & Sons; 2000.
birth. Semin Perinatol 2003;27:185-93. p. 321-45.
29. Bergelin I, Valentin L. Patterns of normal change in cervical length 34. Buhimschi IA, Kramer WB, Buhimschi CS, Thompson LP, Weiner
and width during pregnancy in nulliparous women: a prospective, CP. Reduction-oxidation (redox) state regulation of matrix metal-
longitudinal ultrasound study. Ultrasound Obstet Gynecol loproteinase activity in human membranes. Am J Obstet Gynecol
2001;18:217-22. 2000;182:458-64.
30. Buhimschi CS, Buhimschi IA, Malinow A, Saade GR, Garfield 35. Andersen HF, Nugent CE, Wanty DS, Hayashi RH. Prediction of
RE, Weiner CP. The forces of labor. Fetal Matern Med Rev risk of preterm delivery by ultrasonographic measurement of
2003;14:273-307. cervical length. Am J Obstet Gynecol 1990;163:859-67.
www.ajog.org
Use of over-the-counter medications during pregnancy

Martha M. Werler, ScD,a,* Allen A. Mitchell, MD,a Sonia Hernandez-Diaz, MD, DrPH,a
Margaret A. Honein, PhD,b and the National Birth Defects Prevention Studyb
Slone Epidemiology Center at Boston University, Boston, MAa; Division of Birth Defects and Developmental
Disabilities, Centers for Disease Control and Prevention, Atlanta, GAb
Received for publication October 18, 2004; revised February 4, 2005; accepted February 17, 2005
KEY WORDS Objective: The most common medications used in pregnancy are nonprescription or over-the-
Pregnancy counter medications, although there has been little research on their risks or safety. We describe
Medication the patterns of over-the-counter medication use among pregnant women.
Epidemiology Study design: Data were collected in 2 case-control studies of birth defects: the Slone
Epidemiology Center Birth Defects Study (BDS) and the National Birth Defects Prevention
Study (NBDPS).
Results: Among 7563 mothers of malformed and nonmalformed offspring in the Slone
Epidemiology Center Birth Defects Study and 2970 mothers of nonmalformed offspring in the
National Birth Defects Prevention Study, acetaminophen, ibuprofen, and pseudoephedrine were
used by at least 65%, 18%, and 15%, respectively. Among women in the Slone Epidemiology
Center Birth Defects Study, the use in pregnancy of aspirin and chlorpheniramine decreased from
1976 to 2004 and of ibuprofen, pseudoephedrine, diphenhydramine, dextromethorphan, and
guaifenesin increased. Among women in the National Birth Defects Prevention Study, the use of
acetaminophen, pseudoephedrine, diphenhydramine, and guaifenesin was higher during preg-
nancy than before pregnancy.
Conclusion: Findings show that over-the-counter medications are used by most pregnant women.
Studies that examine specific over-the-counter medications in relation to specific birth defects are
necessary to better inform pregnant women about risks and safety.
Ever since the thalidomide tragedy, there has been agulant warfarin,1 the anticonvulsant valproic acid,2
concern that medication use in pregnancy can cause and the acne medication isotretinoin.3 Because each of
adverse fetal outcomes. Indeed, other medications have these medications requires a prescription, the identifica-
since been identified as teratogens, such as the antico- tion of associated fetal effects was facilitated by medical
record or pharmacy documentation. For over-the-
counter (OTC) medication use in pregnancy, the only
Supported by a cooperative agreement from the Centers for confirmed association is between late pregnancy aspirin
Disease Control and Prevention. use and intracranial hemorrhage in the newborn infant.4
* Reprint requests: Martha M. Werler, ScD, Slone Epidemiology
Center at Boston University, 1010 Commonwealth Ave, Boston, MA
There is a dearth of studies on OTC drugs, in part
02215. because approaches to the study of their potential fetal
E-mail: mwerler@slone.bu.edu effects are complicated by the lack of a paper trail. In
doi:10.1016/j.ajog.2005.02.100
772 Werler et al
Table Prevalence of specific OTC medication use in pregnancy
BDS* NBDPSy Trimester (%)

Medication Pregnancy (%) Pregnancy (%) Prepregnancy (%)z First Second Third
Analgesic 76.1 70.4 56.9 59.3 46.5 51.5
Acetaminophen 69.8 65.5 47.6 54.2 50.5 48.0
Ibuprofen 24.8 18.4 21.1 16.0 8.2 8.6
Aspirin 8.0 4.3 4.2 3.8 2.0 2.1
Naproxen 4.3 4.0 5.3 3.5 1.7 1.6
Decongestant 27.7 16.0 5.8 8.1 8.9 6.3
Pseudoephedrine 25.1 15.4 5.5 7.9 8.4 5.9
Other decongestant 2.6 0.9 0.3 0.3 0.6 0.4
Antihistamine 14.8 7.5 4.3 5.3 4.7 3.8
Chlorpheniramine 4.3 3.0 1.2 1.5 1.6 1.0
Diphenhydramine 7.8 2.9 1.0 1.6 1.4 1.8
Loratadine 2.9 1.9 1.2 1.3 1.1 0.7
Doxylamine 1.1 1.4 0.9 0.9 0.5 0.3
Brompheniramine 0.4 0.5 0.2 .02 .02 .01
Cough medication 12.9 8.8 2.6 3.5 3.5 2
Guaifenesin 9.2 6.2 0.9 2.1 2.8 1.8
Dextromethorphan 8.0 3.4 1.6 1.7 1.4 0.8
* A total of 7563 mothers of offspring with and without birth defects were interviewed between 1998 and 2004.
y
A total of 2970 mothers of offspring without birth defects whose estimated date of delivery was between 1997 and 2001.
z
Prepregnancy period is the 3 months before the estimated date of conception.
granting OTC approval, the US Food and Drug Ad- care hospitals and control subjects without birth defects
ministration might take available data on pregnancy from the same catchment areas are identified within 5
exposures and fetal outcomes into consideration, but months after delivery. Mothers are interviewed by
adequate evidence, particularly on safety for specific trained nurses within 6 months after delivery in person
birth defects, typically is not available. Hence, although (1976-mid 1998) or by telephone (mid 1998-present).
the availability of a drug as an OTC product may reflect Standardized questions are asked about demographic,
safety for its use by the nonpregnant population, it does reproductive, and medical factors; behaviors (eg, smok-
not necessarily extend to safety for use during preg- ing, alcohol use); diet; and medication use. Information
nancy. Nonetheless, the general perception that OTC on medication use is obtained by questions on drugs
medications are safe for the general public and therefore taken for a list of illnesses (eg, cold, flu, cough, sinus
safe for pregnant women may have led, in part, to large infection, or congestion), on categories of medications
proportions of pregnant women taking these products. (eg, nasal sprays), and on selected specifically named
Conversely, this ignorance about the safety of OTCs can medications (eg, Tylenol, Sudafed, Advil, Aleve). For
raise concerns about risks to the fetus among the many reported medications, women are asked to retrieve the
women who take an OTC medication before recognizing package, if possible. In 1999, a medication identification
that they are pregnant. booklet was introduced, with color images of O350
We describe the extent of OTC medication use during specific OTC products. Women who reported the use of
pregnancy and discuss the potential public health a cough, cold, or analgesic product were asked to refer
implications. to the booklet to help them identify which specific
product they had taken. The use of generic forms of
Material and methods medications also were recorded.
The National Birth Defect Prevention Study
Since 1976, the Boston University Slone Epidemiology (NBDPS) is an on-going population-based case-control
Center Birth Defects Study (BDS) has been interviewing study of birth defects.5 Case subjects with major struc-
the mothers of infants with a range of birth defects. tural malformations and a random sample of control
During this time, the study has interviewed O23,000 infants (live births with no major birth defects) whose
mothers of offspring with and without birth defects from estimated date of delivery was from November 1997
the greater metropolitan areas of Boston (1976-present), onward are identified in Arkansas, California, Iowa,
Philadelphia (1977-present), Toronto (1978-present), Georgia, Massachusetts, New Jersey, New York, North
and San Diego (2001-present). Cases with major struc- Carolina, Texas, and Utah. Standardized interviews of
tural birth defects are identified from birth, and tertiary case and control mothers are conducted by trained
Werler et al 773
Figure 1 Secular trends of specific OTC medication use in the first trimester among 20,251 Birth Defects Study mothers of
offspring with and without birth defects.
774 Werler et al
Figure 2 Analgesic, decongestant, and antihistamine use in pregnancy by maternal demographic factors among 2970 mothers of
offspring without birth defects in the National Birth Defects Prevention Study.
interviewers by telephone within 24 months after the because San Diego became part of the study only recently
estimated date of delivery about a range of factors that (2001). Mothers of offspring with and without birth
are similar to those in the BDS. Questions are asked defects are combined for this analysis, because rates of
about medications that were taken for specific illnesses use of specific medications were not appreciably different
(eg, cold or flu) and about specifically named products between the 2 groups. For an examination of recent rates
(eg, Tylenol, Advil, Aspirin, Aleve). of medication use during pregnancy, data on 7563 BDS
For women in both the BDS and the NBDPS, reported participants who were interviewed between 1998 and 2004
products were linked to their active ingredients by the were included. For an examination of secular trends of
Slone Epidemiology Center Drug Dictionary. This tool medication use, data on 20,251 BDS participants who
allowed us to examine all agents that are typically taken as were interviewed between 1976 and 2004 were included.
OTC medications. For example, a report of Tylenol NBDPS data were restricted to Arkansas, California,
Allergy medication would be considered exposure to Iowa, Georgia, Massachusetts, New Jersey, New York,
acetaminophen, pseudoephedrine, and chlorphenir- and Texas because North Carolina and Utah became part
amine, which are its 3 active components. We included of the study only recently (2003). Interviews of 2970
all exposures to all agents that are available OTC, even mothers of non-malformed live births who were delivered
when such an agent may have been prescribed. Informa- between October 1997 and June 2001 were included
tion on dose was not collected. In the BDS, pregnancy is for this analysis. For both studies, OTC medications
defined as the period beginning with the last menstrual excluded vitamin, mineral, and herbal products.
period and ending at delivery. In the NBDPS, pregnancy is
defined as the period beginning 2 weeks after the last Results
menstrual period and ending at delivery. BDS data were
restricted to participants from the Boston and Philadel- In the 1998 to 2004 BDS data, the top 10 medications
phia centers because, in Canada, overall use tends to be that were taken in pregnancy, in rank order, were
less common and different products are available and acetaminophen, ibuprofen, pseudoephedrine, aspirin,
Werler et al 775
Figure 3 Adjusted (for maternal race or ethnicity, age, and years of education) rates of analgesic, decongestant, and antihistamine
use by state among 2970 mothers of offspring without birth defects in the National Birth Defects Prevention Study.
naproxen, diphenhydramine, guaifenesin, albuterol, nancy. These decreases in use were countered by a 7%
amoxicillin, and dextromethorphan. In the 1997 to increase in the use of acetaminophen during the same
2001 NBDPS data, the corresponding medications time frame. The use of all 4 analgesic agents (acetamin-
were acetaminophen, ibuprofen, antibiotics (not other- ophen, ibuprofen, naproxen, and aspirin) decreased
wise specified), amoxicillin, pseudoephedrine, naproxen, from the first to the second and third trimesters. Among
guaifenesin, aspirin, diphenhydramine, and chlorphenir- the cold, allergy, and cough medications, pseudoephed-
amine. In both the BDS and NBDPS, 8 of the top 10 rine and guaifenesin use increased from prepregnancy
products were available OTC. We compared the prev- to the second trimester, then decreased in the third
alence of use in pregnancy of OTC categories and trimester. With the exception of diphenhydramine, spe-
specific agents for overlapping periods of the BDS and cific antihistamines showed minor decreases across
NBDPS (Table). In both studies, acetaminophen was trimesters, as did the antitussive dextromethorphan.
the most commonly taken product, with at least 65.5% The increase in diphenhydramine use from 1.0% during
of women taking it at some point during pregnancy. the prepregnancy interval to 1.8% in the third trimester
Ibuprofen and pseudoephedrine were the next most was primarily due to the use of Benadryl.
commonly used products, with at least 15% of women BDS data reveal secular trends of selected specific
exposed in pregnancy. Aspirin and naproxen were used OTC products for a 28-year period (Figure 1). As
less commonly, but they were nonetheless used by at aspirin use decreased during the 1980s, acetaminophen
least 4% of women. Cough medicines and antihista- use increased; as acetaminophen use leveled off in the
mines were used by at least 7% of women. 1990s, ibuprofen use increased after it became available
Also included in Table is OTC medication use OTC (1984), followed by an increase in naproxen use
among NBDPS women by trimesters of pregnancy and after it became available OTC (1994). Pseudoephedrine
the 3-month period before the onset of pregnancy. use increased up until the early 1990s, when it stabilized
Among analgesic products, there was a 5% decrease in at prevalences of 15% to 18%. Among antihistamines,
ibuprofen use and a 2% decrease in naproxen use chlorpheniramine use has decreased, while diphenhy-
between prepregnancy and the first trimester of preg- dramine use has increased. Also, the use of loratadine, a
776 Werler et al
nonsedating antihistamine, increased from 0.2% when pregnancy. Interestingly, products that contain aceta-
it first became available by prescription to 3.7% in 2003 minophen (eg, Tylenol), pseudoephedrine (eg, Sudafed),
after it became available OTC. Both guaifenesin and chlorpheniramine (eg, Dristan), diphenhydramine (eg,
dextromethorphan use also increased over the past 2 Benadryl), and guaifenesin (eg, Robitussin) often are
decades. Ranitidine also increased in use from 0.6% in included on these lists.9,10 Women might interpret the
1996 to 2000 to 1.6% in 2001 to 2004, after OTC receipt of a list of ‘‘safe’’ medications as an encourage-
availability in 1995 (data not presented). ment for use should symptoms occur, whereas before
We examined demographic and regional patterns of pregnancy, they might have gone untreated.
OTC medication use in the NBDPS data. In Figure 2, The reported use of analgesics, decongestants, and
rates of analgesic, decongestant, and antihistamine use antihistamines was higher for white, non-Hispanic
in pregnancy are presented for categories of maternal women, women with more than a high school education,
race or ethnicity, years of education, and age. Rates of and women who were at least 20 years of age. Similar
analgesic and decongestant use were higher for white demographic patterns were observed in 2 separate
women, women with at least a high school education, studies that were conducted in the eastern United States
and women who were at least 20 years of age. Antihis- in the 1980s.6,7
tamine use showed the same pattern for maternal When these demographic factors were taken into
education and age, but the rates were similar for white, account, the rates of analgesic and antihistamine use
black, and Asian American women and lower for were similar across the states. For decongestants, the
Hispanic women. adjusted rates were lowest for the 3 northeastern states
We adjusted the rates of medication use by standard- (Massachusetts, New York, and New Jersey).
izing each state to the racial or ethnic, age, and educa- For reported use of OTC medication during the first
tion distributions of all states combined. Figure 3 shows trimester, prevalences in the NBDPS are generally
adjusted rates of analgesic, decongestant, and antihista- similar to those in a study of women who were delivered
mine use by state. Adjustment for these 3 demographic in 1995 and were interviewed within 96 hours of deliv-
factors accounted for much of the variability of analge- ery.8 However, the reported uses of some medications
sic and antihistamine use across states, whereas the were lower in the third trimester in the NBDPS, possibly
adjusted rates for decongestant use showed more vari- because of the longer interval between pregnancy and
ability. interview. Differences in data collection methods might
also account for the slightly higher prevalences of
Comment specific OTC products any time in pregnancy in the
BDS rather than in the NBDPS. For example, at the
OTC medication use during pregnancy is extremely beginning of the interview, the NBDPS asks questions
common, as observed in the present findings and in about occurrences of cold and flu and the medications
several earlier studies in the United States.6-8 Both the taken to treat them, whereas the BDS asks questions
BDS and NBDPS show that approximately two-thirds about illnesses and medication use toward the end of the
of women take acetaminophen and that approximately interview. The reporting of cough, cold, or allergy medi-
1 in 6 women takes a decongestant or ibuprofen during cations might be more accurate in the NBDPS rela-
pregnancy. Although the use of some medications, such tive to the BDS because some women may grow weary
as aspirin and chlorpheniramine, has decreased over the of reporting exposures toward the end of the lengthy
years, most usage has increased during the past 2 interview. Conversely, BDS interviews are conducted
decades; ibuprofen, naproxen, diphenhydramine, dex- closer to the time of delivery and include more prompts
tromethorphan, and loratadine have continued to infor specific OTC products. Also, a medication identifi-
crease in the most recent years. Further, rates of use for cation booklet is used to help women to identify the
acetaminophen, pseudoephedrine, chlorpheniramine, di- product that was taken in a product line, which results
phenhydramine, doxylamine, and guaifenesin in the in, for example, a larger proportion of acetaminophen-
first, second, or third trimester of pregnancy are actually exposed women reporting the use of a combination pro-
higher than during the 3 months before pregnancy. duct in the BDS (23%) than in the NBPDS (12%). The
Although such increases in use may be due to actual enhanced reporting of combination products results
increases in upper respiratory symptoms during preg- in more exposures to pseudoephedrine, antihistamines,
nancy, it is more likely that these changes may reflect a guaifenesin, and dextromethorphan, which are available
more relaxed attitude regarding the use of OTC drugs in in various combinations with acetaminophen in a single
pregnancy. Indeed, health care providers and Internet product.
access could be partially responsible for this pattern. The potential for inaccurate recall is problematic in
Some health care providers supply lists, both as hand- retrospective studies. For OTC medications, accurate
outs to their patients and on the Internet to the public, recall can be more difficult because their use during
of medications that they deem to be safe to take during pregnancy tends to be viewed more casually than does
Werler et al 777
the use of prescription drugs, the use tends to be of obtain empiric evidence of whether such exposures are
shorter duration than for prescription products, and the safe. The methodologic challenges to the study of OTC
use tends to be ‘‘as needed’’ rather than on a particular medications in pregnancy must be overcome so that the
schedule, which makes it more difficult to recall the risks and safety of specific products can be identified,
precise exposure period. Recall of OTC medication use critical information can be provided to drug manufac-
can be enhanced by prompting in multiple ways within turers and public health authorities and, most impor-
the interview, such as asking for products by indication tantly, both clinicians and women can be allowed to
and by product name.11 Women can be asked to retrieve make more informed treatment decisions. Further, when
the medication package if they still have it, so that the risks are identified, that new knowledge can lead to
exact name can be recorded. insights into causal pathways.
Many OTC products are marketed according to the
symptoms they treat and frequently contain more than
1 active ingredient (for example, Tylenol PM, Tylenol
Allergy & Sinus, Tylenol Cold and Flu, and Tylenol Acknowledgments
Cough, each contain acetaminophen in combination
We thank the Centers for Birth Defects Research and
with various antihistamines, decongestants, and anti-
Prevention in Arkansas, California, Georgia, Iowa,
tussives). One way to help women recall the exact
Massachusetts, New Jersey, New York, and Texas for
products that were taken is to ask questions about or
their data; Kathy Kelley, Research Pharmacist at Slone
provide pictures of specific products.
Epidemiology Center, for her assistance with drug
Specialized studies are necessary to generate infor-
classification; and the mothers who participated in the
mation about the risks and safety of OTC medication
BDS or NBDPS.
use during pregnancy. Structural birth defects include
many types of malformations that are heterogeneous
with respect to their cause. Because known teratogenic
medications tend to affect the development of specific References
types of birth defects rather than birth defects overall, it
is important to assess the use of specific medications in 1. Pauli RM, Hall JG. Warfarin embryopathy. Lancet 1979;2:144.
relation to specific outcomes.12 For example, isotreti- 2. Robert E, Guibaud P. Maternal valproic acid and congenital
neural tube defects. Lancet 1982;2:937.
noin embryopathy includes certain types of characteris-
3. Lammer EJ, Chen DT, Hoar RM, Agnish ND, Benke PJ, Braun JT,
tic brain, ear, and heart defects,3 and valproic acid et al. Retinoic acid embryopathy. N Engl J Med 1985;313:837-41.
increases the risk of neural tube defects (NTDs).2 If 4. Briggs GG, Freeman RK, Yaffee SJ. Drugs in pregnancy and
OTC medications were high-risk teratogens (like thalid- lactation, 5th ed. Baltimore: Lippincott, Williams & Wilkins; 2003.
omide or isotretinoin), they would most likely have 5. Yoon PW, Rasmussen SA, Lynberg MC, Moore CA, Anderka M,
come to attention. However, without careful and di- Carmichael SL, et al. The National Birth Defects Prevention
Study. Public Health Rep 2001;116(suppl):32-40.
rected study, it is possible that smaller risks of specific 6. Buitendijk S, Bracken MB. Medication in early pregnancy: prev-
defects are going undetected. From a public health alence of use and relationship to maternal characteristics. Am J
perspective, it is useful to compare the impact of a Obstet Gynecol 1991;165:33-40.
teratogenic prescription drug with a putative teratogenic 7. Rubin JD, Ferencz C, Loffredo C, Baltimore-Washington Infant
OTC drug. If valproic acid increases the risk of NTDs Study Group. Use of prescription and non-prescription drugs in
pregnancy. J Clin Epidemiol 1993;46:581-9.
approximately 10-fold and 1% of pregnancies are 8. Splinter MY, Sagraves R, Nightengale B, Rayburn WF. Prenatal
exposed, the drug would be responsible for an estimated use of medications by women giving use at a university hospital.
360 NTD cases in the United States annually. For South Med J 1997;90:498-502.
comparison, consider a hypothetic OTC drug that is 9. WebMD Health. Pregnancy: taking medicine during. Last accessed
taken by 20% of women and that increases NTD risk September 7, 2004. Available at: http://my.webmd.com/content/
article/51/40801.
only 2-fold; although the risk is lower than that of 10. Camino Medical Group Health Education. Patient handouts,
valproic acid, its far greater prevalence of use would medications safe to take in pregnancy. Last accessed September
result in that OTC drug being responsible for approx- 7, 2004. Available at: http://caminomedicalgroup.com/handouts/
imately 800 NTD cases per year. The common use of medssafe_preg.html.
OTC medications in pregnancy necessitates further 11. Mitchell AA, Cottler LB, Shapiro S. Effect of questionnaire design
on recall of drug exposure in pregnancy. Am J Epidemiol
studies to establish safety or to identify risks. 1986;123:670-6.
In conclusion, because most pregnant women ingest 12. Mitchell AA. Systematic identification of drugs that cause birth
at least 1 OTC medication, it is imperative that we defects: a new opportunity. N Engl J Med 2003;349:2556-9.
www.ajog.org
Comparison of the TDx-FLM II and lecithin to

sphingomyelin ratio assays in predicting fetal
lung maturity
Tamina Winn-McMillan, MD, Brad S. Karon, MD, PhD*
Department of Pathology, Sunrise Hospital and Medical Center, Las Vegas, NV
Received for publication November 1, 2004; revised December 3, 2004; accepted January 18, 2005
KEY WORDS Objective: We evaluated the usefulness of the TDx-FLM II and lecithin to sphingomyelin (L/S)
Fetal lung maturity ratio assays in predicting fetal lung maturity.
TDx-FLM II Study design: We retrospectively reviewed 218 consecutive paired TDx-FLM II and L/S ratio
L/S ratio results. Women who delivered viable infants within 72 hours of amniotic fluid collection
(n = 109) were included in the analysis of sensitivity and specificity. Concordance between tests
was determined for all women tested during the study period, and in the subset of women who
delivered viable infants within 72 hours of amniotic fluid collection.
Results: There were 9 respiratory distress syndrome (RDS)-affected infants born during the study
period. Both the TDx-FLM II and L/S ratios had 100% sensitivity in detecting RDS at their best
apparent cut-offs. There was a trend towards increased specificity of the L/S ratio compared with
the TDx-FLM II (80% for L/S vs 73% for FLM II). The overall concordance between the TDx-
FLM II and L/S ratio was approximately 75%.
Conclusion: The TDx-FLM II and L/S ratios are both sensitive tests for RDS; however, there is
not good concordance between the two. The results provide new insight into the optimal use, in
sequential or reflex cascade testing, of the TDx-FLM II and L/S ratio.
Respiratory distress syndrome (RDS), or hyaline fetal surfactant phospholipid in changing the biochem-
membrane disease, is a major cause of death in new- ical properties of maternal amniotic fluid.
borns. RDS is caused by insufficient surfactant pro- Of the tests for fetal lung maturity, two have emerged
duction by the neonatal lung, and can be predicted in as the most common tests performed to assess fetal
part by laboratory measurements of fetal lung maturity. surfactant production. The lecithin to sphingomyelin
Several laboratory measurements of fetal lung maturity (L/S) ratio is considered by many the gold standard
have been developed, all of which rely on the effect of for assessing fetal lung maturity; however, this test relies
on time-consuming and technically difficult thin layer
chromatography. The TDx-FLM II assay by Abbott
* Reprint requests: Brad S. Karon, Mayo Clinic, Department of
Laboratory Medicine and Pathology, 200 First Street SW, Rochester,
Laboratories (Abbott Park, Ill) is an automated fluo-
MN 55905. rescence polarization assay that is easy to perform, fast,
E-mail: Karon.bradley@mayo.edu and quantitatively measures the surfactant phospholipid
doi:10.1016/j.ajog.2005.01.045
Winn-McMillan and Karon 779
present in amniotic fluid reported as milligrams surfac- Material and methods

tant present per gram of albumin.1-5
Previous studies of the TDx-FLM II have found This study was a retrospective evaluation of consecutive
that the test is a reliable predictor of fetal lung physician-ordered TDX-FLM II results that included an
maturity. One recent study found that the TDx-FLM order for L/S ratio, collected over a 41-month period
II test, using a cut-off value of 45 mg surfactant per between August 2000 and December 2003 at a single site
gram albumin, was a better indicator of fetal lung (Sunrise Hospital). Over this period, 90% of all physi-
maturity than the L/S ratio.6 Another study also cian orders for TDX-FLM II included an order for L/S
suggested that the TDx-FLM test (in this case the first ratio. During the study period, 218 paired TDX-FLM II
generation TDx-FLM, not the TDx-FLM II used and L/S tests were performed on 207 women. Samples
currently) may have better sensitivity in predicting fetal tested included amniotic fluid from amniocentesis on
lung maturity than the L/S ratio, but the L/S ratio may women presenting with signs of preterm labor, vaginal
have better specificity.4 In this context, the sensitivity pool samples from women presenting with preterm
of a test for fetal lung maturity refers to the probability labor, and samples from asymptomatic women before
of an immature result or index in an infant with scheduled cesarean section. Samples visibly contami-
RDS, while the specificity refers to the probability of nated with bilirubin, blood, or meconium were rejected
a mature result in an infant without RDS. Others have from analysis.
suggested that TDx-FLM II values must be interpreted Data from women who delivered infants within 72
with regard to the gestational age of the neonate. In hours of paired FLM II and L/S ratio tests (n = 109)
one such study, a polytomous regression program for were included in the analysis of fetal outcomes. The
probability estimation was used to combine risks for sensitivity and specificity of the FLM II at its apparent
RDS based on gestational age and TDx-FLM II best cut-off (45 mg/g) was compared with the L/S ratio
values. Although the optimal cut-off for detecting at the two cut-offs commonly used in clinical practice
RDS (not considering gestational age) was 45 mg (2.0 and 2.5), both including and excluding women who
surfactant/g albumin, the authors found that the received steroids at the time of amniotic fluid collection
regression model still predicted significant risk for or between the time of fluid collection and delivery.
RDS (even with TDx-FLM II values R45 mg/g) at Because the number of affected infants (infants with
gestational ages less than 34 weeks.7 RDS) was expected to be small, we did not design the
The aims of the present study were to compare the present study to define best cut-offs for the FLM and
TDx-FLM II and L/S ratio in predicting fetal lung L/S ratio tests. Rather, we sought to evaluate the
maturity. Because previous similar studies were limited performance of the two tests in our population using
by the small number of RDS-affected infants born, we best apparent cut-offs based on previous studies and
did not attempt in this study to define the optimal cut- clinical practice.
offs of the TDx-FLM II or L/S ratio tests using ROC Infants who were treated with surfactant, or were
analysis or other means. Rather, we compared the placed on a ventilator more than 24 hours, or required
sensitivity and specificity of the TDx-FLM II test using continuous positive airway pressure for more than 24
its best apparent cut-off determined from previous hours, were given the diagnosis of RDS, provided that
studies (45 mg surfactant/g albumin) with the L/S ratio no other congenital anomalies were present that could
at cut-offs used commonly in clinical practice (L/S ratio explain the need for respiratory support. This clinical
of 2.0 or 2.5). definition of RDS is identical to that used in the study
In addition, we sought to determine the overall by Fantz et al6 in order to make the data sets
concordance between TDx-FLM II and L/S ratio results comparable. Twin gestations were included in the study
(ie, how often a mature prediction on one test corre- but were counted as one data point because both twins
sponded to a mature prediction on the other). This is had the same outcome in all cases of twin gestation. The
important because laboratories often perform the TDx- study design was within guidelines provided by the
FLM II result first, and subsequently reflex an L/S ratio Institutional Review Board at Sunrise Hospital.
if the result falls outside of a defined range. To date, All TDx-FLM II analysis was performed on two Abbott
no study has compared a large series of TDx-FLM II TDx-FLX analyzers (Abbott Laboratories) at Sunrise
and L/S ratio results. Because 90% of TDx-FLM II Hospital according to manufacturer’s instructions. The
orders during the study period were accompanied by an L/S ratio was determined by an outside laboratory
order for L/S ratio, we were able to obtain the first large (Quest Diagnostics, Las Vegas, Nev). The cut-off for
concordance data set to compare the results of consec- fetal lung maturity was a ratio of 2.0 at the beginning of
utive paired TDx-FLM II and L/S tests. This concor- the study period; however, it changed during the study
dance data will be useful for both laboratories period to its present value of 2.5. Results of L/S ratio
and clinicians in establishing protocols for reflex L/S were typically available 6 to 8 hours after amniotic fluid
testing. collection, whereas TDx-FLM II results were typically
780 Winn-McMillan and Karon
Table I Diagnostic efficiency of TDx-FLM II and L/S ratio in identifying RDS

Predictive value (%, CI)
Cut-off Sensitivity (%, CI) Specificity (%, CI) Mature result Immature result
FLM II (R45 mg/g) 100 (66-100) 73 (63-81) 100 (95-100) 25 (12-42)
L/S ratio (R2.0) 100 (66-100) 80 (71-87) 100 (95-100) 31 (15-51)
L/S ratio (R2.5) 100 (66-100) 62 (52-71) 100 (94-100) 19 (9-33)
available within 1 to 2 hours after fluid collection. The

Table II Gestational age, TDx-FLM II values, and L/S ratio
distribution of gestational ages at time of amniotic fluid values for the 9 RDS-affected infants
collection ranged from 30 to 37 weeks. Over half the
Gestational age TDx-FLM II L/S ratio
women were tested at 34 to 35 weeks gestational age,
3
while less than 5% were tested at less than 32 weeks. 32 22.0 1.5:1
Statistical analysis was performed using GraphPad 336 21.7 1.6:1
InStat version 3 for Windows 2000 (GraphPad Soft- 342 22.3 1.6:1
345 29.3 1.9:1
ware, San Diego, Calif). Sensitivity refers to the prob-
344 26.7 1.6:1
ability of an immature test result in a patient with RDS. 350 26.2 0.9:1
Specificity refers to the probability of a mature result in 330 20.5 1.5:1
a patient without RDS. The predictive value of a mature 340 22.2 1.6:1
result is the percentage of patients with a mature result 316 38.2 1.5:1
who did not have RDS. The predictive value of an
immature result is the percentage of patients with an
immature result who did have RDS.
Excluding women who received antenatal steroids at
the time of amniotic fluid collection or between the time
Results of amniotic fluid collection and delivery, there were 96
women who delivered infants within 72 hours of
A total of 218 paired TDx-FLM II and L/S ratio results amniotic fluid analysis. Among this population there
were performed on 207 women over the study period. Of were 8 RDS-affected infants born. All 8 women who
these 207 women, 109 delivered infants within 72 hours delivered RDS-affected infants had TDx-FLM II amni-
of testing. There were 9 infants who met the criteria for otic fluid values less than 45 mg/g, and L/S ratios less
RDS, and 100 unaffected infants born. All of the women than 2.0 (sensitivity 100%). Among the 88 women who
who delivered RDS-affected infants had TDx-FLM II delivered unaffected infants, 68/88 (77%) had a TDx-
values on amniotic fluid less than 45 mg surfactant/g FLM II value on amniotic fluid R45mg/g, while 74/88
albumin, and L/S ratios less than 2.0. Thus, the (84%) had an L/S ratio R2.0.
sensitivity of the TDx-FLM II test at a cut-off of 45 We also sought to determine the concordance be-
mg/g, and the L/S ratio at a cut-off of 2.0, was 100% tween the TDx-FLM II test at its apparent best cut-off
(Table I). The range of TDx-FLM II values among (45 mg/g) and the L/S ratio at its apparent best cut-off
women who delivered affected infants was 20-38 mg/g, (2.0). Among the 109 test results evaluated on women
and the range of L/S ratios was 0.9 to 1.9. The who delivered infants within 72 hours of TDx-FLM II
gestational age, TDx-FLM II, and L/S ratio values of and L/S ratio tests, the overall concordance rate
the RDS-affected infants are listed in Table II. (percentage of the time that a mature result on one
Among the 100 women who delivered unaffected test correlated with a mature test on the other, and
infants, a TDx-FLM II value of R45 mg/g was obtained immature results correlated) was 79%. In 7% of samples
in 73/100 cases (73% specificity). For the L/S ratio, 80/100 tested in this population, the TDx-FLM II predicted
women had a value R2.0 (80% specificity), while 62/100 fetal lung maturity (R45 mg/g), while the L/S ratio
women had a value R2.5 (62% specificity). This repre- predicted immaturity (!2.0). In 14% of samples tested,
sents a trend toward increased specificity of the L/S ratio the L/S ratio predicted maturity (R2.0), while the TDx-
(at a cut-off of 2.0) vs the TDx-FLM II test (at a cut-off FLM II predicted immaturity (!45 mg/g) (Table III).
of 45 mg/g), although the difference in specificity did not Among all 209 samples tested, the overall concor-
reach statistical significance. The specificity of the L/S dance between the TDx-FLM II (at a cut-off of 45 mg/g)
ratio at a cut-off of 2.0 was significantly better than at and the L/S ratio (at a cut-off of 2.0) was 76%. In 4% of
a cut-off of 2.5 (P ! .05). The sensitivity, specificity, samples tested in this group, the TDx-FLM II test
predictive value of a mature result, and predictive value predicted fetal lung maturity (R45 mg/g), while the L/S
of an immature result are listed in Table I. ratio predicted immaturity (!2.0). In 19% of samples
Winn-McMillan and Karon 781
Table III Concordance between TDx-FLM II and L/S results Table IV Concordance between TDx-FLM II and L/S results in
in women who delivered infants within 72 hours of amniotic all women who had both FLM II and L/S results performed
fluid collection (represents 90% of FLM II tests performed during study period)
Number (%) in each category Number (%) in each category
FLM II !45 mg/g FLM II R45mg/g FLM II !45 mg/g FLM II R45mg/g
L/S R2.0 15 (14%) 67 (60%) L/S R2.0 42 (19%) 96 (44%)
L/S !2.0 21 (19%) 8 (7%) L/S !2.0 71 (33%) 9 (4%)
tested, the L/S ratio predicted maturity (R2.0), while L/S results R2.5, suggesting increased sensitivity of the
the TDx-FLM II predicted immaturity (!45 mg/g) TDx-FLM II compared with the L/S ratio.
(Table IV). Our study is the first to compare a full set of
corresponding TDx-FLM II and L/S ratio values on
women who delivered viable infants within 72 hours of
Comment amniotic fluid collection. We did not observe a difference
in sensitivity between the TDx-FLM II and L/S ratio at
Three recent studies have examined the utility of the their apparent best cut-offs (both had a sensitivity of
TDx-FLM II assay for the evaluation of fetal lung 100%). However, we did observe a trend towards
maturity.6-8 The studies by Fantz et al6 and Kesselman increased specificity of the L/S ratio compared to the
et al8 were retrospective reviews of consecutive TDx- TDx-FLM II, which did not reach statistical signifi-
FLM II values from women who delivered viable infants cance. The specificity of the L/S ratio at a cut-off of 2.0
within 72 hours of amniotic fluid collection. Both studies was significantly better than at a cut-off of 2.5 (Table I).
included small numbers of RDS-affected infants (15 in Similar to other studies on the TDx-FLM II test, we
the study by Fantz et al, 7 in the study by Kesselman were limited by the small number of RDS-affected
et al) because of the low prevalence of RDS in the study infants born during the study period. In addition, our
populations (8.1% and 8.5%, respectively). Fantz et al6 study was retrospective so clinician bias in establishing
found that no affected infants were born when the TDx- the diagnosis of RDS (based on either TDx-FLM II or
FLM II from maternal amniotic fluid was greater than L/S results or both) cannot be eliminated. However, our
or equal to 45 mg/g. In the study by Kesselman et al,8 2 study adds significantly to the available information
RDS-affected infants were born to women whose comparing the performance of the TDx-FLM II and L/S
amniotic fluid TDx-FLM II value was indeterminate ratio tests.
(40-55 mg/g), while no affected infants were born when In comparing results of various studies, however,
the FLM II value was R55 mg/g. a few cautionary notes are necessary. First, studies have
The study by Kaplan et al7 differed from the others in used different clinical definitions of RDS, which can
that (1) it was a case-finding study of RDS-affected affect the prevalence of the disease. We used the same
infants that also had amniotic TDx-FLM II values clinical definition of RDS as used in Fantz et al,6 and the
available, and (2) the results were expressed as a prob- prevalence of RDS in our study was nearly identical to
ability model including both FLM II value and gesta- that observed in Fantz et al6 and Kesselman et al.8
tional age. Because a case-finding method was used in Second, Kaplan et al7 demonstrated that even with
this study, the prevalence of RDS (15%) and number of ‘‘mature’’ TDx-FLM II values of R45 mg/g, significant
affected infants (46) were greater than observed in risk for RDS exists at less than 34 weeks gestational age.
previous studies. The study found that the single best It should be noted that the majority of women in our
cut-off for the TDx-FLM II test for predicting fetal lung study had amniotic fluid collection performed at 34 to 35
maturity was 45 mg surfactant/g albumin. weeks gestational age, while less than 5% of women were
Using the first generation TDx-FLM test, one pre- tested at less than 32 weeks gestational age. Gestational
vious study found that the sensitivity of the TDx-FLM age is clearly an important consideration in interpreting
(at its apparent best cut-off of 70 mg/g) was superior to the results of the TDx-FLM II and L/S ratio.7,9
that of the L/S ratio (at a cut-off of 2.0), while the Our study is the first to evaluate the concordance
specificity of the L/S ratio was superior to that of the between the TDx-FLM II and L/S ratio tests using
TDx-FLM.4 The study by Fantz et al6 used the second a large series of consecutive samples. There was con-
generation TDx-FLM II assay, but did not compare cordance of results (mature or immature prediction on
a full set of FLM II and L/S results (only 55/185 FLM II both tests) on 79% of tests performed on women who
results had a corresponding L/S ratio). However, this delivered within 72 hours of amniotic fluid collection,
study did note that 3/15 RDS-affected infants did have and on 76% of results overall. The most common
782 Winn-McMillan and Karon
discrepant result was for the TDx-FLM II to predict 2. Herbert W, Chapman JF, Schooner MM. Role of the TDx
fetal lung immaturity, while the L/S test predicted fetal FLM assay in fetal lung maturity. Am J Obstet Gynecol 1993;168:
808-12.
lung maturity. Given that in most situations the TDx- 3. Blumenfeld TA, Stark RI, James LS, George JD, Dyrenfurth I,
FLM II result will be available first, our results suggest Freda VJ, et al. Determination of fetal lung maturity by fluores-
that there is value in performing the L/S ratio when the cence polarization of amniotic fluid. Am J Obstet Gynecol
results of the TDx-FLM II suggest fetal lung immatu- 1978;130:782-7.
rity. This conclusion is based on the available data 4. Hagen E, Link JC, Arias F. A comparison of the accuracy of the
TDx FLM assay, lecithin-sphingomyelin ratio, and phosphatidyl-
(including ours), which suggest equal or better specificity glycerol in the prediction of neonatal respiratory distress syndrome.
of the L/S test compared with the TDx-FLM II. Our Obstet Gynecol 1993;82:1004-8.
data do not support a role for reflex L/S testing when 5. Bonebrake RG, Towers CV, Rumney CV, Reimbold P. Is fluores-
the TDx-FLM II suggests fetal lung maturity. This cent polarization reliable and cost efficient in a fetal lung maturity
conclusion is based on the available data suggesting cascade? Am J Obstet Gynecol 1997;177:835-41.
6. Fantz CR, Powell C, Karon B, Parvin CA, Hankins K, Dayal M,
equal sensitivity between the 2 tests, and the relatively et al. Assessment of the diagnostic accuracy of the TDx-FLM II to
uncommon finding (4%-7% of samples, Tables III and predict fetal lung maturity. Clin Chem 2002;48:761-5.
IV) of an immature L/S result occurring with a mature 7. Kaplan LA, Chapman JF, Bock JL, Santa Maria E, Clegan S,
TDx-FLM II result. Because of the limitations inherent Huddleston DJ, et al. Prediction of respiratory distress syndrome
in all studies of this type (small number of RDS-affected using the Abbott FLM-II amniotic fluid assay. Clin Chim Acta
2002;326:61-8.
infants), further studies are necessary to determine the 8. Kesselman EJ, Figueroa R, Garry D, Maulik D. The usefulness of
optimal use of the TDx-FLM II and L/S ratio tests. the TDx/TDxFLx fetal lung maturity II assay in the initial
evaluation of fetal lung maturity. Am J Obstet Gynecol
References 2003;188:1220-2.
9. Tanasijevic MJ, Wybenga DR, Richardson D, Greene MF, Lopez
1. Apple FS, Bilodeau L, Preese LM, Benson P. Clinical implementa- R, Winkelman JW. A predictive model for fetal lung maturity
tion of a rapid automated assay for assessing fetal lung maturity. employing gestational age and test results. Am J Clin Path
J Reprod Med 1994;39:883-7. 1994;102:788-93.
www.ajog.org
Increasing maternal parity predicts neonatal adiposity:

Pune Maternal Nutrition Study
Niranjan P. Joshi, MD, DNB,a Smita R. Kulkarni, MSc,a Chittaranjan S.
Yajnik, MD, FRCP,a,* Charudatta V. Joglekar, MS,a Shobha Rao, PhD,b
Kurus J. Coyaji, MD,a Himangi G. Lubree, MSc,a Sonali S. Rege, MSc,a
Caroline H. D. Fall, DM, FRCP, FRCPCHc
Diabetes Unit, KEM Hospital and Research Centre,a Pune, Maharashtra, India; Agharkar Research Institute,b Pune,
India; MRC Environmental Epidemiology Unit, University of Southampton, Southampton, United Kingdomc
Received for publication September 4, 2004; revised December 16, 2004; accepted January 11, 2005
KEY WORDS Objective: This study was undertaken to study the effect of parity on maternal and neonatal
Parity characteristics.
Maternal size Study design: Maternal anthropometry, diet, micronutrient status, biochemistry, and physical
Maternal metabolism activity were measured during pregnancy and detailed neonatal size recorded in 770 pregnancies
Neonatal adiposity in rural Maharashtra, India.
Results: Increasing parity was associated with larger offspring birth weight, skinfold thicknesses,
and abdominal circumference, but not head circumference and length. Compared with
primiparous women, multiparous women were older, less adipose, and more physically active
but had similar education, socioeconomic status, nutritional intake, and weight gain during
pregnancy. They had lower circulating concentrations of hemoglobin, albumin, ferritin, glucose,
and insulin and lower total leucocyte counts at 18 and 28 weeks’ gestation. There was no
difference in their husbands’ body size. The relationship between maternal parity and neonatal
weight and adiposity was significant independent of the difference in maternal characteristics.
Conclusion: Increasing maternal parity predicts increasing adiposity in the newborn infant. This
may result from maternal nutritional, cardiovascular, or immunologic factors.
Increasing maternal parity is associated with increas- a few studies have measured birth size other than
ing birth weight in the infant. This has been shown in weight. The Pune Maternal Nutrition Study (PMNS)
both cross-sectional studies and serial studies of the is a population-based study of maternal nutrition
same women across more than one pregnancy.1-3 Only and fetal outcome in 6 villages near the city of
Pune, Maharashtra, India.4-6 Maternal nutritional and
Supported by the Wellcome Trust, London, and the Medical metabolic status was measured before and during
Research Council, United Kingdom.
pregnancy, and detailed anthropometric measurements
* Reprint requests: Chittaranjan S. Yajnik, MD, FRCP, Diabetes
Unit, KEM Hospital and Research Centre, Rasta peth, Pune, 411011,
of the babies were made at birth. We have used data
Maharashtra, India. from this study to examine the relationship of parity to
E-mail: diabetes@vsnl.com different neonatal measurements of the offspring.
doi:10.1016/j.ajog.2005.01.020
784 Joshi et al
Material and methods tory at KEM Hospital, Pune. Assays for insulin, vitamin
C, red cell folate, and ferritin were performed in South-
The PMNS methodology has been described previ- ampton and Cambridge, UK. Maternal hemoglobin
ously.4-6 In brief, 2675 married women of child-bearing concentrations and total leucocyte, red blood cell, and
age (excluding sterilized couples), living in 6 villages platelet counts were measured with a Coulter T540
(total population 35,000) were identified by a house-to- analyser (Beckman Coulter UK Ltd, Buckinghamshire,
house survey. Of these, 2466 women agreed to take part UK). Serum albumin and plasma glucose and triglyceride
in the study. The majority were illiterate, belonged to concentrations were measured with Randox kits (Ran-
subsistence farming families, and worked on the farm as dox Laboratories Ltd, Crumlin, Northern Ireland) on an
well as doing household work. Field workers visited the Abbott Spectrum autoanalyzer (Abbott Laboratories,
women monthly to record their menstrual dates, and Abbott Park, Ill). Erythrocyte folate and serum ferritin
every 3 months to record anthropometry (weight, height, concentrations were measured with radioimmunoassays
head circumference, and triceps, biceps, subscapular, and (Beckton Dickinson UK Ltd, Oxford, UK).4 Serum
suprailiac skinfold thicknesses). Maternal prepregnant vitamin C concentrations were measured with an ascor-
fat mass was calculated by using the 4 skinfolds.7 Women bate oxidase-orthophenylene diamine assay on a Roche
who missed 2 successive periods were examined by Cobas Bio Centrifugal analyzer with fluorescence attach-
ultrasound at 15 to18 weeks to confirm pregnancy and ment.4 Plasma insulin concentrations were measured on
assess gestational age.8 Gestational age was derived from the Access Immunoassay System (Sanofi Pasteur Diag-
the last menstrual period unless it differed from the nostics, Chaska, Minn) with the use of a 1-step chemilu-
sonographic estimate by more than 2 weeks, in which minescent immunoenzymatic assay. Insulin resistance
case the latter was used. Women entered the study if was calculated from fasting glucose and insulin concen-
a singleton pregnancy of less than 21 weeks’ gestation trations by using the homeostasis model assessment
was confirmed. Enrollment of pregnant women equation (HOMA).10
(n = 814) began in June 1994 and ended in April 1996. The babies were measured by 1 of 3 trained field
Their socioeconomic status was assessed with a standard- workers within 72 hours of birth. IOV studies revealed
ized questionnaire, which derives a composite score a CV of 9.5% (range 7.5-10.8) for triceps and 7.5%
based on occupation and education of the head of the (range 4.0-8.8) for subscapular skinfold. At the time of
household, caste, type of housing, and family ownership these measurements, all babies were exclusively breast
of animals, land, and material possession.9 Ethical fed. Birth weight was measured to the nearest 25 g by
permission for the study was granted by the KEM using a Salter spring balance and crown-heel length to
Hospital Ethical Committee, and by the local village the nearest 0.1 cm using a portable stadiometer (Pedo-
leaders. All participants gave signed informed consent. baby, ETS, JMB, Brussels, Belgium). Triceps and sub-
For all pregnant women, additional data were col- scapular skinfold thicknesses were measured to the
lected in a standardized manner at 18 G 2 weeks’ and nearest 0.2 mm, on the left side of the body, with
28 G 2 weeks’ gestation. These included the anthropom- Harpenden skinfold calipers (CMS Instruments, Lon-
etry of the women themselves (same measurements as don, UK). Occipito-frontal head circumference and
prepregnancy) and their husbands (weight and height, midupperarm circumference (MUAC) were measured
waist and hip circumferences). A total of 21 trained to the nearest 0.1 cm with fiberglass tapes (CMS Instru-
observers made the measurements over the period of the ments). Abdominal circumference was measured at the
study, with an average of 11 observers at any given time. level of umbilicus in expiration. Placental weight was
A total of 9 interobserver variation (IOV) studies were recorded to the nearest 5 g with the use of Ishida scales,
performed during the study. The median (IQR) co- after trimming off the umbilical cord and membranes.
efficient of variation (CV) for triceps skinfold was 8.4% Effect of seasonality was investigated for maternal
(7.2-11) and for subscapular skinfold 2.7% (1.8-12.1). and neonatal measurements in a standardized manner:
Maternal dietary intakes were measured with a semi- summer (April-July), rainy season (August-November),
weighed 24-hour recall method and a food frequency and winter (December-March).
questionnaire.4 Physical activity was assessed with
a structured questionnaire that recorded the time spent
Statistical methods
in a range of activities and derived a total daily score,
based on the relative energy expenditure of the different Mothers were divided into 3 parity groups, which we
activities.6 Fasting blood samples were taken at both time define as follows: women who were pregnant for the
points in pregnancy; in addition, at 28 weeks’ gestation an first (primipara), second (para 2), and third or more
oral glucose tolerance test was carried out, with blood times (multipara). Data represented as mean G SD
samples taken fasting and 120 minutes after a standard unless otherwise specified. Variables with skewed dis-
75-g oral glucose load. Hemogram, glucose, lipids, and tributions (subscapular and triceps skinfold thicknesses,
albumin were measured at Diabetes Research Labora- and maternal ferritin and insulin concentrations) were
Joshi et al 785
Table I Mean (SD) maternal characteristics according to parity

Characteristics Primipara Para 2 Multipara P P*
N 245 268 257
Age (y) 18.9 (2.2) 21.0 (2.7) 24.0 (3.4) ! .001 d
Illiterate (%), College educated (%) 17.7, 10.7 18.0, 13.0 19.0, 6.6 .13 d
Socioeconomic score 27.1 (6.7) 27.0 (6.6) 26.4 (7.2) .19 d
Before pregnancy
Height (cm) 152.5 (4.7) 151.8 (5.2) 151.3 (4.9) ! .01 .04
Weight (kg) 42.3 (4.9) 41.6 (5.1) 41.1 (5.0) ! .01 .04
BMI (kg/m2) 18.2 (1.9) 18.0 (1.9) 17.9 (1.9) .09 .06
Head circumference (cm) 52.1 (1.4) 52.3 (1.5) 52.3 (1.5) 0.21 .81
Subscapular skin fold (mm)y 11.3 (8.9-14.1) 9.9 (7.7-12.3) 9.3 (7.5-11.6) ! .001 ! .001
Fat mass (kg) 9.6 (2.6) 8.8 (2.7) 8.4 (2.6) ! .001 ! .001
At 28 wks’ gestation
Physical activity score 62.1 (25.9) 64.9 (26.5) 69.8 (25.4) ! .001 ! .001
Weight gain (kg) 3.2 (2.0) 3.7 (2.4) 3.5 (2.1) .09 .06
Energy intake (kcal) 1678 (485) 1672 (499) 1671 (506) .89 .71
Hemoglobin (g/L) 113.0 (16.0) 112.0 (14.0) 108.0 (15.0) ! .001 ! .001
Total leucocyte count (!109/L) 9.6 (2.1) 9.1 (2.0) 8.7 (1.8) ! .001 ! .001
Red blood cell count (!1012/L) 3.9 (0.4) 3.9 (0.5) 3.8 (0.4) .47 .22
Serum albumin (g/L) 37.0 (3.0) 36.0 (3.0) 36.0 (3.0) ! .001 ! .001
Plasma ferritin (pmol/L)y 29.2 (15.7-53.9) 22.5 (15.7-40.0) 22.5 (15.7-43.1) .84 ! .01
Erythrocyte folate (nmol/L) 1062.8 (407.9) 1049.2 (396.6) 1026.5 (407.9) .389 .09
120-min plasma glucose (mmol/L) 4.6 (1.2) 4.4 (1.1) 4.2 (0.9) ! .01 ! .01
120-min plasma insulin (pmol/L)y 95.0 (38.0-148.0) 77.0 (36.8-143.5) 60.5 (25.8-107.3) ! .01 ! .001
Insulin resistance (HOMA-R) 0.59 (0.37-0.87) 0.55 (0.37-0.88) 0.50 (0.34-0.75) .45 .83
Plasma triglycerides (mmol/L) 1.5 (0.5) 1.5 (0.6) 1.5 (0.4) .25 .05
Systolic blood pressure (mm Hg) 115.3 (9.4) 112.3 (9.2) 110.4 (9.0) ! .001 ! .001
At birth
Gestational age (wk) 38.9 (2.1) 38.8 (1.9) 38.8 (1.8) .67 .64
Preterm (%) 13.5 9.7 12.1 .40 d
P values indicate the significance of linear trend using analysis of variance.
* P after adjusting for maternal age and socioeconomic status.
y
Median and interquartile range for log transformed variables.
log transformed to satisfy assumptions of normality. in all 3 parity groups, but a higher percentage of
The significance of the mean difference between groups multiparous women underwent medical termination of
was analyzed by analysis of variance. We also tested for pregnancy (n = 10; 3.8%) than primipara (0%) or para
linearity of trend for maternal and offspring character- 2 (n = 4; 1.5%).
istics against parity. Multiple linear regression was used Of the 770 mothers who completed pregnancy, 245
to determine the effects of maternal and paternal (32%) were primipara, 268 (35%) were para 2, and 257
characteristics on the size of the newborn infants. (33%) were multipara (Table I). Of the 257 multiparous
Analyses were carried out with SPSS for Windows mothers, 167 were para 3, 59 were para 4, 19 were para
(version 10.0) (SPSS, Chicago, Ill). 5, 8 were para 6, 2 were para 7, and 1 each of parity 8
and 9. Mothers of higher parity were older, but the
mean age in the multiparous group was still only 24
Results years. There were no associations between parity and
education level or socioeconomic status. Before preg-
Of 814 women with confirmed pregnancies, 10 did not nancy, women of higher parity were shorter and lighter
have prepregnant anthropometry recorded, 4 were and had thinner skinfolds and lower fat mass. These
beyond 21 weeks’ gestation at recruitment, and 3 had associations remained significant after adjusting for
major fetal anomalies detected on early ultrasound scan maternal age, socioeconomic status, and observer.
and subsequently terminated the pregnancy. Of the Maternal head circumference was similar in all parity
remaining 797 women, 12 spontaneously aborted, 14 groups. Paternal weight, height, and body mass index
underwent a medical termination of pregnancy, and 1 did not vary with maternal parity (data not shown). At
died from pregnancy-induced hypertension. Percentages 28 weeks’ gestation, women of higher parity had higher
of women who had spontaneous abortions were similar physical activity scores. Though physical activity was
786 Joshi et al
Table II Mean (SD) neonatal measurements according to maternal parity

Boys (338) Girls (293)
Maternal parity Maternal parity
Characteristics Primipara Para 2 Multipara P P* Primipara Para 2 Multipara P P*
N 111 117 110 84 109 100
Weight (g) 2625 (334) 2757 (360) 2778 (353) ! .001 ! .001 2518 (319) 2638 (358) 2639 (343) .02 ! .001
Length (cm) 47.8 (2.3) 48.2 (1.9) 48.3 (1.7) .09 .06 47.3 (2.0) 47.3 (2.0) 47.4 (1.8) .09 .43
Ponderal 24 (2.0) 25 (2.0) 25 (2.0) ! .01 .02 24 (2.0) 25 (3.0) 25 (3.0) .02 ! .01
index (kg/m3)
Skinfolds
Subscapular 3.6 4.2 4.2 ! .001 ! .001 4.0 4.3 4.4 ! .001 ! .001
(mm)y (3.4 - 4.2) (3.6-4.8) (3.8-4.8) (3.5-4.4) (3.8-4.8) (3.8-5.0)
Triceps (mm)y 3.8 4.2 4.2 ! .001 ! .001 4.1 4.2 4.3 .04 .03
(3.2-4.4) (3.6-4.8) (3.5-4.6) (3.6-4.6) (3.8-4.8) (3.7-4.8)
Circumferences
Head (cm) 33.3 (1.3) 33.4 (1.3) 33.4 (1.1) .75 .45 32.7 (1.3) 32.8 (1.2) 32.8 (1.2) .49 .17
Abdomen (cm) 28.3 (2.0) 28.8 (1.9) 28.9 (1.8) .03 .03 28.2 (1.9) 28.8 (1.9) 28.8 (1.9) .06 ! .01
MUAC (mm) 9.6 (0.9) 9.7 (0.9) 9.8 (0.8) .08 .09 9.5 (0.9) 9.8 (0.9) 9.7 (0.9) .47 .04
Placental 350.0 (67.0) 371.1 (76.6) 375.4 (84.0) .02 .02 346.0 (66) 361.0 (79) 354 (80) .51 .22
weight (g)
P values indicate the significance of linear trend using analysis of variance.
* P after adjusting for gestational age at delivery.
y
Median and interquartile range for log transformed variables.
related to season (highest in winter, least in summer), the and 120 g for girls in the multipara group compared with
parity and physical activity association was unaffected the primipara group. Birth weight was not related to
by seasonality. Weight gain during pregnancy, dietary season at birth (P = .296) and the birth weight difference
intakes (energy, protein, carbohydrate and fat intakes, among the 3 parity groups was unaffected by adjusting
and intakes of green leafy vegetables and dairy prod- for seasonality. Ponderal index and abdominal circum-
ucts) and erythrocyte folate and plasma vitamin C ference were also larger in babies born to mothers of
concentrations were similar in all parity groups (even higher parity, and subscapular and triceps skinfolds
after adjustment for seasonality). Women of higher increased strongly with increasing parity. A substantial
parity had lower hemoglobin concentrations, total component of this relationship was due to smaller
leucocyte counts, serum albumin concentrations, plasma skinfold thickness in babies of primipara women com-
ferritin, 120-minute glucose and insulin concentrations, pared with those of multipara women, but there was also
and systolic blood pressure. Red blood cell and platelet a continuous relationship between maternal parity and
counts, fasting glucose and insulin, plasma triglyceride offspring adiposity. Offspring size, including skinfold
concentrations, insulin resistance, and diastolic blood thicknesses, were related to gestational age at delivery.
pressure did not vary with parity. Findings were similar However, the relationship between maternal parity and
for all these maternal measurements recorded at 18 offspring size was unaffected by adjusting for gestation at
weeks’ gestation (data not shown). Average gestation at delivery and observer. Season at birth was not a signifi-
delivery and the percentage of preterm deliveries (!37 cant predictor of offspring adiposity, and the adjustment
weeks’ gestation) and stillbirths were similar in all 3 for seasonality did not alter the relationship between
parity groups. maternal parity and offspring adiposity. In contrast,
Of 770 babies delivered, we excluded 8 who were neonatal length, head circumference, and MUAC were
stillborn, 9 with major congenital anomalies, 51 who similar in all parity groups. These findings were similar
were not measured within 72 hours of birth, 69 who were for boys and girls, except that MUAC rose with in-
born preterm, 1 born to the mother who developed creasing parity in girls, and placental weight was higher in
gestational diabetes, and 1 whose mother had pregnancy- boys but not girls born to mothers of higher parity. These
induced hypertension. The babies lost to evaluation were results were unaffected by adjustment for seasonality of
similarly distributed in the 3 parity groups: primipara, 50 delivery and in para 2 and multiparous women for the
(20.4%), para 2, 40 (14.9%), and multipara, 47 (18.3%) interpregnancy interval.
(P = .261). Our analysis is limited to the remaining 631 Table III shows the relationships of parity, the infant’s
term live born babies (Table II). Birth weight rose with sex and gestational age at delivery, and parity-related
increasing parity; the mean was higher by 150 g for boys maternal nutritional and biochemical characteristics to
Joshi et al 787
Table III Multiple linear regression analysis of maternal parity and other maternal characteristics, fetal sex and gestational age, and
paternal size, as predictors of birth weight, neonatal subscapular skinfold thickness, and head circumference
Sum of neonatal
subscapular and triceps Neonatal head
Birth weight (G) skinfolds (mm) circumference (cm)
Univariate Multivariate Univariate Multivariate Univariate Multivariate
b P b P b P b P b P b P
Parity (3 groups) 47.42 ! .001 48.68 .02 0.27 ! .001 0.32 ! .01 0.025 .54 0.0027 .97
Gestation (wk) 109.42 ! .001 100.05 ! .01 0.19 ! .01 0.17 ! .01 0.44 ! .001 0.430 ! .001
Sex (1: male, 2: female) 116.94 ! .001 103.25 ! .01 0.17 .19 0.09 .53 0.61 ! .001 0.507 ! .001
Socioeconomic score 1.15 .58 3.92 .14 0.004 .67 0.007 .59 0.002 .79 0.0093 .32
Age (y) 12.19 ! .01 6.34 .33 0.07 ! .001 0.01 .66 0.01 .41 0.032 .17
Mother’s height (cm) 11.81 ! .001 8.88 ! .01 0.005 .97 0.005 .75 0.02 .03 0.020 .11
Prepregnant fat mass (kg) 26.43 ! .001 15.19 .03 0.06 ! .01 0.03 .29 0.07 ! .001 0.027 .26
Activity score 1.10 .04 1.75 .02 0.002 .40 0.002 .58 0.005 ! .01 0.0066 ! .01
Hemoglobin (g/L) 24.49 .02 31.94 ! .01 0.02 .11 0.04 .48 0.01 .78 0.017 .08
Leucocyte count (!109/L) 13.83 .09 1.27 .89 0.02 .67 0.02 .69 0.03 .39 0.0029 .92
Serum albumin (g/L) 147.04 ! .01 96.97 .10 0.78 ! .01 0.63 .02 0.47 ! .01 0.396 .05
Plasma ferritin (pmol/L) 2.15 .12 1.18 .48 0.01 .08 0.008 .34 0.002 .97 0.0044 .45
120 min glucose (mmol/L) 1.00 .22 0.33 .72 0.007 .04 0.01 ! .01 0.005 .11 0.0020 .52
120 min insulin (pmol/L) 0.15 .29 0.06 .67 0.004 .48 0.001 .32 0.0002 .72 0.0055 .72
Systolic blood 1.03 .51 0.09 .96 0.009 .16 0.002 .83 0.005 .35 0.0041 .52
pressure (mm Hg)
Husband’s height (cm) 4.79 .04 1.87 .51 0.006 .55 0.005 .73 0.02 ! .01 0.017 .09
Husband’s BMI (kg/m2) 16.73 .01 7.56 .25 0.03 .18 0.03 .37 0.03 .11 0.0010 .97
BMI, Body mass index.
newborn weight, sum of skinfold thicknesses, and head lower systolic blood pressure. However, the effect of
circumference. Neonatal measurements were strongly maternal parity on neonatal adiposity remained signifi-
related to gestational age at delivery and fetal sex (boys cant after adjusting for these characteristics.
were heavier and had larger head circumferences than Our population was rural and younger than in most
girls). As already described, birthweight and neonatal other studies; the mean age of the multiparous group was
skinfold thickness, but not head circumference, were only 24 years compared with 28 or more years else-
larger in babies born to mothers of higher parity. Several where.11 The lower adiposity of multiparous mothers in
of the other maternal variables were also related to Pune contrasts with other published studies, all of which
newborn size. Increased maternal height, larger prepreg- report higher body mass index and fat mass in multip-
nant fat mass, and lower hemoglobin and serum albumin arous women.11,12 Dietary intakes were similar in all
concentrations during pregnancy were all associated with parity groups, but multiparous women were more
a larger newborn size. In a multivariate analysis, which physically active. Their lower adiposity may be a result
included all maternal variables and which differed in the 3 of a high physical workload associated with both caring
parity groups and paternal size (height and body mass for the family and working on the farm, combined with
index), the effects of maternal parity on birthweight and repeated childbearing at a young age. The parity
neonatal adiposity were little changed and remained association was independent of socioeconomic status
statistically significant. which itself was quite homogeneous for this rural
farming population and also independent of seasonality
Comment of birth. In contrast to findings in other populations, in
which maternal insulin resistance and the risk of diabetes
In a large population-based study of rural Indian tend to increase with rising parity,13,14 multiparous
pregnant women, we have shown that increasing mater- mothers in our study had lower postload plasma glucose
nal parity is associated with increasing offspring adipos- and insulin concentrations.15 This is likely to be because
ity, there was increased neonatal soft tissue but not of their lower adiposity and higher levels of physical
increased skeletal size (length and head circumference). activity. Women of higher parity had lower blood
Multiparous mothers in our study were lighter, thinner, hemoglobin, plasma albumin, and ferritin concentra-
and more physically active than primiparous mothers. tions, which may result from greater plasma volume
They also had lower hemoglobin, serum albumin and expansion described in multiparous women.16 Against
plasma glucose, insulin and ferritin concentrations, and this, however, was the lack of association between parity
788 Joshi et al
and other hematologic parameters that would be ex- fat of babies born to mothers of higher parity could
pected to fall with plasma volume expansion, such as red indicate ‘‘better’’ fetal nutrition, perhaps as a result of
blood cell and platelet counts. Alternatively, lower ‘‘better’’ maternal physiologic adaptations to pregnancy
hemoglobin, albumin, and ferritin concentrations could (vasodilatation, plasma volume expansion, and immu-
indicate poorer nutritional status in the multiparous nologic changes) or to more effective placental transfer
group, perhaps as a result of successive pregnancies. of nutrients. Equally, it could indicate fetal ‘‘malnutri-
Finally, this may be a result of reverse causality, larger tion,’’ for example, the fetuses of mothers who are
fetuses extracting more nutrients from the mother. The undernourished as a result of successive pregnancies
lower leucocyte count in mothers of higher parity has may receive inadequate substrate for skeletal and/or
not, to our knowledge, been reported elsewhere. This lean mass growth, and thus incorporate more energy
could, again, be attributed to greater plasma volume into fat stores. The maternal metabolic variables that we
expansion or poorer nutritional status in multiparous measured could be considered proxies for maternal
women, or possibly a reduced immune response to adaptations: vasodilatation (blood pressure), plasma
pregnancy as a result of higher parity.17,18 We favor volume expansion (hemoglobin, serum albumin, and
the latter explanation especially because there was no plasma ferritin) and immune status (leucocyte count and
association of parity with red cell and platelet counts. plasma ferritin), and/or markers of maternal nutritional
As reported consistently in many different popula- status: (hemoglobin, serum albumin, and plasma ferri-
tions, the babies born to multiparous women in the tin). The vascular adaptation and nutritional deficiency
PMNS were heavier.1-3,19-21 There was no demonstrable interpretations would provide opposing explanations of
effect of parity on neonatal length and head circumfer- our findings. These measurements, however, did not
ence, but the offspring of multiparous women had explain the relationship of parity to birth size, which
thicker triceps and subscapular skinfolds, and were may be due to the relative crudeness of our measures, or
thus both peripherally and centrally more adipose. The point to some other aspect of parity as the mechanism
relationship between maternal parity and offspring mediating larger newborn size.
adiposity was continuous across parity though the The strengths of our study were that it was population
difference between babies of primiparous women and based, monitored mothers from before conception and
the rest was the most striking. Offspring of multiparous throughout pregnancy, recorded gestational age accu-
women also had larger abdominal circumferences, in- rately, measured maternal nutritional and metabolic
dicating either greater abdominal adiposity or larger status, and recorded detailed neonatal anthropometry.
visceral size. Few other studies have examined neonatal The main limitation was that the data were cross
size other than birthweight in relation to parity, but the sectional, and hence some unmeasured maternal factors
findings appear similar to ours. Among 4206 consecutive may vary between parity groups. Even in longitudinal
live births in Ethiopia, increasing parity predicted larger studies in the same mothers, it is difficult to separate the
birth weight but not length and head circumference.3 effects of parity from other confounders such as advanc-
Only a few studies have measured neonatal skinfold ing maternal age and secular changes in lifestyle and diet.
thicknesses: both showed, like us, that increasing parity In summary, the PMNS mothers were multiparous at
was associated with larger skinfold thickness or fat mass a younger age than in most other populations. They
in newborn infants.22,23 In the whole population of were lighter, less adipose, and more physically active
PMNS mothers, neonatal weight, skinfold thicknesses, than primiparous mothers, but did not differ in their
and abdominal circumference were strongly positively dietary intakes or circulating nutrient levels. Their
related to maternal fat mass.24 However, maternal fat babies were heavier and more centrally and peripherally
mass decreased with increasing parity in rural Indian adipose, independent of maternal and paternal body
mothers in contrast with other population.11,12 Thus, size, and measures of maternal nutritional and meta-
more adipose offspring born to (less adipose) multipa- bolic status. This appears to be therefore a direct effect
rous mothers presents a paradox. of parity. More research is required into maternal
We can only speculate as to the reasons for this physiologic changes associated with increasing parity.30
increased neonatal adiposity, and as to whether it is an Future epidemiologic studies of long-term outcomes
advantage or disadvantage. We have already reported should examine the relationship of birth order to adult
that as a population, the PMNS babies were consider- health outcomes, especially those related to adiposity
ably smaller at birth than white babies born in the such as insulin resistance and type 2 diabetes.
United Kingdom, but were relatively adipose.5 We
suggested that this may have survival advantages in
the neonatal period, but could lead to increased obesity Acknowledgments
in later life,25 and thus partly explain the South Asian
propensity for high percentage body fat, insulin re- We are grateful to the community and to the pregnant
sistance,26-28 and type 2 diabetes.29 The increased body women and their families for their cooperation. We
Joshi et al 789
thank Professor C. N. Hales and the laboratory staff 12. Roman H, Robillard PY, Verspyck E, Hulsey TC, Marpeau L,
of the Department of Clinical Biochemistry, Adden- Barau G. Obstetric and neonatal outcomes in grand multiparity.
brooke’s Hospital, Cambridge, for insulin assays; David 13. Kritz-Silverstein D, Barrett-Connor E, Wingard DL. The effect of
Collis and the staff of the Special Haematology Labo- parity on the later development of non-insulin–dependent diabetes
ratory, Southampton General Hospital, Southampton, mellitus or impaired glucose tolerence. N Engl J Med 1989;
UK for ferritin and folate assays; and Drs Chris Bates, 321:1214-9.
Glynn Harvey, and Jonathan Perkins at the Medical 14. Kritz-Silverstein D, Barrett-Connor E, Wingard DL, Friedlander
NJ. Relation of pregnancy history to insulin levels in older, non-
Research Council Resource Center for Human Nutri- diabetic women. Am J Epidemiol 1994;140:375-82.
tion Research, Cambridge, UK, for Vitamin C assays. 15. Hanley AJ, McKeown-Eyssen G, Harris SB, Hegele RA, Wolever
We also thank Dr A. D. Agate, Director, Agharkar TM, Kwan J, et al. Association of parity with risk of type 2
Research Institute and Dr V. N. Rao, Director, the diabetes and related metabolic disorders. Diabetes Care 2002;
KEM Hospital Research Centre for providing the 25:690-5.
16. van Buul EJ, Steegers EA, Jongsma HW, Eskes TK, Thomas CM,
facilities for this collaborative research. We acknowl- Hein PR. Haematological and biochemical profile of uncompli-
edge the contributions made to the study by Dr cated pregnancy in nulliparous women; a longitudinal study. Neth
Siddhivnayak Hirve, Dr Arun Kinare, Dr Monesh J Med 1995;46:73-85.
Shah, Dr Asit Natekar, Dr Manoj Chinchwadkar, Dr 17. Skjaerven R, Wilcox AJ, Lie RT, Irgens LM. Selective fertility and
Binu John, Dr Anuja Bisht, Dr Mahananda Bhavikatti, the distortion of perinatal mortality. Am J Epidemiol 1988;
128:1352-63.
Mrs Poonam Gupta, Mrs Parveen Bharucha, and Miss 18. Stene LC, Magnus P, Lie RT, Sevik O, Joner G. Norwegian
Vanessa Cox. Childhood Diabetes Study Group. Maternal age and paternal age
at delivery, birth order, and risk of childhood onset type 1 diabetes:
population based cohort study. BMJ 2001;323:369-71.
19. Chakraborty R, Roy M, Dass SR. Proportion with low birth
References weight infants in an Indian population and its relationship with
maternal age and parity. Hum Hered 1975;25:73-9.
1. Wilcox MA, Chang AMZ, Johnson IR. The effect of parity on 20. Dhall K, Bagga R. Maternal determinants of birth weight of North
birth weight using successive pregnancies. Acta Obstet Gynecol Indian babies. Indian J Pediatr 1995;62:333-44.
Scand 1996;75:459-63. 21. Malik S, Ghibiyal RG, Udani R, Waingankar P. Maternal
2. Juntunen KST, Laara EMJ, Kauppila AJI. Grand grand multi- biosocial factors affecting low birthweight. Indian J Pediatr 1997;
parity and birth weight. Obstet Gynecol 1997;90:495-9. 64:373-7.
3. Feleke Y, Enquoselassie F. Maternal age, parity and gestational 22. Catalano PM, Drago NM, Amini SB. Factors affecting fetal
age on the size of the newborn in Addis Ababa. East Afri Med J growth and body composition. Am J Obstet Gynecol 1995;
1999;76:468-71. 172:1459-63.
4. Rao S, Yajnik CS, Kanade A, Fall CHD, Margetts BM, Jackson 23. Guihard-Costa AM, Papiernik E, Kolb S. Maternal predictors
AA, et al. Intake of micronutrient rich foods in rural Indian of subcutaneous fat in the term newborn. Acta Paediatr 2004;
mothers and size of their babies at birth. J Nutr 2001;131:1217-24. 93:346-9.
5. Yajnik CS, Fall CHD, Coyaji KJ, Hirve SS, Rao S, Barker DJP, 24. Fall CHD, Yajnik CS, Rao S, Coyaji KJ, Shier RP. The effects of
et al. Neonatal anthropometry: the thin-fat Indian baby; the Pune maternal body composition before birth on fetal growth: the Pune
Maternal Nutrition Study. Int J Obesity 2003;27:173-80. Maternal nutrition and Fetal Growth Study. In: O’Brien PMS,
6. Rao S, Kanade A, Margetts BM, Yajnik CS, Lubree HG, Rege Wheeler T, Barker DJP, editors. Fetal programming: influences on
S, et al. Maternal activity in relation to birth Size in rural India: development and disease in later life. London: RCOG Press; 1999.
the Pune Maternal Nutrition Study. Eur J Clin Nutr 2003;57: 25. Neel JV. Diabetes mellitus: a ‘thrifty’ genotype rendered detri-
531-42. mental by ‘progress’? Am J Hum Genet 1962;14:353-62.
7. Durnin JV, Womersley J. Body fat assessed from total body 26. McKeigue PM, Shah B, Marmot MG. Relation of central obesity
density and its estimation from skinfold thickness: measurements and insulin resistance with high diabetes prevalence and cardio-
on 481 men and women aged from 16 to 72 years. Br J Nutr 1974; vascular risk in South Asians. Lancet 1991;337:382-6.
32:77-97. 27. Banerji MA, Faridi N, Atluri R, Chaiken RL, Lebovitz HE. Body
8. Hadlock FP. Sonographic estimation of fetal age and weight. Fetal composition, visceral fat, leptin and insulin resistance in Asian
Ultrasound 1990;28:39-50. Indian men. J Clin Endocrinol Metab 1999;84:137-44.
9. Pareek U, Trivedi G. Reliability and validity of socio-economic 28. Chandalia M, Abate N, Garg A, Stray-Gundersen J, Grundy SM.
scales. Indian J Appl Psychol 1964;1:34-40. Relationship between generalized and upper body obesity to
10. Matthews DR, Hosker JP, Rudenski AS, Naylor BA, Treacher insulin resistance in Asian Indian men. J Clin Endocrinol Metab
DF, Turner RC. Homeostasis model assessment: insulin resistance 1999;84:2329-35.
and beta-cell function from fasting glucose and insulin concen- 29. Fall CHD, Stein CE, Kumaran K, Cox V, Osmond C, Barker DJP,
trations in man. Diabetologia 1985;28:412-9. et al. Size at birth, maternal weight, and type 2 diabetes in South
11. Rodrigues MLCF, Da Costa THM. Association of maternal India. Diabetic Med 1998;15:220-7.
experience and changes in adiposity measured by BMI, waist: 30. Odent M, Tsufino J. Studies exploring health in relation to
hip ratio and percentage body fat in urban Brazilian women. Br intrauterine life should look at birth order [letter]. BMJ
J Nutr 2001;85:107-14. 1997;314:1416.
www.ajog.org
Misoprostol induces cervical nitric oxide release in

pregnant, but not in nonpregnant, women
Mervi Väisänen-Tommiska, MD, Tomi S. Mikkola, MD, PhD,
Olavi Ylikorkala, MD, PhD*
Department of Obstetrics and Gynecology, Helsinki University Central Hospital, Helsinki, Finland
KEY WORDS Objective: The cells of the human uterine cervix synthesize nitric oxide, which may be a factor in
Cervical ripening cervical ripening. We studied the effect of misoprostol on cervical nitric oxide release in
Induction of labor nonpregnant and pregnant women.
Nitric oxide Study design: Seventy-two nonpregnant (n = 15) and pregnant (n = 57; 26 in early pregnancy,
metabolite in 31 in late pregnancy) women were treated with either vaginal misoprostol (n = 54) or vaginal
cervical fluid placebo (n = 18). The dose of misoprostol was 400 mg in nonpregnant and early pregnancy
Placebo-controlled group, and 25 mg in late pregnancy group. Serial cervical fluid samples, collected before and up to
Pregnancy termination 3 hours after misoprostol/placebo, were assessed for the concentration of nitric oxide metabolites
Prostaglandin by means of the Griess reaction.
Results: Placebo had no effect on cervical fluid nitric oxide metabolite level. In 1 to 3 hours,
misoprostol induced 4.3- to 5.2-fold elevations in cervical fluid Nox concentrations in early
pregnancy (P ! .01), and 4.4- to 18.2-fold elevations in late pregnancy (P ! .01), but these
responses did not differ significantly from each other. Misoprostol had no effect on cervical fluid
nitric oxide metabolites in nonpregnant women. There was a trend towards a relationship
between cervical nitric oxide stimulation after misoprostol and cervical ripening.
Conclusion: Vaginal misoprostol stimulates cervical nitric oxide release in pregnancy. This
suggests a joint action of nitric oxide and prostaglandins in cervical ripening.
It is well established that various cells in the human appears to be iNOS.3,4 Sex steroids,5,6 cytokines,7,8 and
uterine cervix express and produce nitric oxide (NO) by prostaglandins9,10 are involved in the regulation of NOS
way of various nitric oxide synthases (NOS; neuronal, activities, and this regulation may operate in both dir-
inducible, and endothelial).1,2 It is impossible to delin- ections.4,11 Nitric oxide may take part in various repro-
eate which NOS in the cervix is quantitatively and ductive processes,4 one of the most established effects
physiologically most important, but during pregnancy it being cervical ripening.5,12,13 It is not known by which
mechanism NO causes this effect, but it may involve
various matrix metalloproteases (MMPs) that degrade
* Reprint requests: Olavi Ylikorkala, MD, FRCOG, Professor,
Department of Obstetrics and Gynecology, Helsinki University Cen-
cervix-stabilizing glycosaminoglycans and collagens.4,14
tral Hospital, PO Box 140, FIN-00029 HUS, Finland. Misoprostol, a prostaglandin E1 analogue, is rou-
E-mail: olavi.ylikorkala@hus.fi tinely used for cervical ripening before termination of
doi:10.1016/j.ajog.2005.02.084
Väisänen-Tommiska, Mikkola, and Ylikorkala 791
Table I Clinical characteristics, the presence and concentration of nitric oxide metabolites in the cervical fluid at baseline of different
study populations (mean G SD, or %; nitric oxide metabolite concentration median, 95% CI)
Study group
Nonpregnant Early pregnancy Late pregnancy
Number of women 15 26 31
Age (y) (range) 39.9 G 8.4*,y (24–52) 28.9 G 7.7* (20–43) 31.4 G 4.2y (23–39)
Nulliparous n (%) 7 (47%) 10 (38%) 13 (42%)
Gestational age (wk) – 8.8 G 1.8 40.4 G 1.7
Nitric oxide metabolite
Detectable n (%) 14 (93%)z 17 (65%)z 25 (82%)z
Concentration (mmol/L) CI 18.6x (15.0–75.2) 11.3x (!3.8–22.0) 52.6x (!3.8–147.4)
* P = .0002.
y
P = .0008.
z
P = .01, compared to early pregnancy.
x
P = .02, compared to late pregnancy.
Table II The concentrations of nitric oxide metabolites (Nox, mmol/L) in cervical fluid of study groups before and after vaginally
administrated misoprostol or placebo (median, 95% CI)
Nonpregnant Early pregnancy Late pregnancy Early and late
misoprostol 0.4 mg misoprostol 0.4 mg misoprostol 0.025 mg pregnancy
(n = 15) (n = 17) (n = 22) placebo (n = 18)
At 0 min 18.6 15.0-75.2 7.1 !3.8-24.4 49.5 !3.8-135.2 31.4 !3.8-67.7
At 60 min 19.2 14.6-58.3 29.9* 17.3-34.8 286.3z 39.7-395.4 39.6 !3.8-127.4
At 120 min 22.0 12.9-52.1 24.4* 12.0-36.5 238.8z 99.5-495.0 33.4 14.2-56.8
At 180 min 23.3 13.2-45.9 26.3y 12.8-41.9 390.6x 103.2-831.1 31.7 10.3-57.3
All P values compared to group specific level at 0 min.
* P = .002.
y
P = .01.
z
P = .003.
x
P = .0005.
early pregnancy and in induction of labor.15-18 Although (Table I). Each woman received written and oral infor-
it can be administered orally, rectally, or buccally, the mation on the purpose and conduct of the study, and
vaginal route of administration is favored in clinical written consent was obtained from each of them.
routine owing to its superior clinical efficacy and lack of The nonpregnant group (n = 15) consisted of healthy
gastrointestinal side effects.16,17 The sensitivity of the women admitted to hospital for tubal sterilization
cervix to misoprostol is enhanced during pregnancy and, within 5 to 14 days after the cessation of preceding
therefore, smaller doses of misoprostol are given in late menstruation (follicular phase). They were healthy and
pregnancy (around 25–50 mg) than in early pregnancy did not use intrauterine contraceptive devices or con-
(around 400–800 mg).16,17 It is not exactly known by traceptive pills. The early pregnancy group comprised
which mechanism misoprostol ripens the cervix, but women (n = 26) who sought termination of pregnancy
stimulatory effects on MMPs and/or on cytokines in the between 7 and 12 weeks of gestation for socioeconomic
cervix may be involved.19,20 reasons. They had not experienced any bleeding or other
Nitric oxide is rapidly converted to stable nitrate and signs of threatened abortion, and on the basis of findings
nitrite (Nox), the overall level of which in cervical fluid in transvaginal ultrasonography, pregnancy was viable.
reflects cervical NO release.21,22 We studied here the effect The late pregnancy group consisted of women (n = 31)
of misoprostol, as used in clinical routine, on cervical NO requiring labor induction because of a gestational age of
release in nonpregnant and pregnant women. 41 weeks or more (based on early pregnancy ultrasono-
graphic examination) (n = 25), or because of cholestasis
of pregnancy (n = 6); the cervixes of these women were
Material and methods unripe (Bishop score !5). No study subject had any
signs of cervical infection, abnormal Pap smear, or a
With permission of the local committee on ethics, we positive Chlamydia test result. No pelvic or cervical
studied 72 healthy nonpregnant and pregnant women palpation was allowed within 4 hours before, and during
792 Väisänen-Tommiska, Mikkola, and Ylikorkala
Figure The levels of nitric oxide metabolites (Nox) in cervical fluid (percentages of initial values, mean G SE) at 1, 2, and 3 hours
after vaginal administration of misoprostol in nonpregnant women (-,-), in women in early pregnancy (-O-), in women in late
pregnancy (-B-), and after placebo in women in early or late pregnancy (->-). For the dose of misoprostol, see Material and methods.
= P ! .01, * = P ! .001 in comparison with baseline. There was no statistically significant difference between women in early and
late pregnancy, and placebo had no effect.
the first 3 study hours because even gentle palpation there for precisely 20 seconds, as described previ-
may cause a release of cervical NO.21 ously.21,22,24 The swab was then washed in 1.5 mL of
Misoprostol (Cytotec, Pharmacia Ltd, Morpeth, UK) physiologic saline for 2 minutes, and the saline samples
or placebo (Placebo, pregelatinized starch, 136 mg, were kept frozen at 21(C until assayed for Nox by
cellulose microcrystalline [90 mol/L], 114 mg, Schering, means of the Griess reaction.21,22,24 Macroscopically
Turku, Finland) was administered to the posterior bloody cervical fluid samples (n = 12) were discarded,
fornix of the vagina under visual control. The dose of leaving 276 samples for analysis. The detection limit of
misoprostol was 400 mg in nonpregnant subjects and in the assay was 3.8 mmol/L, and the intra-assay and
the early pregnancy group, and 25 mg in the late interassay coefficients of variation were 1.6% and
pregnancy group; this dosage was selected on the basis 2.4%, respectively. To reduce the impact of interassay
of the results of previous trials.16-18 The study was variation, the samples were assayed in only 7 batches.
conducted as a placebo-controlled trial as far as possible Cervical ripening was assessed by one investigator
in our clinical setting. The misoprostol group consisted (M V-T). In the nonpregnant and early pregnancy
of 15 nonpregnant and 39 pregnant women (17 women groups, the cervix was judged to be softened if a Hegar
in early pregnancy and 22 in late pregnancy), and the dilator of size 7 could be introduced into the cervix
placebo group of 18 pregnant women (9 parous women without force 3 hours after the application of miso-
with early pregnancies and 9 women with late pregnan- prostol or placebo; otherwise, it was tight. In the late
cies); no placebo was used in nonpregnant women. pregnancy group, cervical ripening was assessed by
Cervical fluid samples were collected before the means of Bishop scores at 4 hours before and 3 hours
application of misoprostol or placebo, and at 1, 2, and after the application of misoprostol or placebo.
3 hours after its application. This study period was After the study period of 3 hours, the women were
judged to be sufficiently long because after vaginal treated according to routine principles in our hospital.
application, misoprostol reaches its peak level in plasma The nonpregnant group underwent laparoscopic tubal
in 80 minutes.23 The samples (288 in all) were collected sterilization, and the early pregnancy group vacuum
by introducing a Dacron swab (DuPont, Wilmington, curettage, both under general anesthesia. Women in
Del) into the cervix under visual control, and keeping it late pregnancy received 1 to 6 additional doses of
Table III Changes in the cervical status and in nitric oxide metabolite (Nox) level to misoprostol or placebo in nonpregnant women
and in women with early pregnancy
Misoprostol 0.4 mg Placebo
Nonpregnant (n = 15) Early pregnancy (n = 17) Early pregnancy (n = 9)
Tight Softened Tight Softened Tight Softened
Number of women 10 5 (33%) 7 10 (59%) 6 3 (33%)
Cervical fluid Nox
Baseline level 17.6 19.3 23.9 4.0 15.0 19.6
Median (95%CI) 15.0-75.2 14.6-58.3 !3.8-39.3 !3.8-30.3 !3.8-29.3 !3.8-41.4
Change from the 5 8 318 438 15 5
baseline at 3 hours
(%) (median)
Women with group 50 60 29* 70* 50 67
specific median Nox
response or above (%)
* P = .09.
misoprostol at 4-hour intervals until initiation of labor; The administration of misoprostol was followed by
after the 3-hour study period, no additional cervical significant elevations in cervical fluid Nox concentra-
fluid samples were collected. tions in 1 to 3 hours both in early (4.3- to 5.3-fold rise)
Categorical data were analyzed by Fisher exact and in late (4.4- to 18.2-fold rise) pregnancy, but these
probability test, and repeated measures analysis of responses did not differ significantly from each other
variance (ANOVA). Medians with their 95% CIs were (Table II, Figure). The response of Nox to misoprostol
used to describe absolute Nox levels. The Nox values in women with cholestasis of pregnancy was similar to
were analyzed using nonparametric tests, such as that in women without it. Nor did the women with post-
the Mann-Whitney U test, Kruskal-Wallis one-way term pregnancy differ in Nox response to misoprostol
ANOVA, and Spearman rank correlation tests. All tests from the women with term pregnancy. Parity did not
were two-sided and carried out using NCSS 2000 affect the response of Nox to misoprostol in any group.
software (NCSS, Inc, Kaysville, Utah). Values of P ! Misoprostol had no effect on cervical fluid Nox concen-
.05 were considered statistically significant. A concentrations in nonpregnant women (P = .68), even if ana-
tration below the detection limit (3.8 mmol/L) was given lyzed on a relative scale (Table II, Figure). Placebo had
an arbitrary value of 3.7 mmol/L. To better describe no effect on cervical fluid Nox levels (P = .50) when
treatment-induced changes in cervical fluid Nox analyzed in all pregnant women (Table II, Figure), or in
levels, we also present the Nox data as percentages of those in early and late pregnancy separately.
pretreatment values. After misoprostol, the cervix was softened in 5 of 15
nonpregnant women (33%), and in 10 of 17 women in
early pregnancy (59%) (Table III). Parity was no factor
Results in this regard because 2 of 7 nulliparous (29%) and 3 of
8 parous (38%) women in the nonpregnant group, and 5
On average, the nonpregnant group was older than the of 10 nulliparous (50%) and 5 of 7 parous (71%) women
pregnant groups (Table I). The baseline cervical fluid in the early pregnancy group had experienced cervical
Nox level was detectable less often in the early preg- softening after misoprostol application. However, use of
nancy group than in the nonpregnant and late preg- placebo was also associated with a softened cervix
nancy groups, but the levels were higher in the late (Hegar 7) in 3 of 9 women in early pregnancy (33%).
pregnancy group than in the nonpregnant or early The baseline levels of Nox, and their responses to
pregnancy groups (Table I). Taking all subjects into misoprostol or placebo, showed no statistically signifi-
account (n = 72), parous women (median 29.3 mmol/L, cant differences between women with softened or tight
95% CI 15.0–48.9) had higher (P = .02) cervical fluid cervixes (Table III). However, in the early pregnancy
Nox concentrations than nulliparous women (median group, elevation of Nox levels (median or more) tended
18.6 mmol/L, 95% CI !3.8–28.0), whereas in no sub- to occur more often (P = .09) in women with a softened
group was parity a factor regarding Nox. The age of the cervix (70%) than with a tight cervix (29%) (Table III).
subject, the duration of amenorrhea, or cholestasis of In the late pregnancy group, elevation of Nox levels
pregnancy were not determinants regarding Nox levels. following misoprostol was not related to changes in
Bishop score (P = .33), but the median cervical fluid dose of misoprostol, or a more prolonged study period,
Nox concentration per 1 Bishop score rose 4-fold (97.7 would have resulted in a release of NO, but we regard it
vs 24.3; P = .04) after misoprostol. The change in as unlikely, although cervical ripening after misoprostol
cervical fluid Nox levels following misoprostol bore no may take 10 hours in nonpregnant women.25,26
relationship to the final dose of misoprostol ultimately In contrast to nonpregnant women, cervical Nox
needed for labor induction, to the interval between concentrations responded to misoprostol by showing
sampling and initiation of labor, or to the need of significant and consistent rises as early as after 1 hour in
cesarean section (18% in the whole series) or vacuum women in early pregnancy. Thus, in early pregnancy, an
extraction (9%). agent appears in the blood, or the cervix, or in both,
which makes cervical NO release responsive to miso-
prostol. Our data do not allow us to identify this agent,
Comment but it may come from the decidua, placenta, corpus
luteum, or from all of them. It is unlikely that this agent
We show here that cervical NO release is induced by a is progesterone because progesterone inhibits the release
vaginally administered, routine dose of misoprostol in of cervical NO in pregnancy,4,5,12,27 and low progester-
pregnant, but not in nonpregnant women. Cervical NO one and elevated cervical Nox levels have been observed
release was assessed by the levels of cervical fluid to precede the onset of clinical abortion in women with
Nox,21,22,24 a method which is reproducible and reflects failed early pregnancy.24 Other candidates could be
accurately the availability of NO.21 Nonpregnant human chorionic gonadotropin (hCG), or inhibins,
women were slightly older than pregnant women, but which appear at high levels during pregnancy, and
because there was no relationship between the age and which have been shown to be involved in the synthesis
Nox levels in the present series, this bias in age was no of NO in various study models.4,5,28,29 Moreover, it is
determinant for the differences in Nox levels observed. not known if this factor affects NO synthesis directly, or
The NO stimulation, which was relatively similar in indirectly, through changes in cytokines4,30 and/or other
early and in late pregnancy, could be accomplished with regulators.8,9,31
a much smaller dose (only 6%) of misoprostol in late The sensitivity of cervical NO release to misoprostol
pregnancy. This effect of misoprostol was specific be- became enhanced from early to late pregnancy because
cause placebo had no effect on cervical NO release either we observed relatively higher and more persistent Nox
in early or in late pregnancy. We could not use placebo responses in late pregnancy, although the dose of
in nonpregnant women, but in view of the lack of effect misoprostol was only 6% of that used in early preg-
of placebo in the pregnant groups, and the lack of effect nancy. The causes of increased sensitivity of NO syn-
of misoprostol in nonpregnant women, a placebo thesis to misoprostol in late pregnancy are not known,
effect on cervical fluid Nox in nonpregnant women but they may be, in part, the same as those responsible
seems unlikely. Our data do not allow us to deduce for the appearance of the sensitivity of NO in early
which isoform of NOS was primarily responsible for pregnancy. A role of hCG in the increased release of NO
misoprostol-induced NO stimulation. However, we be- appears unlikely in late pregnancy because circulating
lieve that it is iNOS, since it is abundantly expressed in levels of hCG fall with advancing gestation. It can be
cervical cells,2,3 and is generally regarded as the most possible that the number of progesterone receptors in
significant NOS in the cervix during pregnancy.4 the cervix becomes reduced already before the initiation
In one of our previous studies,21 fewer nonpregnant of labor,32 and thus, local progesterone insufficiency
women (46%) showed detectable cervical fluid Nox than may ensue. This could be one explanation for the
in the present work (93%), although we employed the stimulation of cervical NO in late pregnancy, and for
same methods. Biological variation in NOS could be one its higher sensitivity to misoprostol. We should also note
explanation for this discrepancy, especially because the that women with established post-term pregnancy were
total number of nonpregnant women studied was mod- characterized with reduced cervical Nox levels.22 There-
est (n = 26). Another explanation, and perhaps a more fore, a possibility exists that these women had a lower
likely one, could be a different cycle phase at the time of Nox capacity than women delivering at term. Against
the sampling; the luteal phase in our previous work21 this speculation speaks the fact that there was no
and the follicular phase in the present study. There is relation between the levels/responses of Nox and gesta-
ample evidence of the inhibitory role of progesterone in tional day, and that women with hepatosis and near-
the release of cervical NO in pregnancy.4 Our data imply term pregnancy showed similar Nox responses as did
that a lack of progesterone made cervical fluid Nox women who were past their dates.
more often detectable in nonpregnant women in the There is abundant evidence that prostaglandins can
present study. However, despite the high detection rate, stimulate the release of NO.4,10,33,34 Our data on the
cervical fluid Nox levels failed to respond to misoprostol significant elevations in NO release following miso-
in 3 hours in these women. We cannot deduce if a higher prostol show that this mechanism may also operate in
the human cervix, perhaps primarily through iNOS. 4. Maul H, Longo M, Saade GR, Garfield RE. Nitric oxide and its
On the other hand, it is also known that NO can induce role during pregnancy: from ovulation to delivery. Curr Pharm
Des 2003;9:359-80.
the expression of cyclo-oxygenase 2,4 and thereby in- 5. Chwalisz K, Garfield RE. Regulation of the uterus and cervix
crease the release of endogenous prostaglandins. Thus, during pregnancy and labor. Role of progesterone and nitric oxide.
misoprostol could perhaps initiate a chain reaction in Ann N Y Acad Sci 1997;828:238-53.
the cervix of pregnant women; the initial NO stimula- 6. Lyons C, Beharry K, Akmal Y, Attenello F, Nageotte MP. In vitro
tion caused by misoprostol is followed by an endoge- response of prostaglandin E2 receptor (EP3) in the term pregnant
rat uterus and cervix to misoprostol. Prostaglandins Other Lipid
nous release of prostaglandins triggered by NO. Both Mediat 2003;70:317-29.
these responses facilitate the cervical ripening essential 7. Sennstrom MB, Ekman G, Westergren-Thorsson G, Malmstrom
for safe delivery. A, Bystrom B, Endresen U, et al. Human cervical ripening, an
The significance of misoprostol-induced NO release inflammatory process mediated by cytokines. Mol Hum Reprod
remains open. However, we found that cervical ripening, 2000;6:375-81.
8. Kelly RW. Inflammatory mediators and cervical ripening.
as assessed by clinical tests, tended to be more pro- J Reprod Immunol 2002;57:217-24.
nounced in the presence of greater NO release. This may 9. Ekerhovd E, Weijdegard B, Brannstrom M, Mattsby-Baltzer I,
imply that NO release, per se, could have potentiated the Norstrom A. Nitric oxide induced cervical ripening in the human:
effect of misoprostol. This is supported by the results of Involvement of cyclic guanosine monophosphate, prostaglandin
clinical studies showing that NO donors, such as iso- F(2 alpha), and prostaglandin E(2). Am J Obstet Gynecol
2002;186:745-50.
sorbide mononitrate and glyceryl trinitrate, result in 10. Goharkhay N, Wing DA, Pan V, McCausland V, Hanna M,
cervical ripening in early and late pregnancy.35-37 It is Naidu YM, et al. The expression of EP3-6 and inducible nitric
also possible that misoprostol (labor)-induced release of oxide synthase messenger RNA are correlated in pregnant and
cervical NO may not only facilitate cervical opening, but misoprostol-treated but not in nongravid or menopausal myome-
also protect the cervical canal against ascending infec- trium. Am J Obstet Gynecol 2002;186:1202-6.
11. Gookin JL, Duckett LL, Armstrong MU, Stauffer SH, Finnegan
tions during miscarriage or labor. This speculation is CP, Murtaugh MP, et al. Nitric oxide synthase stimulates prosta-
supported by the fact that high amounts of iNOS/NO glandin synthesis and barrier function in C. parvum-infected
have been reported to kill bacteria and parasites in porcine ileum. Am J Physiol Gastrointest Liver Physiol
various studies.4 A dual role of cervical NO release is 2004;287:G571-81.
physiologically meaningful in women before abortion or 12. Chwalisz K, Garfield RE. Nitric oxide as the final metabolic
mediator of cervical ripening. Hum Reprod 1998;13:245-8.
labor. 13. Shi L, Shi SQ, Saade GR, Chwalisz K, Garfield RE. Studies of
In conclusion, vaginal misoprostol stimulates cervical cervical ripening in pregnant rats: effects of various treatments.
NO release in pregnancy. This suggests a joint action of Mol Hum Reprod 2000;6:382-9.
NO and prostaglandins in cervical ripening. 14. Yoshida M, Sagawa N, Itoh H, Yura S, Korita D, Kakui K, et al.
Nitric oxide increases matrix metalloproteinase-1 production in
human uterine cervical fibroblast cells. Mol Hum Reprod
Acknowledgments 2001;7:979-85.
15. Song J. Use of misoprostol in obstetrics and gynecology. Obstet
Gynecol Surv 2000;55:503-10.
We thank Eira Halenius, RN, for excellent technical
16. Goldberg AB, Greenberg MB, Darney PD. Misoprostol and
assistance. The Clinical Research Fund of Helsinki pregnancy. N Engl J Med 2001;344:38-47.
University Central Hospital, the Finnish-Norwegian 17. Goldberg AB, Carusi DA, Meckstroth KR. Misoprostol in
Medical Foundation, the Emil Aaltonen Foundation, gynecology. Curr Womens Health Rep 2003;3:475-83.
the Research Foundation of Biomedicum, Helsinki, the 18. Alfirevic Z. Oral misoprostol for induction of labour. Cochrane
Database of Systematic Reviews [database online]. Updated 2004.
Research Foundation of Orion Corporation, and the
19. Denison FC, Calder AA, Kelly RW. The action of prostaglandin
Finnish Medical Association, Duodecim supported this E2 on the human cervix: Stimulation of interleukin 8 and
study. inhibition of secretory leukocyte protease inhibitor. Am J Obstet
Gynecol 1999;180:614-20.
20. Ledingham MA, Denison FC, Kelly RW, Young A, Norman
References JE. Nitric oxide donors stimulate prostaglandin F(2alpha) and
inhibit thromboxane B(2) production in the human cervix during
1. Ekerhovd E, Brannstrom M, Weijdegard BBS, Norstrom A. Nitric the first trimester of pregnancy. Mol Hum Reprod 1999;5:
oxide synthases in the human cervix at term pregnancy and effects 973-82.
of nitric oxide on cervical smooth muscle contractility. 21. Vaisanen-Tommiska M, Nuutila M, Aittomaki K, Hiilesmaa V,
Am J Obstet Gynecol 2000;183:610-6. Ylikorkala O. Nitric oxide metabolites in cervical fluid during
2. Ledingham MA, Thomson AJ, Young A, Macara LM, Greer IA, pregnancy: further evidence for the role of cervical nitric oxide in
Norman JE. Changes in the expression of nitric oxide synthase in cervical ripening. Am J Obstet Gynecol 2003;188:779-85.
the human uterine cervix during pregnancy and parturition. 22. Vaisanen-Tommiska M, Nuutila M, Ylikorkala O. Cervical nitric
Mol Hum Reprod 2000;6:1041-8. oxide release in women postterm. Obstet Gynecol 2004;103:
3. Tschugguel W, Schneeberger C, Lass H, Stonek F, Zaghlula MB, 657-62.
Czerwenka K, et al. Human cervical ripening is associated with an 23. Zieman M, Fong SK, Benowitz NL, Bantskter D, Darney PD.
increase in cervical inducible nitric oxide synthase expression. Biol Absorption kinetics of misoprostol with oral or vaginal adminis-
Reprod 1999;60:1367-72. tration. Obstet Gynecol 1997;90:88-92.
24. Vaisanen-Tommiska M, Mikkola TS, Ylikorkala O. Increased 31. Osman I, Young A, Ledingham MA, Thomson AJ, Jordan F,
release of cervical nitric oxide in spontaneous abortion before Greer IA, et al. Leukocyte density and pro-inflammatory cytokine
clinical symptoms: a possible mechanism for preabortal cervical expression in human fetal membranes, decidua, cervix and myo-
ripening. J Clin Endocrinol Metab 2004;89:5622-6. metrium before and during labour at term. Mol Hum Reprod
25. Preutthipan S, Herabutya Y. Vaginal misoprostol for cervical 2003;9:41-5.
priming before operative hysteroscopy: a randomized controlled 32. Stjernholm-Vladic Y, Wang H, Stygar D, Ekman G, Sahlin L.
trial. Obstet Gynecol 2000;96:890-4. Differential regulation of the progesterone receptor A and B in the
26. Fernandez H, Alby JD, Tournoux C, Chauveaud-Lambling A, human uterine cervix at parturition. Gynecol Endocrinol
DeTayrac R, Frydman R, et al. Vaginal misoprostol for cervical 2004;18:41-6.
ripening before operative hysteroscopy in pre-menopausal women: 33. Ogando D, Farina M, Ribeiro ML, Perez Martinez S, Cella M,
a double-blind, placebo-controlled trial with three dose regimens. Rettori V, et al. Steroid hormones augment nitric oxide synthase
Hum Reprod 2004;19:1618-21. activity and expression in rat uterus. Reprod Fertil Dev
27. Chwalisz K, Shao-Qing S, Garfield RE, Beier HM. Cervical 2003;15:269-74.
ripening in guinea-pigs after a local application of nitric oxide. 34. Timoshenko AV, Lala PK, Chakraborty C. PGE2-mediated
Hum Reprod 1997;12:2093-101. upregulation of iNOS in murine breast cancer cells through the
28. Faletti A, Perez Martinez S, Perotti C, de Gimeno MA. Activity of activation of EP4 receptors. Int J Cancer 2004;108:384-9.
ovarian nitric oxide synthase (NOs) during ovulatory process in 35. Thomson AJ, Lunan CB, Cameron AD, Cameron IT, Greer IA,
the rat: Relationship with prostaglandins (PGs) production. Nitric Norman JE. Nitric oxide donors induce ripening of the human
Oxide 1999;3:340-7. uterine cervix: a randomised controlled trial. BJOG 1997;104:1054-7.
29. Reis FM, Faletti A, Luisi S, Bifulco G, Cauci S, Quadrifoglio F, 36. Chanrachakul B, Herabutya Y, Punyavachira P. Randomized
et al. High concentrations of inhibin A and inhibin B in ovarian comparison of glyceryl trinitrate and prostaglandin E2 for cervical
serous cystadenoma: relationship with oestradiol and nitric oxide ripening at term. Obstet Gynecol 2000;96:549-53.
metabolites. Mol Hum Reprod 2000;6:1079-83. 37. Ekerhovd E, Bullarbo M, Andersch B, Norstrom A. Vaginal
30. Estevez A, Farina M, Franchi A, Johnson C, Vega M, Motta AB. administration of the nitric oxide donor isosorbide mononitrate
Interleukin-1beta up-regulates nitrite production: effects on ovar- for cervical ripening at term: a randomized controlled study.
ian function. Nitric Oxide 2004;10:92-100. Am J Obstet Gynecol 2003;189:1692-7.
www.ajog.org
N-acetyl-transferase phenotype and risk for preeclampsia

Petra L. M. Zusterzeel,a René H. M. te Morsche,b Maarten T. M. Raijmakers,b
Eva Maria Roes,a Wilbert H. M. Peters,b Régine P. M. Steegers-Theunissen,c,d
Eric A. P. Steegersd,*
Departments of Obstetrics and Gynecology,a Gastroenterology,b and Epidemiology and Biostatistics,c Radboud
University Nijmegen Medical Center, Nijmegen, The Netherlands; and the Department of Obstetrics and Gynecology,
Erasmus University Medical Center Rotterdam, Rotterdam, The Netherlandsd
Received for publication April 26, 2004; revised November 30, 2004; accepted January 10, 2005
KEY WORDS Objective: This study was undertaken to determine whether the N-acetyltransferase (NAT)
N-acetyltransferase phenotype contributes to the susceptibility for the development of preeclampsia.
Biotransformation Study design: The NAT acetylator status was determined by measuring urinary caffeine
Preeclampsia metabolites in 134 nonpregnant women with a history of preeclampsia and in 109 control women
HELLP syndrome with uncomplicated pregnancy. The c2 and logistic regression analyses were used for statistical
evaluation of differences in acetylator status.
Results: Significantly more fast acetylators were found among the women with a history of
preeclampsia (46.3%) than among the controls (25.4%). Fast acetylators showed an odds ratio of
2.5 (95% CI 1.4-4.3) for preeclampsia. No differences in the acetylator status were found between
women with a history of preeclampsia only and those with the HELLP syndrome as well.
Conclusion: The fast NAT acetylator status, which may result in altered NAT detoxifaction
capacity, is associated with preeclampsia.
Preeclampsia represents one of the most serious Many enzymes, including the cytochromes P450,5
medical disorders complicating human pregnancies. glutathione S-transferases,6 epoxide hydrolases,5,7 and
However, factors responsible for the initiation and pro- N-acetyltransferases (NATs)5 are involved in the me-
gression of the disease process remain elusive. There is tabolism of endogenous and exogenous compounds.
growing evidence that an imbalance between toxic com- Women with an altered enzyme activity for glutathione
pounds, such as lipid peroxides and oxygen free radicals, S-transferase Pl or epoxide hydrolase were shown to
and detoxifying and scavenging substances might con- have an increased risk for preeclampsia.8,9 NAT has
tribute to the pathophysiology of preeclampsia.1-4 never been studied in this context.
NATs are phase 2 enzymes, and 2 distinct isoenzymes
are known: NAT1 and NAT2. Both NAT enzymes are
involved in bioinactivation (N-acetylation) as well as
* Reprint requests: Eric A. P. Steegers, MD, PhD, Department of
Obstetrics and Gynecology Erasmus University Medical Center
bioactivation (O- and N,O-acetylation), and use co-
Rotterdam, PO Box 2040, 3000 CA Rotterdam, The Netherlands. enzyme A as a cofactor. Reactions catalyzed by NAT
E-mail: e.a.p.steegers@erasmusmc.nl generally result in less toxic metabolites (detoxification
doi:10.1016/j.ajog.2005.01.012
798 Zusterzeel et al
Figure 1 Schematic overview of caffeine metabolism. CYP1A2, Cytochrome P450 1A2; NAT, N-acetyltransferase; XO, xanthine
oxidase; ?, unknown substratum.
reactions). However, several derivates, notably in concert an increased risk for (smoking-related) lung or bladder
with oxidative metabolism by cytochrome P450-mediated cancer and endometriosis, whereas a fast acetylator status
reactions, are substrates for additional metabolism by has been associated with an increased risk for colon
NAT and result in more reactive compounds.5 Known cancer.11-14
exogenous substrates of NAT isoenzymes include aryl- The objective of this study was to investigate whether
amines and arylhydroamines, found in cigarette smoke,10 the NAT2 acetylator status, as determined by measure-
whereas NAT2 is involved in the metabolism of caffeine. ment of caffeine metabolites in urine, was associated
A slow NAT2 acetylator status has been associated with with the risk for preeclampsia.
Zusterzeel et al 799
Figure 2 Distribution histogram showing the urinary AAMU/1X ratios for women with a history of preeclampsia (A) and
controls (B). Arrow, cutoff point separating slow from fast NAT2 acetylators. AAMU, 5-acetylamino-6-amino-3-methyluracil;
1X, 1-methylxanthine.
Material and methods secondary, or intermediate vocational education) or

high (higher secondary, higher vocational, or university
The study group consisted of 134 nonpregnant women education).
who had been previously admitted for preeclampsia to the The institutional review board approved the study
antenatal wards or intensive care units of the Radboud and a written informed consent was obtained from each
University Nijmegen Medical Center (median time in- study subject.
terval from delivery to study 13 months, range 3-42), 85 of Analysis of NAT2 acetylator status was performed
whom showing the hemolysis elevated liver enzymes, low by quantifying caffeine metabolites in urine without
platelets (HELLP) syndrome as well. Of these 134 giving an extra caffeine load, by using a modification of
women, 48 were also used in a study concerning the role the method as described by Tang et al.16 The NAT2
of epoxide hydrolase in preeclampsia.9 Preeclampsia was acetylator status was established by estimating the ratio
defined as the occurrence after 20 weeks’ gestation of of the caffeine metabolites 5-acetylamino-6-amino-3-
a diastolic blood pressure greater than 90 mm Hg (phase 5 methyluracil (AAMU) and 1-methylxanthine (1X) in
Korotkoff sound) on 2 or more occasions at least 4 hours urine (Figure 1). As can be seen in the reaction scheme
apart, and concordant proteinuria (urinary protein of Figure 1, an unknown metabolite of caffeine is
greater than 0.3 g/L in a 24-hour urine collection period). converted by NAT2 into 5-acetylamino-6-formyl-3-
The HELLP syndrome was defined as the simultaneous methyluracil (AMFU), which is converted in vitro into
occurrence of a platelet count less than 100 ! 109/L, AAMU, which is quantified. Another pathway, not
serum aspartate aminotransferase and serum alanine catalyzed by NAT2, leads to 1X. Therefore, the NAT2
aminotransferase concentrations greater than 70 IU/L, acetylator status can be estimated from the ratio
and lactic dehydrogenase more than 600 IU/L. A group of AAMU (NAT catalyzed product)/1X (non-NAT cata-
109 nonpregnant women served as controls. They all lyzed product); ie, a high AAMU/1X ratio points at
experienced normotensive pregnancies only (data ob- a fast NAT acetylator phenotype.
tained by questionnaire). Seventy-two of these women To determine the NAT2 acetylator status, all women
were also used as controls in a study comparing the NAT2 collected a first-void urine sample, which was stored at
acetylator status in women with a child with orofacial cleft ÿ30(C within 1 hour until analysis. The concentrations
or spina bifida in comparison with women having healthy of AAMU and 1X in urine were determined by high-
children.15 The median time interval from the last delivery performance liquid chromatography (HPLC) (model
to the moment of study was 14 months (range 9-36). All 590 solvent delivery system coupled with a WISP
women were Dutch white and not pregnant or breastfeed- autosampler/injector and a 440 fixed-wavelength ultra-
ing at the time of study. violet detector) by using a Bio-Gel TSK-20 gel filtration
All women filled in a questionnaire at home that was column (Bio-Rad, Mississauga, Canada). Calibration
checked for missing or unclear answers by an additional curves were made by adding solutions containing known
personal interview, and in this way data on number of concentrations of AAMU and 1X to blank urine.
pregnancies, smoking habits, use of oral contraceptives Frozen urine samples were warmed up to room temper-
and education levels were obtained. The education level ature and vibrated ultrasonically to dissolve coprecipi-
was categorized as low (lower vocational, intermediate tated 1X. Then, 50 mL of 0.15 mol/L NaOH was added
Table I General characteristics of women with a history of Table II The distribution of the NAT acetylator status in
preeclampsia and control women during their last pregnancy controls and women with preeclampsia
Preeclampsia Controls Acetylator Preeclampsia Controls OR
n = 134 n = 109 status n = 134 n = 109 (95% CI)
Maternal age (y) 29 (20-38) 31 (18-40) Slow 72 (53.7) 81 (74.3) 1.0 (reference)
Primigravid 79%* 34% Fast 62 (46.3) 28 (25.7) 2.5 (1.4-4.3)*
Gestational age at 32 (26-40)* 39 (36-40) Values are given in numbers (percentages).
delivery (wks) * P ! .005, preeclampsia vs controls.
Diastolic blood 110 (92-140) ND
pressure (mm Hg)
Hemoglobin (g/dL) 12.8 (10.7-14.9) ND
Hematocrit (L/L) 0.36 (0.19-0.45) ND
was performed. If necessary, the OR was corrected for
Platelet count (!109/L) 94 (38-295) ND confounders (level of education, smoking, and use of
Serum AST (IU/L) 84 (11-2610) ND oral contraceptives at the time of urine sampling).
Serum ALT (IU/L) 75 (8-1920) ND Statistical significance was taken as P ! .05.
Serum LDH (IU/L) 650 (65-7670) ND
Serum creatine (mmol/L) 76 (49-215) ND
Serum uric acid (mmol/L) 0.40 (0.22-0.80) ND Results
Urinary protein/ 2.8 (0.3-27.6) ND
creatinine ratio General characteristics of women with a history of
(g/10 mmol) preeclampsia and controls are shown in Table I.
Values are given as median (range) or number (percentages), as Smoking habits (19% in cases vs 18% in controls), use
appropriate. of contraceptives (37% in cases vs 42% in controls), and
ND, Not determined; AST, aspirate aminotransferase; ALT, alanine education level (47% low education level and 53% high
aminotransferase; LDH, lactic dehydrogenase. education level in cases vs 42% and 58% respectively in
* P ! .01 preeclampsia vs controls (Mann-Whitney U test or c2
test).
controls) did not differ between cases and controls.
Women in the preeclamptic group delivered at a signif-
icantly lower gestational age than the controls (33 [range
26-38] vs 39 [range 36-40]; P ! .01). There were
to 50 mL urine to convert AFMU into AAMU. After 10 significantly more primiparae among the women with
minutes, 50 mL of 0.15 mol/L HCl and 50 mL of a 0.05% a history of preeclampsia than among the controls (79%
acetic acid solution were added. The samples were vs 33%, P ! .01).
vigorously shaken after each step. An aliquot of 20 mL The distribution of the AAMU/1X ratio in both cases
of the mixture was injected, and elution was performed and controls appeared to be bimodal.16,17 The cutoff
with 0.05% acetic acid at a flow-rate of 1.0 mL/min, and point separating slow from fast acetylators was graph-
monitored by measuring absorption at 263 nm. The ically determined at a ratio of 2.2 (Figure 2). We found
column was washed before each run. Quantification was significantly more fast acetylators in the preeclampsia
performed by measuring the area under the curve at group than in the control group (46.3% vs 25.7%,
retention times belonging to AAMU and 1X. The respectively; P ! .005) (Table II). Women with the fast
concentrations of these 2 metabolites in urine were acetylator status had an increased risk of 2.5 (95% CI
determined by comparison with the calibration curves. 1.4-4.3) for preeclampsia. No differences were found in
The NAT2 acetylator status was calculated from the acetylator status between women with a history of
ratio AAMU/1X; fast acetylators have a higher preeclampsia who either did or did not concurrently
AAMU/1X ratio than slow acetylators. The cutoff point have HELLP syndrome develop (46.9% vs 45.9% fast
separating slow from fast acetylators was determined acetylators, respectively; P = .96).
graphically. No association was found between the acetylator
The Mann-Whitney U test or c2 test, as appropriate, status and smoking habits, use of oral contraceptives, or
was used to assess statistical significance of differences in level of education. Although more primipara women
the population characteristics between cases and con- were found in the preeclampsia group, the OR did not
trols. The difference between groups with respect to slow change when adjusting for number of pregnancies.
or fast NAT2 acetylator status was tested with c2
analysis with Yates’ correction in 2 ! 2 contingency
tables. Women with an increased risk for preeclampsia Comment
were assumed to have the fast acetylator status. This risk
was estimated by calculating the odds ratio (OR) with Metabolism of endogenous and exogenous toxins nor-
95% CI for fast acetylators compared with slow mally requires phase 1 modification, followed by phase 2
acetylators. Furthermore, logistic regression analysis conjugation reactions. Increasing evidence suggests an
Zusterzeel et al 801
important role for biotransformation enzymes in pre- amounts of caffeine intake via coffee, tea, or soft drinks
eclampsia.2,18-20 The risk for preeclampsia may be in almost all individuals investigated was sufficient to
modified by variation in individual detoxification activ- make a reliable and reproducible estimate of each
ities, which may disturb the balance between phase 1 individual’s NAT phenotype by using the method as
and 2 enzymes. NAT enzymes are phase 2 enzymes that also applied by us. Indeed, in all 243 individuals
are involved in both bioinactivation and bioactivation, investigated in this study, without prior caffeine load,
depending on the substrate.5 However, because NATs the urinary levels of caffeine metabolites were high
are involved in both bioactivation as well as inactivation enough to be measured accurately. Levels of caffeine
reactions, the use of coenzyme A as a cofactor, the intake do influence the urinary levels of the various
a priori hypothesis that fast acetylators are less prone to caffeine metabolites; however, because changes in the
develop preeclampsia may be too simple. Several deri- levels of these metabolites, as a result of changes in
vates, notably in concert with oxidative metabolism by caffeine load, will follow the same trend, there seems to
cytochrome P450 catalyzed reactions are substrates for be hardly any effect on the AAMU/1X ratio.14,16,21
additional metabolism by NATs, resulting in more In conclusion, we demonstrated that the fast NAT
reactive compounds, which subsequently may harm status is associated with preeclampsia, which suggests
important cellular functions. It is not yet known which that interindividual variations in the metabolism of
reactive intermediates formed by NAT are biologically exogenous or endogenous toxic compounds may in-
plausible in terms of increasing susceptibility to pre- fluence the risk for preeclampsia. Clinical use of our
eclampsia, but reactive oxygen species such as lipid findings here may be 2-fold. First, it may be wise to
peroxides might be among them. After now having avoid an excessive intake of some NAT2 substrates,
established an association between preeclampsia and such as coffee, whereby stimulation of the NAT2
a fast acetylator phenotype, further research should be enzyme activity is limited during pregnancy. Second,
performed to identify possible relevant NAT substrates early screening for the NAT2 acetylator status may be of
or factors involved, which may be relevant for future additional value in the early recognition of individuals
prevention or treatment of preeclampsia. at high risk for preeclampsia.
In this case-control study we found that women with
the fast NAT2 acetylor status were more prone to have
preeclampsia develop. Higher acetylation may lead to Acknowledgments
more bioactivation of certain, yet unknown, compounds
leading to an imbalance between toxic compounds and We thank M. van Drongelen for laboratory assistance
detoxifying and scavenging substances. This finding is in and Dr I. Van Rooij for the collection of samples.
agreement with previous results of our group on
impaired detoxification in preeclampsia. We recently
showed that polymorphisms in genes encoding for the References
biotransformation enzymes glutathione S-transferase P1
and epoxide hydrolase, both of which may lead to 1. Higgs GA, Vane JR. Inhibation of cyclo-oxygenase and lip-
altered enzyme activity of the corresponding enzyme are oxygenase. Br Med Bull 1983;39:265-70.
associated with preeclampsia.8,9 Furthermore, we found 2. Uotila J, Tuimala R, Aarnio T, Pyykko K. Erythrocyte glutathi-
one peroxidase activity in hypertensive complications of preg-
reduced amounts of glutathione S-transferase P1 in both nancy. Gynecol Obstet Invest 1990;29:259-62.
placental and decidual tissue of preeclamptic women 3. Hubel CA, Kozlov AV, Kagan BE, Evans RW, Davidge ST,
compared with levels in healthy pregnant controls.20 McLaughlin MK, et al. Decreased transferrin and increased
We determined the NAT2 acetylator status in non- transferrin saturation in sera of women with preeclampsia:
pregnant women with a history of preeclampsia and implications for oxidative stress. Am J Obstet Gynecol
1996;175:692-700.
in women with uncomplicated pregnancies only. As 4. Roes EM, Raijmakers MTM, Zusterzeel PLM, Knapen MFCM,
Cascorbi et al17 showed a correlation of at least 90% Peters WHM, Steegers EAP. Deficient detoxifying capacity in
between the NAT2 acetylator status and the NAT2 the pathophysiology of preeclampsia. Med Hypotheses 2000;
genotype, it seems unlikely that pregnancy itself will 55:415-8.
affect the NAT2 acetylator status. 5. Wormhoudt LW, Commandeur JN, Vermeulen NPE. Genetic
polymorphisms of human N-acetyltransferase, cytochrome P450,
No association was found between NAT acetylator glutathione S-transferase, and epoxide hydrolase enzymes: rele-
status and smoking, use of oral contraceptives, age, or vance to xenobiotic metabolism and toxicity. Crit Rev Toxicol
education, which was in agreement with other stud- 1999;29:59-124.
ies.14,15,21,22 We did not investigate dietary intake of 6. Knapen MFCM, Zusterzeel PLM, Peters WHM, Steegers EAP.
caffeine, and we determined the NAT2 acetylator status Glutathione and glutathione-related enzymes in reproduction:
a review. Eur J Obstet Gynecol Reprod Biol 1999;82:171-84.
by measuring urinary caffeine metabolites without prior 7. Seidegard J, Ekstrom G. The role of human glutathione S-
extra caffeine load, because it has been demonstrated transferase and epoxide hydrolases in the metabolism of xeno-
earlier by Tang et al16 and Le Marchant et al14 that the biotics. Environ Health Perspect 1997;105(Suppl 4):791-9.
8. Zusterzeel PLM, Visser W, Peters WHM, Merkus JMWM, and spina bifida: N-acetyltransferase phenotype, maternal smok-
Nelen WLDM, Steegers EAP. Polymorphism in the glutathione ing, and medication use. Teratology 2002;66:260-6.
S-transferase P1 gene and risk for preeclampsia. Obstet Gynecol 16. Tang BK, Kadar D, Qian L, Iriah J, Yip J, Kalow W. Caffeine as
2000;95:50-4. a metabolic probe: validation of its use for acetylator phenotyping.
9. Zusterzeel PLM, Peters WHM, Visser W, Hermsen KJM, Merkus Clin Pharmacol Ther 1991;49:648-57.
JMWM, Steegers EAP. A polymorphism in microsomal epoxide 17. Cascorbi I, Drakoulis N, Brockmöller J, Mauer A, Sperling
hydrolase gene is associated with preeclampsia. J Med Genet K, Rots I. Arylamine N-acetyltransferase (NAT2) mutations
2001;38:234-7. and their allelic linkage in unrelated Caucasian individuals:
10. King CM, Land SJ, Jones RF, Debiec-Rychter M, Lee MS, Wang correlation with phenotypic activity. Am J Hum Genet 1995;
CY. Role of acetyltransferases in the metabolism and carcinoge- 57:581-92.
nicity of aromatic amines. Mutat Res 1997;376:123-8. 18. Loverro G, Greco P, Capuano F, Carone G, Selvaggi L. Lip-
11. Okkels H, Sigsgaard T, Wolf H, Autrup H. Arylamine N- operoxidation and antioxidant enzymes activity in pregnancy
acetyltransferase 1 (NAT1) and 2 (NAT2) polymorphisms in complicated with hypertension. Eur J Obstet Genecol Reprod
susceptibility to bladder cancer: the influence of smoking. Cancer Biol 1996;70:123-7.
Epidemiol Biomarkers Prev 1997;6:225-31. 19. Wang Y, Walsh SW. Antioxidant activities and mRNA expression
12. Baranova H, Canis M, Ivaschenko T, Albuisson E, Bothorishvilli of superoxide dismutase, catalase, and glutathione peroxidase in
R, Baranov V, et al. Possible involvement of arylamine normal and preeclamptic placentas. J Soc Gynecol Investig
N-acetyltransferase 2, glutathione S-transferase M1 and T1 genes 1996;3:179-84.
in the development of endometriosis. Mol Human Reprod 20. Zusterzeel PLM, Peters WHM, de Bruyn MAH, Merkus JMWM,
1999;5:636-41. Steegers EAP. Glutathione S-transferase isoenzymes in decidua
13. Hirvonen A. Polymorphic NATs and cancer predisposition. IARC and placenta of preeclamptic pregnancies. Obstet Gynecol
Sci Publ 1999;148:251-70. 1999;94:1033-8.
14. Le Marchant L, Sivamaran L, Franke AA, Custer LJ, Wilkens LR, 21. Vistisen K, Poulsen HE, Loft S. Foreign compounds metabolism
Lau AF, et al. Predictors of N-acetyltransferase activity: should capacity in man measured from metabolites of dietary caffeine.
caffeine phenotyping and NAT2 genotyping be used interchange- Carcinogenesis 1992;13:1561-8.
ably in epidemiological studies? Cancer Epidemiol Biomarkers 22. Rothman N, Hayes RB, Bi W, Caporosa N, Broly F, Woosley
Prev 1996;5:449-55. RL, et al. Correlation between N-acetyltransferase activity and
15. Van Rooy IALM, Groenen PMW, van Drongelen M, te Morsche NAT2 genotype in Chinese males. Pharmacogenetics 1993;3:
RHM, Peters WHM, Steegers-Theunissen RPM. Orofacial clefts 250-5.
www.ajog.org
Profound hypotension and associated electrocardiographic

changes during prolonged cord occlusion in the near term
fetal sheep
Bert Wibbens, MD,a,b Jenny A. Westgate, MD, PhD,a Laura Bennet, PhD,a
Vincent Roelfsema,a Harmen H. De Haan, MD, PhD,c Christian J. Hunter, MD, PhD,d
Alistair J. Gunn, MBChB, PhDa
From the Departments of Physiology and Obstetrics and Gynecology, University of Auckland, Auckland, New
Zealand a; Department of Obstetrics and Gynaecology, Academic Medical Hospital, Groningen, The Netherlandsb;
Department of Obstetrics and Gynaecology, Isala Clinics, Zwolle, The Netherlandsc; and Center for Perinatal Biology,
Loma Linda University School of Medicine, Loma Linda, CA d
Received for publication November 15, 2004; revised December 23, 2004; accepted January 25, 2005
KEY WORDS Objective: To determine whether the onset of fetal hypotension during profound asphyxia is
Asphyxia reflected by alterations in the ratio between the T height, measured from the level of the PQ
Fetal electrocardio- interval, and the QRS amplitude (T/QRS ratio) and ST waveform.
graphic monitoring Study design: Chronically instrumented near-term fetal sheep received complete occlusion of the
T/QRS ratio umbilical cord for either 8 (n = 6) or 15-min (n = 9).
ST segment Results: Cord occlusion led to sustained bradycardia and severe acidosis. Mean arterial blood
Sheep pressure initially rose and then fell to a nadir of 32.6 G 2.6 mm Hg in the 8-min group and 9.3 G
1.0 mm Hg in the 15-min group (P ! .001). The T/QRS ratio rose initially in parallel with mean
arterial blood pressure and then reduced as mean arterial blood pressure fell but remained
significantly above baseline. Biphasic ST waveforms during occlusion occurred in only 2 fetuses,
but biphasic and negative waveforms occurred during reperfusion in the 15-min group, with
a significant rise in troponin T levels (0.58 G 0.46 versus 0.02 G 0.01 ng/mL at 6 h, P ! .01).
Conclusion: Elevation of the T/QRS ratio does not identify fetal hypotension during severe
hypoxia, but abnormal waveforms in the recovery phase may indicate developing cardiac
injury.
In experimental studies the presence and severity of

hypotension is one of the major factors associated with
Supported by the grants from the Health Research Council of New neural injury.1-3 Ideally, fetal surveillance would identify
Zealand, Lottery Grant Board of New Zealand, Auckland Medical
Research Foundation, and VSB Fonds and Stichting Nuffic (The
not simply the presence of hypoxia but also those fetuses
Netherlands). whose adaptation to asphyxia is beginning to fail, as
Reprints not available from the authors. shown by the development of hypotension. The presence
doi:10.1016/j.ajog.2005.01.058
804 Wibbens et al
of elevation of the ST segment of the fetal electrocardio- Using sterile techniques, catheters were placed in the
gram (ECG), typically measured relative to the QRS right femoral artery and vein, left and right brachial
complex (the ratio between the T height and the QRS artery and right vein, and the amniotic sac. Ultrasound
amplitude [T/QRS ratio]), is a marker for fetal hypoxia blood flow probes (size 3S, Transonic Systems Inc,
in clinical and experimental studies.4-8 However, the Ithaca, NY) were placed around the left femoral artery
relationship between ST waveform changes and the for measurement of femoral blood flow (FBF). ECG
development of fetal compromise is complex.9,10 Indeed, electrodes were placed subcutaneously over the right
the appearance of biphasic and negative ST waveforms shoulder and chest at apex level and sewn across the
between variable decelerations induced by cord occlu- chest to record the fetal ECG. An inflatable silicone
sion in fetal sheep seems to be more predictive of fetal occluder was placed around the umbilical cord of all
compromise than the changes in T wave height or shape fetuses (In Vivo Metric, Healdsburg, Calif). All leads
during decelerations.9 It is unclear whether it was the were exteriorized through the maternal flank, and
extreme brevity of hypotension during each brief vari- a maternal long saphenous vein was catheterized to
able deceleration, which limited the apparent utility of provide access for postoperative care and killing of the
the ST waveform hypoxic changes in that model, or sheep. Eighty milligrams gentamicin (Rousell, Auck-
whether they were related not to hypotension per se but land, New Zealand) was administered into the amniotic
to some other factor such as myocardial injury.11,12 sac prior to the closure of the uterus. Amniotic fluid lost
To dissect these factors, we used a model of during surgery was replaced using normal saline warmed
sustained hypoxia induced by umbilical occlusion in to 37(C.
near-term fetal sheep in which responses during initial A period of 4 to 6 days postoperative recovery was
compensation (hypertension) and later decompensation allowed before experiments commenced, during which
(hypotension) could be contrasted. This study examined time antibiotics were administered daily for 5 days IV to
the relationship between ST waveform and T/QRS the ewe (600 mg benzylpenicillin sodium [Crystapen]
ratio changes and the development of hypotension. and 80 mg gentamicin). Fetal catheters were maintained
We compared the effects of 8 min of occlusion, which patent by continuous infusion of heparinized saline (40
was predicted to cause only mild hypotension,13 with 15 U.mL1at 0.2 mL.hour1), and the maternal catheter
min of occlusion, which is associated with profound was maintained by daily flushing.
hypotension,14 on postasphyxial recovery of mean
arterial blood pressure (MAP), ST waveform shape Recordings
changes, and fetal levels of a marker of cardiac injury,
troponin T.15 Fetal arterial and venous blood pressure, corrected for
amniotic fluid pressure (Novatrans II, MX860, Medex
Inc, Ohio), heart rate, ECG, and FBF (T208 Ultrasonic
Material and methods Flowmeter, Transonic Systems Inc) were recorded
continuously from 24 h before the experiment until
Surgery 72 h afterward. The MAP signal was collected at 64 Hz
Romney/Suffolk fetal sheep were instrumented between and low pass filtered at 30 Hz. The ECG signals were
119 and 126 days of gestation (term = 147 days) under stored to disk using custom software (Labview for
general anesthesia (2% halothane in O2) using sterile Windows, National Instruments Ltd, Austin, Tex).
techniques.9,13,16 All procedures were approved by the They were analog filtered between 0.05 and 80 Hz and
Animal Ethics Committee of the University of Auck- digitized at 512 Hz.
land. Food but not water was withdrawn 18 h before Experimental procedures
surgery. Ewes were given 5 mL of Streptopen (Procaine
penicillin [250,000 IU] and Dihydrostreptomycin Fetuses were randomly assigned to either the 8-min
[250mg/mL], Pitman-Moore, Wellington, New Zealand) occlusion group (n = 6) or the 15-min occlusion group
intramuscularly for prophylaxis 30 min prior to the start (n = 9). Fetal asphyxia was induced by a rapid inflation
of surgery. Anesthesia was induced by intravenous (IV) of the umbilical occluder with sterile saline of a defined
injection of Alphaxalone and Alphadolone (3 mg/kg, volume (6-8 mL) known to completely inflate the
Schering-Plough Animal Health Ltd, Wellington, New occluder. The success of occlusion was confirmed by
Zealand) and general anesthesia maintained using 2% to observation of an immediate sharp rise in MAP and
3% halothane in O2. The depth of anesthesia and a fall in fetal heart rate (FHR) to below 100 beats/min.
maternal respiration were constantly monitored by On completion of the occlusion period, the occluder was
trained anesthetic staff. Under anesthesia a 20-gauge deflated over 10 to 15 seconds to prevent excessively
IV catheter was placed in a maternal front leg vein, and rapid changes in the circulating blood volume. When
the ewes were placed on a constant infusion isotonic bradycardia persisted for more than 30 seconds or fetal
saline drip to maintain maternal fluid balance. blood pressure did not increase to over 25 mm Hg in
Wibbens et al 805
Table I Gestational age at surgery, weight at postmortem, and survival time after umbilical cord occlusion in the 2 groups
Group No. Weight (g) SEM Gestation (d) SEM Survival (hr) SEM
8 min 6 2837.8 134.4 122.1 1.2 72.0 0
15 min 9 3349.3 184.2 124.4 0.4 16.0 7.3
Table II Fetal acid-base balance and serum glucose and lactate levels for the 8- and 15-minute occlusion groups*
Group Baseline 2 min 7 or 12 min 30 min post 1 h post 2 h post 4 h post 6 h post
pH
15 min 7.36 G 0.01 7.22 G 0.017 6.93 G 0.01y 7.2 G 0.02y 7.25 G 0.07 7.20 G 0.06y 7.06 G 0.10y 7.10 G 0.10y
8 min 7.38 G 0.01 7.23 G 0.04 7 G 0.01 7.31 G 0.01 7.36 G 0.01 7.37 G 0.01 7.37 G 0.01 7.38 G 0.01
PCO2
15 min 45.2 G 2.1 69.9 G 3.7 116.7 G 0.02 48.5 G 2.28 49.0 G 3.5y 50.6 G 3.0y 54.8 G 1.7y 50.2 G 2.2y
8 min 45.7 G 1.1 67.5 G 5.9 104.2 G 5.0 46.1 G 1.55 45.0 G 0.7 45.3 G 0.9 45.1 G 0.6 45.4 G 0.8
pO2
15 min 21.5 G 0.9 5.9 G 0.9 8.8 G 0.6 26.3 G 1.25 22.6 G 2.7 20.7 G 1.5 23.2 G 2.4 20.8 G 1.96
8 min 21.8 G 1.1 5.6 G 0.8 5.5 G 0.6 22.8 G 1.03 21.5 G 1.6 21.8 G 1.2 22.1 G 1.9 22.8 G 1.5
Lactate
15 min 0.75 G 0.10 1.96 G 0.2 5.6 G 0.2y 5.2 G 0.3y 5.9 G 0.9y 6.5 G 1.3y 10.2 G 2.8y 9.6 G 2.7y
8 min 0.91 G 0.10 2.1 G 0.6 4.6 G 0.3 3.5 G 0.2y 2.81 G 0.3 2.2 G 0.2 1.5 G 0.2 1.1 G 0.2
Glucose
15 min 0.6 G 0.1 0.3 G 0.1 1.1 G 0.1 1.6 G 0.2y 1.3 G 0.2 1.1 G 0.2 1.0 G 0.2 1.0 G 0.1
8 min 0.7 G 0.1 0.4 G 0.1 0.8 G 0.1 1.2 G 0.1 1.0 G 0.1 0.9 G 0.1 1.0 G 0.1 0.9 G 0.1
* The final blood sample during occlusion was taken at 7 minutes in the 8-minute group and 12 minutes in the 15-minute group.
y
P ! .05, analysis of variance versus 8-minute group. Data are mean G SEM.
the first 60 seconds after release of occlusion, then a intervals. For each averaged waveform, the ratio be-
dose of 0.3 mL (0.1 mL/kg estimated weight) of 1:10 000 tween the T height, measured from the level of the PQ
epinephrine was given by slow IV push. interval, and the QRS amplitude was calculated (T/QRS
Fetal arterial blood samples (0.3 mL) were taken at ratio) to give a measure of T-wave amplitude. The raw
baseline (60 minutes prior to occlusion), after 2 and ECG data for each averaged waveform were visually
either 7 minutes (8-minute occlusion group) or 12 assessed by an experienced researcher (J. W.) to identify
minutes (15-minute occlusion group) during occlusion ST waveform shape changes and verify software iden-
and then 30 minutes and 1, 2, 3, 4, 6, 24, 48, and 72 tification of the T-wave.
hours after release of the occluder for pH, blood gas, The effects of occlusion on time sequence data were
hematocrit, and hemoglobin determination (845 blood evaluated by repeated-measures 1-way analysis of var-
gas analyzer and cooximeter, Ciba-Corning Diagnostics, iance (SPSS, version 10, SPSS Inc, Chicago, Ill). The
Mass) and for glucose and lactate measurements (model baseline period was taken as the mean of the hour before
2300, YSI, Yellow Springs, Ohio). Samples (0.5 mL) occlusion. When an effect of time was found, data were
were frozen at 80(C for later measurement of cardiac compared with baseline using Dunnett’s post-test. In-
troponin T values using the Elecsys 2010 immunoassay tergroup comparisons for selected data were performed
system (Roche, Boehringer-Mannheim, Germany). On by analysis of variance. Statistical significance was
completion of the experiment, the ewes and fetuses were accepted when P ! .05. Data are mean G SEM.
killed by an overdose of sodium pentobarbitone (9 g IV
to the ewe: Pentobarb 300, Chemstock International,
Christchurch, New Zealand). Results
Data analysis and statistics Table I shows the number of fetuses, mean survival time,
and the range of survival in each group. There were no
Off-line analysis of data was performed using custom- differences in mean gestational age at surgery in each
ized Labview for Windows (National Instruments Ltd). group or fetal weight at postmortem. All fetuses had
One-second averages of MAP were calculated for each normal blood gases and glucose and lactate levels by the
fetus; these data were used to derive the baseline MAP standards of our laboratory (Table II). There was
before and after occlusions, and the minimum MAP a significant postinsult mortality in the 15-minute group
at the end of each occlusion. The ECG waveform was but not the 8-minute group. Seven lambs required
averaged with respect to the S wave over 5-second epinephrine i.v. for fetal resuscitation during reperfusion.
806 Wibbens et al
Figure 1 The time sequence of changes during and for 30 Figure 2 The time sequence of changes during and for 30
minutes after asphyxia induced by complete umbilical cord minutes after asphyxia induced by complete umbilical cord
occlusion for 8 minutes (upper panel) and 15 minutes (lower occlusion for 8 minutes (upper panel) and 15 minutes (lower
panel) on MAP (closed circles), and T/QRS ratio (open circles). panel) on FHR (closed circles) and FBF (open circles). The
The gray bar denotes the period of occlusion. Values are gray bar denotes the period of occlusion. Values are plotted at
plotted at 1-minute intervals. Note the transient period of 1-minute intervals. Asterisk, P ! .05 versus baseline (FHR,
hypertension in the 8-minute group but not the 15-minute solid line; FBF, dashed line). Mean G SEM.
group during recovery. Asterisk denotes P ! .05 versus
baseline (MAP, solid line; T/QRS, dashed line). Mean G SEM.
umbilical cord occlusion. In both groups MAP initially
rose after occlusion and then fell, reaching baseline values
Three did not respond; 4 showed a brisk response but after 6.5 G 0.2 minutes and 6.8 G 0.2 minutes (N.S.) in the
then progressively deteriorated and died within 15 hours 8- and 15-minute groups, respectively. The nadir of MAP
after the end of occlusion. was 32.6 G 2.6 mm Hg (versus 43.6 G 1.6 baseline) in the
Changes in fetal blood gases, glucose, 8-minute group versus 9.3 G 1.0 mm Hg in the 15-minute
group (versus 44.2 G 1.8 baseline). The initial hyperten-
and lactate
sion was associated with profound peripheral vasocon-
Umbilical cord occlusion was associated with metabolic striction with very low FBF (Figure 2). However, this
acidemia, hypoxia, and hypercapnia, which were more effect was only temporary: FBF increased again after 5 to
severe in the 15-minute group (Table II). Following 6 minutes, to approximately 25% of baseline values.
release of occlusion, the 8-minute group showed nor-
malization of arterial pH values, PO2, and pCO2 within Hemodynamic changes after occlusion
30 minutes after release of the occluder. In contrast, the
After release of occlusion, the 8-minute group showed
15-minute group showed a continuing metabolic acido-
rapid spontaneous recovery of both MAP and FHR
sis, with a mild but persistent hypercarbia (Table II).
(Figures 1 and 2), whereas the 15-minute group showed
Hemodynamic changes during umbilical slow recovery of MAP and FHR. Epinephrine admin-
cord occlusion istration was required in 7 of the 9 fetuses in the 15-
minute group, starting at a mean of 2.6 minutes after
Figures 1 and 2 show the changes in fetal MAP, T/QRS release of occlusion. MAP returned to baseline values in
ratio, FHR, and FBF during and immediately after 0.7 G 0.2 minutes in the 8-minute group and 5.3 G 0.9
Wibbens et al 807
Figure 3 A, With the onset of occlusion, most animals showed negative T waves, followed by a rapid inversion toward a positive
T wave and progressive ST elevation. B, After failure of compensation, typically the ST segment flattened, with progressive reduction
in height and lengthening of the Q-T interval. C, Example of the tombstone-like shape of the ST segment. D, Example of the fork-
like shape of the QRS segment. E, Postasphyxial recovery of MAP was associated with a rapid, deep inversion of the T wave,
accompanied by initially a negative and then a positive ST-waveform. F, Once MAP had recovered, 4 animals showed ST
depression, with a negative T wave, followed by a progressive return to baseline appearance.
808 Wibbens et al
minutes in the 15-minute group. In the 8-minute but not After occlusion, 1 fetus in the 8-minute group de-
the 15-minute group, this was followed by transient veloped biphasic ST waveforms, for 225.0 seconds,
hypertension, lasting approximately 10 minutes. The 8- whereas the remainder rapidly returned to baseline ST
minute group showed immediate overshoot tachycardia waveform appearances. In the 6 surviving 15-minute
after release of occlusion, lasting 2 minutes, followed by group fetuses postasphyxial recovery of MAP was
transient normalization and then a further transient rise associated with development of deeply negative T-waves
(Figure 2). In contrast, the 15-minute group showed and biphasic ST waveforms in 4 fetuses (Figures 3, E
a slower rise, which followed epinephrine in all but 2 and F); all but one had received epinephrine. When seen,
cases, and then a longer period of tachycardia. biphasic waveforms lasted for 206 G 71 seconds. ST
FBF flow during recovery in the 8-minute group depression was seen in 2 fetuses (for 6 and 9 minutes,
showed a pattern of temporary normalization lasting 4 respectively). By 30 minutes after release of the occluder,
minutes, followed by a rapid fall to approximately 50% all fetuses showed positive ST waveforms, and only 1
of the baseline values (Figure 2). In contrast, FBF in the had a deeply negative T wave.
15-minute group showed a much smaller initial increase,
which did not reach baseline levels, followed by severe
Troponin T
sustained hypoperfusion. Postasphyxial levels of troponin T were significantly
greater in the 15-minute group than in the 8-minute
T/QRS ratio and ST waveform changes group. Troponin levels in the 15-minute group rose from
0.02 G 0.01 ng/mL at baseline to 0.38 G 0.12 at 3 hours
The initial T-wave orientation was variable prior to and 0.58 G 0.21 at 6 hours, P ! .01 versus the 8-minute
occlusion. However, all fetuses, including those with group). In contrast, there was no significant change in
initially negative T waves, showed rapid T-wave and ST the 8-minute group, with levels of 0.02 G 0.01 ng/mL at
(positive) elevation at the onset of occlusion (Figures 1 baseline, 0.02 G 0.01 at 3 hours, and 0.025 G 0.006 at
and 3, A). Peak values of MAP were reached after 2.5 G 6 hours.
0.5 minutes versus 1.9 G 0.1 minutes of occlusion (8-
and 15-minute groups, respectively, N.S.), whereas peak Comment
values of T/QRS were reached at 2.9 G 0.6 minutes
versus 3.7 G 0.4 minutes (N.S.). With continued This is the first study to report fetal ECG changes during
occlusion, the onset of fetal hypotension was accompa- and after a catastrophic insult. Consistent with 1 previous
nied by a fall of T/QRS height back toward normal acute study,14 15 minutes of cord occlusion in the late-
values (Figure 3, B). In a few cases, negative T waves gestation fetal sheep was associated with profound
were seen briefly. However, the average T/QRS ratio hypotension, requirement for resuscitation with i.v.
remained significantly greater than baseline levels in both epinephrine, and a subsequent high mortality. Such un-
groups until the end of occlusion (P ! .05, Figure 1). predictable acute, catastrophic events, such as abruptio
The ST waveform showed a typical sequence of placentae or umbilical cord prolapse, continue to be a
changes, which is illustrated in Figure 3. With the onset significant cause of perinatal morbidity.17
of occlusion, all 15 fetuses in the 2 groups rapidly Dynamic ECG changes, particularly biphasic or
developed ST elevation. One 15-minute fetus initially negative ST waveforms, are suggested to reflect anaer-
showed a tombstone-like shape of ST waveform eleva- obic myocardial metabolism and depletion of myocar-
tion with heart block (Figure 3, C), which resolved after dial glycogen reserve, augmented by b-adrenergic
several minutes. Biphasic ST waveforms were seen only stimulation.5,18 Therefore, we hypothesized that the
in 1 15-minute fetus, after 14 minutes of occlusion, and profound hypotension would be associated with ST
in 1 8-minute fetus intermittently (for a total of 225 waveform changes during sustained occlusion. Initially,
seconds). In the 15-minute group, the ST waveform there was a dramatic elevation in ST and T-wave height
elevation gradually resolved toward the end of occlu- at a time of good fetal compensation, as demonstrated
sion, as shown in Figure 3, B. by the marked hypertension, mediated by increased
In the 15-minute group, the degree of ST waveform peripheral vascular resistance. Once MAP began to
elevation gradually reduced as the occlusion continued, fall, however, the T/QRS ratio also fell, although it
as shown in Figure 3, B. Three fetuses showed bizarre remained mildly elevated on average despite the de-
ST waveform shapes toward the end of the occlusion velopment of profound metabolic acidosis and hypo-
period, 1 with a tombstone appearance and heart block tension, in part secondary to failure of femoral
and 2 with fork-like QRS shapes similar to bundle vasoconstriction.19
branch block patterns (Figure 3, D). These fetuses did Thus, the elevation of the T/QRS ratio was greatest
not respond to epinephrine after release of occlusion and at the time of maximal fetal compensation. We speculate
died soon after reperfusion. ST depression was never that this early maximal elevation corresponds with the
seen during occlusion. time of maximal anaerobic cardiac metabolism, and the
Wibbens et al 809
subsequent attenuation of ST waveform height reflects damage in the fetal sheep. Am J Obstet Gynecol 1992;167:
depletion of the key substrate, cardiac glycogen, leading 1423-30.
2. Fujii EY, Takahashi N, Kodama Y, Roman C, Ferriero DM,
to a progressive reduction in anaerobic cardiac metab- Parer JT. Hemodynamic changes during complete umbilical cord
olism.5 These results are consistent with the finding, in occlusion in fetal sheep related to hippocampal neuronal damage.
studies in fetal lambs of prolonged moderate asphyxia Am J Obstet Gynecol 2003;188:413-8.
induced by reduced uterine perfusion20 and of brief 3. de Haan HH, Gunn AJ, Williams CE, Gluckman PD. Brief
repeated cord occlusion,9 that T/QRS elevation did not repeated umbilical cord occlusions cause sustained cytotoxic
cerebral edema and focal infarcts in near-term fetal lambs. Pediatr
indicate the development of fetal compromise. Thus, Res 1997;41:96-104.
these data suggest that elevation of the T/QRS ratio in 4. Greene KR, Dawes GS, Lilja H, Rosen KG. Changes in the ST
isolation is a marker for anaerobic cardiac metabolism waveform of the fetal lamb electrocardiogram with hypoxemia.
due to fetal hypoxia rather than for the development of Am J Obstet Gynecol 1982;144:950-8.
hypotension.21 5. Hokegard KH, Eriksson BO, Kjellmer I, Magno R, Rosen KG.
Myocardial metabolism in relation to electrocardiographic
There are some data to suggest that the onset of changes and cardiac function during graded hypoxia in the fetal
significant fetal acidosis and hypotension may be lamb. Acta Physiol Scand 1981;113:1-7.
reflected by the appearance of negative and biphasic 6. Rosen KG, Dagbjartsson A, Henriksson BA, Lagercrantz H,
ST waveforms,8,9 possibly reflecting myocardial ische- Kjellmer I. The relationship between circulating catecholamines
mia.22 In the present study, this hypothesis was not and ST waveform in the fetal lamb electrocardiogram during
hypoxia. Am J Obstet Gynecol 1984;149:190-5.
confirmed because the majority of fetuses showed 7. Westgate JA, Harris M, Curnow JS, Greene KR. Randomised trial
a consistent elevation of the ST segment during occlu- of cardiotocography alone or with ST waveform analysis for
sion, with brief, transient biphasic ST waveforms in only intrapartum monitoring. Lancet 1992;340:194-8.
2 fetuses. In contrast, however, during recovery from 8. Amer-Wahlin I, Hellsten C, Noren H, Hagberg H, Herbst A,
asphyxia, biphasic waveforms or ST depression was seen Kjellmer I, et al. Cardiotocography only versus cardiotocography
plus ST analysis of fetal electrocardiogram for intrapartum fetal
transiently in 5 of the 6 survivors in the 15-minute group monitoring: a Swedish randomised controlled trial. Lancet
(and 1 animal in the 8-minute group). It is likely that 2001;358:534-8.
these ST waveform changes reflect at least in part the 9. Westgate JA, Bennet L, Brabyn C, Williams CE, Gunn AJ. ST
development of myocardial injury, as shown by waveform changes during repeated umbilical cord occlusions in
a marked rise in troponin T values in the 15-minute near-term fetal sheep. Am J Obstet Gynecol 2001;184:743-51.
10. de Haan HH, Gunn AJ, Gluckman PD. Fetal heart rate changes
group. The rise was markedly more than adult normal do not reflect cardiovascular deterioration during brief repeated
values of ! 0.03 mg/L and was associated with pro- umbilical cord occlusions in near-term fetal lambs. Am J Obstet
gressive cardiovascular decompensation. Similarly, evi- Gynecol 1997;176:8-17.
dence of reversible cardiac injury associated with 11. Barberi I, Calabro MP, Cordaro S, Gitto E, Sottile A, Prudente D,
hypotension has reported in near-term fetal sheep after et al. Myocardial ischaemia in neonates with perinatal asphyxia.
Electrocardiographic, echocardiographic and enzymatic correla-
repeated umbilical cord occlusion12 and in newborn tions. Eur J Pediatr 1999;158:742-7.
infants with hypoxic-ischemic encephalopathy.11 12. Gunn AJ, Maxwell L, de Haan HH, Bennet L, Williams CE,
In conclusion, in a model of sustained asphyxia, the Gluckman PD, et al. Delayed hypotension and subendocardial
T/QRS ratio was maximally elevated during the early injury after repeated umbilical cord occlusion in near-term fetal
phase of fetal compensation, with elevated arterial blood lambs. Am J Obstet Gynecol 2000;183:1564-72.
13. Hunter CJ, Bennet L, Power GG, Roelfsema V, Blood AB,
pressure and peripheral vasoconstriction. This elevation Quaedackers JS, et al. Key neuroprotective role for endogenous
reduced as compensation failed. This pattern may reflect adenosine A1 receptor activation during asphyxia in the fetal
the timing of maximal cardiac anaerobic metabolism. sheep. Stroke 2003;34:2240-5.
Abnormal ST waveforms occurred in only a few fetuses 14. Thorngren-Jerneck K, Ley D, Hellstrom-Westas L, Hernandez-
and were also not helpful in discriminating the de- Andrade E, Lingman G, Ohlsson T, et al. Reduced postnatal
cerebral glucose metabolism measured by PET after asphyxia in
velopment of fetal hypotension. However, transient near term fetal lambs. J Neurosci Res 2001;66:844-50.
biphasic waveforms and ST depression were seen in 15. Moller JC, Thielsen B, Schaible TF, Reiss I, Kohl M, Welp T,
many in the 15-minute group during resuscitation. This et al. Value of myocardial hypoxia markers (creatine kinase and its
group showed a subsequent marked elevation of tropo- MB-fraction, troponin-T, QT-intervals) and serum creatinine for
nin T levels and a very high postresuscitation mortality, the retrospective diagnosis of perinatal asphyxia. Biol Neonate
1998;73:367-74.
consistent with developing cardiac injury. These data 16. George S, Gunn AJ, Westgate JA, Brabyn C, Guan J, Bennet L.
support the concept that clinical ST waveform monitor- Fetal heart rate variability and brainstem injury after asphyxia in
ing should focus on the intervals immediately following preterm fetal sheep. Am J Physiol Regul Integr Comp Physiol
FHR decelerations. 2004;287:R925-33.
17. Westgate JA, Gunn AJ, Gunn TR. Antecedents of neonatal
encephalopathy with fetal acidaemia at term. Br J Obstet Gynaecol
References 1999;106:774-82.
18. Hokegard KH, Karlsson K, Kjellmer I, Rosen KG. ECG-changes
1. Mallard EC, Gunn AJ, Williams CE, Johnston BM, Gluckman in the fetal lamb during asphyxia in relation to beta-adrenoceptor
PD. Transient umbilical cord occlusion causes hippocampal stimulation and blockade. Acta Physiol Scand 1979;105:195-203.
810 Wibbens et al
19. Jensen A, Hohmann M, Kunzel W. Dynamic changes in organ 21. Westgate JA, Bennet L, Gunn AJ. Fetal ECG monitoring. Fetal
blood flow and oxygen consumption during acute asphyxia in fetal Neonatal Med Rev 2002;13:119-39.
sheep. J Dev Physiol 1987;9:543-59. 22. Widmark C, Jansson T, Lindecrantz K, Rosen KG. ECG
20. de Haan HH, Ijzermans AC, de Haan J, Hasaart TH. The T/QRS waveform, short term heart rate variability and plasma catechol-
ratio of the electrocardiogram does not reliably reflect well- being in amine concentrations in response to hypoxia in intrauterine
fetal lambs [see comments]. Am J Obstet Gynecol 1995;172:35-43. growth retarded guinea-pig fetuses. J Dev Physiol 1991;15:161-8.
www.ajog.org
Hyperemesis gravidarium: Epidemiologic findings

from a large cohort
Jennifer L. Bailit, MD, MPH
Division of Maternal Fetal Medicine, Department of Obstetrics and Gynecology, Cleveland, OH
KEY WORDS Objective: This study was undertaken to quantify the frequency, clinical course, charges, and
Hyperemesis outcomes of hyperemesis gravidarum.
gravidarium Study design: California birth certificate data linked with maternal and neonatal hospital
Complications of discharge data in 1999 were used (N = 520,739). Hyperemesis was defined by ICD-9 codes. The
pregnancy frequency, estimated charges, and demographic characteristics associated with hyperemesis
Epidemiology patients were assessed. Maternal and neonatal perinatal outcomes were compared by maternal
hyperemesis status.
Results: Hyperemesis complicated 2,466 of 520,739 births. The average length of stay was 2.6
days and the average charge was $5,932. Singleton hyperemesis infants were smaller (3,255 vs
3,380 g; P ! .0001 and more likely to be small for gestational age (29.21% vs 20.8%; P ! .0001).
Conclusion: Hyperemesis occurs in 473 of 100,000 live births and is associated with significant
charges. Infants of mothers with hyperemesis have lower birth weights and the mothers are more
likely to have infants that are small for gestational age.
Nausea and vomiting of pregnancy are a known and there has been 1 population-based study regarding
common symptom of pregnancy affecting 50% to 90% hyperemesis in Sweden using 1970s data,5 there have
of pregnant women.1 Hyperemesis, intractable nausea, been no population-based studies in the United States.
and vomiting of pregnancy, is rarer but more serious. Therefore, this study proposes to study the incidence of
Hyperemesis can lead to dehydration, metabolic distur- hospitalization for hyperemesis per 100,000 live births,
bances, and nutritional deficits requiring hospitaliza- the demographic characteristics of women hospitalized
tion, and, if left untreated, death.2 Current estimates of for hyperemesis, charges for treatment, maternal clinical
the incidence of hyperemesis are poor, but range from course, and perinatal outcomes associated with this
0.3% to 2% of pregnancies.3-5 Most of these estimates condition in the United States.
are from small cohort studies, many of which were
performed in the 1930s through the 1950s.3 Although
Methods
After obtaining Institution Review Board approval from
Support provided by the Women’s Reproductive Health Research MetroHealth Medical Center and the Committee for
(WRHR) Career Development Program; K12: HD98004.
Abstract presented as a poster at the Society for Maternal-Fetal
Human Protection in the State of California, we ob-
Medicine in Reno, Nev, February 10, 2005. tained the 1999 patient discharge data (PDD) linked
Reprints not available from the author. birth cohort data set from the State of California. The
doi:10.1016/j.ajog.2005.02.132
812 Bailit
average gestational age at admission, length of stay,

Table I Demographic characteristics of women with and
without hyperemesis charges, and admissions per patient were described.
Lastly, neonatal outcomes of mothers with and without
Patients with Patients without
hyperemesis were compared. We analyzed gestational
hyperemesis hyperemesis
age at delivery, birth weight, percent of infants weight-
Variable N = 2433* N = 518,306 P value
ing less than 2500 g, percent of infants weighting less
Maternal age (y) 27 27.7 !.0001 than 1500 g, fetal deaths after 20 weeks, neonatal
(avg)
deaths, large for gestational age (LGA) (greater than
White (%) 74 79.4 !.0001
Hispanic (%) 38.6 46.0 !.0001
90th percentile), and small for gestational age (SGA)
High school 59.4 58 .2 (less than 10th percentile) for singleton and multiple
education (%) gestations separately.6 Deaths were stratified by gesta-
Married (%) 65.0 69.2 !.0001 tional age (24-30, 31-36, 37-40, and greater than
Medicaid (%) 38.3 37.1 .22 41weeks). All comparisons were made with Student t
Nulliparity (%) 39.1 38.5 .53 tests and c2 where appropriate. Lastly, diagnoses of
Cesarean rate 26.2 22.4 !.0001 hyperemesis infants who died were examined.
overall (%)
Nulliparous 22.5 23.2 .62
cesarean rate (%)
Multiparous 28.5 21.9 !.0001 Results
cesarean rate (%)
Multiple 6.7 2.7 !.0001
In 1999 there were 520,739 live births in the State of
gestations (%) California that met our entry criteria. There were 2,466
patients who were admitted antenatally with a primary
* Demographic data were missing on 33 hyperemesis patients.
diagnosis of hyperemesis. The rate of hyperemesis was
473 per 100,000 live births. Thirty-three hyperemesis
PDD linked data set links birth certificate data with patients did not have matching birth records so that
maternal and neonatal hospital discharge records for 9 demographic data were not available on those patients.
months before and 12 months after delivery. All live Demographic characteristics of women with hyperemesis
births are recorded in this file, including those that occur are seen in Table I. Hyperemesis patients are more likely
out of a hospital setting. Data are also linked with infant to be younger and nonwhite. They are less likely to be
and fetal death certificates. Fetal death certificates are Hispanic or married. There are no differences in Medicaid
issued when the deceased fetus was greater than 20 coverage, nulliparity, or education level. Hyperemesis
weeks’ gestational age. patients are more likely to have a multiple gestation and
We removed records with a high likelihood for data overall more likely to have a cesarean delivery.
entry error. We removed maternal age younger than 11 The median gestational age for admission for hyper-
or older than 55 years. We removed birth weight less emesis is 11.3 weeks. There were a total of 3245
than 100 g or greater than 7000 g and infants delivered admissions with the primary diagnosis of hyperemesis.
after 45 weeks. To ensure that all hyperemesis admis- With 2466 patients who had the primary diagnosis of
sions were during the current pregnancy, we calculated hyperemesis, this means that there were an average of
the due date from the gestational age at delivery and the 1.3 admissions per patient. The average length of stay
delivery date. From this data, we calculated a concep- for a hyperemesis admission is 2.6 days. Admissions for
tion date, and excluded all admissions before the con- hyperemesis were no more likely to be on a weekend
ception date. We also excluded hyperemesis admissions than admissions for other conditions (30.3% vs 31%;
that were more than 2 weeks after the due date. P = .37). The average charge for an antenatal hyper-
Antenatal hospital admissions with the primary di- emesis admission was $5,932. Hyperemesis patients had
agnosis of hyperemesis were identified from Interna- slightly more infants anomalies (1.2% without hyper-
tional Classification of Diseases, ninth edition, (ICD-9) emesis vs 1.7% with hyperemesis P = .05) The total
codes. All ICD-9 codes for vomiting and malnutrition charges for all antenatal hyperemesis admissions in
inside and outside of pregnancy were used to identify California were $18,000,000 per year.
patients with hyperemesis, as 1 study has indicated that The average age at delivery for singleton with hyper-
up to 46% of hospital admissions on pregnant women emesis was statistically but not meaningfully decreased
do not have an obstetric code as the primary code.1 (Table II). The average birth weight was decreased in
ICD-9 codes used included: 643, 787, 261, 262, 263, singleton hyperemesis patients. Singleton patients with
276.5, 276.0, 269.8, and 269.9. We described demo- hyperemesis had more infants under 2500 g (7.8% vs
graphic characteristics of patients with and without the 5.1%; P ! .0001) and under 1500 g (1.54% vs 1.1%;
primary diagnosis of hyperemesis and compare them. P = .04) than singleton patients without hyperemesis.
Characteristics of the hyperemesis admissions, such as In addition, singleton hyperemesis patients had more
Bailit 813
Table II Singleton outcomes of patients with and without Table III Multiple gestation outcomes of patients with and
the primary diagnosis of hyperemesis without the primary diagnosis of hyperemesis
Patients with Patients without Patients with Patients without
hyperemesis hyperemesis Hyperemesis hyperemesis
Variable N = 2270 N = 486,505 P value Variable N = 163 N = 13,728 P value
Gestational age 39.0 39.2 !.0001 Gestational age 35.6 35.8 .32
at delivery (wks) at delivery (wks)
Birth weight (g) 3255 3380 !.0001 Birth weight (g) 2219 2371 .006
Birth weight 7.8 5.1 !.0001 Birth weight 65.0 51.4 .001
!2500 g (%) !2500 g (%)
Birth weight 1.54 1.1 .04 Birth weight 13.5 10.9 .29
!1500 g (%) !1500 g (%)
SGA (%) 29.21 20.8 !.0001 SGA (%) 75.5 73.6 .59
LGA (%) 9.8 11.2 .04 LGA (%) 1.2 4.4 .05
Fetal death 0.71 0.43 .05 Fetal death 2.45 1.47 .3
O20 wks (%) O20 wks (%)
Neonatal death 0.44 0.30 .07 Neonatal death 3.07 2.12 .58
overall (%) overall (%)
24-30 wks (%) 8.33 7.1 !.0001 24-30 wks (%) 0 9.82 .56
31-36 wks (%) 1.47 0.44 .09 31-36 wks (%) 0 .35 .75
37-40 wks (%) 0.06 0.08 .84 37-40 wks (%) 0 .09 .91
R41 wks (%) 0.23 0.21 .78 R41 wks (%) 16.6 3.32 .09
Maternal pregnancy 17,938 7,499 !.0001 Maternal pregnancy 26,258 17,014 !.001
charges charges
SGA infants (29.21 vs 20.8; P ! .0001) and fewer LGA live births in the form of birth certificate data. Thus, we
infants (9.8 vs 11.2; P = .04). There were more fetal present a ratio of hyperemesis to live births rather than a
deaths (0.71 vs 43; P = .05), but overall no difference in hyperemesis in pregnancy rate.
neonatal death rates. There were slightly more infant The numerator for hyperemesis is also difficult to
deaths in the neonatal period for infants born between determine. Many women have nausea and vomiting
24 and 30 weeks. Pregnancy hospitalization charges during their pregnancy that is considered normal and is
were significantly higher. treated as an outpatient.
Multiple gestations with hyperemesis and without To capture clinically meaningful hyperemesis, we
hyperemesis showed no difference in gestational age at chose to look at hyperemesis that was severe enough
delivery (Table III). Multiple gestations with hyperem- to require hospital admission. We classified patients as
esis have lower birth weights, more low birth weight having hyperemesis by looking at the primary diagnosis
infants, and higher hospitalization charges for preg- ICD-9 code. We did not describe patients who had a
nancy. Overall, they did not have higher rates of fetal or hyperemesis qualifying ICD-9 code as a secondary
neonatal death. diagnosis. It is possible that medically complex patients
Twenty-seven infants of mothers admitted for hyper- admitted for another reason also had hyperemesis.
emesis died. All but 3 died during the birth admission. Thus, our estimate of 473 per 100,000 deliveries should
Of the 27 infants, 13 were intrauterine demises. Of the 14 be considered conservative.
liveborn infants who died, 8 were premature and data on Previously published reports of hyperemesis have
gestational age was missing on 1. Three of the liveborn quoted a rate of 0.3% to 2%.3-5 The largest study on
infants were anomalous. hyperemesis was from 1973 to 1981 Swedish registry
data.5 They showed a hyperemesis rate of 0.3%. Our
Comment study has similar numbers of cases but needed only
1 year of data. The large number of cases in a short time
The rate of hyperemesis in pregnancy has been difficult increases our ability to study hyperemesis admissions
to estimate. The appropriate denominator for this rate is without having to account for changes in practice over
hard to define. Databases of all pregnancies in a pop- time. The 2 population-based studies, the Kallen study
ulation do not exist. From studies of early pregnancy, and ours, show a rate on the lower end of the published
we know that up to 31% of all conceptions are lost range.3-5
before 24 weeks and that 22% of those are not even Hyperemesis infants are born earlier and weight less,
clinically recognized.7 Population databases do exist for and in singletons, they have higher rates of being SGA.
814 Bailit
Although the 125 g difference between average weights nationally for inpatient management of hyperemesis.
is clinically meaningless, the increased number of SGA In April 2004, the American College of Obstetricians
infants is more concerning. Fetal death rates are also and Gynecologists (ACOG) published guidelines for
higher in hyperemesis singletons. However, from this treating nausea and vomiting of pregnancy.9 These were
study, we are unable to determine the cause of fetal evidence-based recommendations for efficacy. However,
death or whether these are directly related to hyperem- the first-line drugs recommended are far cheaper than
esis or not. Importantly, overall neonatal death rates do those at the bottom of the algorhythm. Charges for
not appear different. hyperemesis after ACOG recommendations should be
Although all live births in California require a birth investigated in the future to see if ACOG publish-
certificate be filed, it is likely that at previable gestational ing recommendations decreases hospitalization charges
ages, data may not be as complete. However, because we associated with hyperemesis.
were interested in fetal and neonatal deaths, we included
all gestational ages less than 45 weeks. Fetal death
certificates should include all fetuses that die over 20 Conclusion
weeks; however, it is possible that incomplete data for
Hyperemesis occurs in 473 per 100,000 live births and is
fetuses less than 20 weeks may affect our results. Thus,
associated with significant charges. Infants of mothers
caution should be in interpreting the increased fetal
with hyperemesis have lower average birth weights and
death rate among singletons with hyperemesis. Our fetal
are more likely to be SGA. Future studies should
death data should be hypothesis generating for future
incorporate inpatient as well as outpatient management
studies and it may not be appropriate to counsel patients
of hyperemesis to better characterize rates of hyperem-
with these numbers.
esis and the true cost of hyperemesis.
Our data did show demographic differences in who is
admitted for hyperemesis. Interestingly, insurance did
not seem to differ between the groups. This suggests that References
financial barriers to outpatient management may be less
important than other types of barriers. Women with 1. Bennett T, Kotelchuck M, Cox C, Tucker M, Nadeau D. Preg-
nancy-associated hospitalizations in the United States in 1991 and
hyperemesis were less likely to be married suggesting 1992: a comprehensive view of maternal morbidity. Am J Obstet
that lack of social support at home may contribute to Gynecol 1998;178:346-54.
the risk of hospitalization. Future studies that incorpo- 2. Broussard C, Richter J. Nausea and vomiting of pregnancy.
rate outpatient and inpatient management of hyperem- Gastroenterol Clin North Am 1998;27:123-51.
esis may be better able to discern characteristics that 3. Fairweather D. Nausea and vomiting in pregnancy. Am J Obstet
Gynecol 1968;102:135-75.
contribute to inpatient versus outpatient management 4. Tsang I, Katz V, Wells S. Maternal and fetal outcomes in hyper-
decisions. Physician preference may also play a part in emesis gravidarum. Int J Gynecol Obstet 1996;55:231-5.
the decision to hospitalize; however, we were not able to 5. Kallen B. Hyperemesis during pregnancy and delivery outcome: a
assess that from our study design and data sources. To registry study. Eur J Obstet Gynecol Reprod Biol 1987;26:291.
ascertain the true costs of hyperemesis, outpatient costs 6. Callen P. Ultrasonography in obstetrics and gynecology. New
York: WB Saunders; 2000.
as well as inpatient costs should be included. Our study 7. Wilcox AJ, Weinberg CR, O’Connor JF, Baird DD, Schlatterer JP,
only examines inpatient costs and as such underesti- Canfield RE, et al. Incidence of early loss of pregnancy. N Engl J
mates the dollars spent on the treatment of hyperemesis. Med 1988;319:189-94.
Charges for inpatient management of hyperemesis 8. Ventura S, Martin J, Curtin S, Menacker F, Hamilton B. Final data
are substantial totaling more than $18 million per year for 1999: national vital statistics reports. Hyattsville, Md: National
Center for Health Statistics; 2001.
in California alone. California performed 13% of all live 9. American College of Obstetricians and Gynecologists. Nausea and
births in the United States in 1999.8 Thus, nationally, we vomiting of pregnancy. ACOG practice bulletin no. 52. Obstet
can estimate that approximately $200,000,000 is charged Gynecol 2004;103:803-15.
www.ajog.org
The involvement of Rho-associated kinases in

agonist-dependent contractions of human
maternal and placental arteries at term
gestation
Mark Wareing, PhD,a Maureen O’Hara,a Fella Seghier, MSc,a
Philip N. Baker, DM,a Michael J. Taggart, PhDa,b,*
Maternal and Fetal Health Research Centre, St Mary’s Hospital,a and Smooth Muscle Physiology Group,b
Cardiovascular Research, University of Manchester, Great Britain
Received for publication September 17, 2004; revised January 5, 2005; accepted February 9, 2005
KEY WORDS Objective: The purpose of this study was to assess the involvement of rho kinase (ROK) in
Human arteries agonist-dependent contraction of omental, myometrial, and placental arteries of pregnant women
Rho kinase at term.
Pregnancy Study design: Wire myography was used to assess if contractions of intact or a-toxin-
permeabilized arteries obtained from women at elective cesarean section were influenced by the
ROK inhibitor Y-27632.
Results: Western blotting indicated the presence of ROKa in each of the 3 tissue types. In intact
human omental, myometrial, or placental arteries, Y-27632 dose-dependently reduced constric-
tions to the thromboxane-mimetic U46619. In permeabilized vessels, U46619 induced substantial
Ca2C-sensitization of contraction that was inhibited by Y27632. The phosphatase inhibitor
calyculin A induced a Ca2Csensitization of contraction similar to that of U46619 in permeabilized
omental arteries that was unaffected by Y-27632, suggesting that ROK may signal via myosin
phosphatase in these vessels.
Conclusion: These results are the first report of the involvement of ROK in the receptor-coupled
constriction of intact and permeabilized arteries from pregnant women.
In human pregnancy many dynamic vascular adapta- substantially, and alterations in receptor-mediated vas-
tions occur to accommodate changing physiologic cir- cular responsiveness may, at least partly, underlie mech-
cumstances. Cardiac output and blood volume increase anisms evoked to cope with this.1,2 Moreover, several
pregnancy complications, for example, preeclampsia and
intrauterine growth restriction, are associated with in vivo
* Reprint requests: Dr Michael Taggart, Maternal and Fetal
Health Research Centre, University of Manchester, St Mary’s Hos-
increased resistance of the maternal and/or placental
pital, Hathersage Road, Manchester M13 0JH, Great Britain. circulations, as well as postpregnancy increased risks
E-mail: Michael.j.taggart@man.ac.uk of maternal thrombosis and cerebral hemorrhage.3,4
doi:10.1016/j.ajog.2005.02.076
816 Wareing et al
In addition, these pregnancy conditions are linked by Material and methods

enhanced vascular sensitivity to constrictor agents.5,6 As
such, it becomes crucial to understand the mechanisms All biopsies were obtained following written informed
regulating agonist-mediated contraction of vascular tis- consent according to local ethical committee guidelines,
sues in normal human pregnancy. in compliance with the World Medical Association Dec-
Receptor-mediated contraction of smooth muscle laration of Helsinki. Myometrial and omental biopsies
involves an increase in intracellular calcium (Ca2Ci) and were obtained from women undergoing elective cesarean
subsequent phosphorylation of myosin regulatory light section at term (37-41 weeks) not in labor. Placental
chains (MLC20).7 An additional mechanism for force biopsies were obtained from women at term undergoing
enhancement exists whereby agonists induce a sensitiza- vaginal or elective cesarean section delivery.
tion of the contractile filaments to the activating cal- Small arteries were dissected from human biopsies
cium.8,9 Such Ca2C sensitization of contraction involves with the aid of a stereomicroscope, and placed in phys-
a sequelae of signalling events, including activation of iological salt solution (PSS) of composition: NaCl 119
heterotrimeric G-proteins, stimulation of the small G- mmol/L, KCl 4.7 mmol/L, MgSO4.7H2O 1.2 mmol/L,
protein rhoA, and subsequent activation of its down- NaHCO3 25 mmol/L, KH2PO4 1.17 mmol/L, K2EDTA
stream effector rho-associated kinase (ROK). A major 0.03 mmol/L, glucose 5.5 mmol/L, CaCl2.2H2O, 1.6
effect reported of ROK is the downstream inactivation of mmol/L at pH 7.4. Placental arteries were identified
the targeting subunit of myosin phosphatase (MYPT1), from the point of insertion of the umbilical artery, and
although this remains a point of some controversy.9-13 traced along the surface of the chorionic plate. Small
Inhibition of MYPT activity would be anticipated to arteries (!500 mm) were identified from successive
result in a net increase in MLC20 phosphorylation, a branching, and dissected free from surrounding tissue.
phenomenon that indeed accompanies Ca2C sensitiza- Myometrial small arteries were identified where they
tion of smooth muscle contraction in animals.10,14 entered into the myometrial surface of the biopsy and
In a wide variety of smooth muscles a water-soluble, dissected free of smooth muscle/connective tissue. Omen-
cell-permeant inhibitor of ROK, Y27632, has been tal small arteries were selected compared to veins by a
extremely effective in reducing agonist-dependent con- visual assessment of their wall to lumen ratio, and their
tractions in vitro of intact and permeabilized prepara- morphology at sites of branching before dissection free
tions.12,15-17 In addition, in various animal models from adipose tissue. Some vessel segments (10-20 mg
involving hyperstimulation of smooth muscle, such as weight) were snap frozen in liquid N2 and stored at
hypertension, coronary vasospasm, or atherosclerotic ÿ80(C until processed for SDS-PAGE and Western
lesions, ROK inhibition has successfully ameliorated blotting (see below). Other fresh arterial segments were
the complications.15,18,19 mounted on 2 40-mm wires in 5 to 7 mL PSS-containing
There are relatively fewer studies investigating the role chambers of a Danish Myotechnologies 610M myograph
of ROK activation in human blood vessels. In athero- (Danish Myotech, Aarhus, Denmark). Myometrial and
sclerotic coronary vessels20 or internal mammary arter- omental arteries were normalized as previously de-
ies21 obtained from bypass operations, or in omental scribed26 (MyoDaq software version 2.2, Danish Myo-
vessels22 from elderly patients undergoing bowel surgery, tech), which allows determination of the vessel stretch to
Y27632 reduced agonist-dependent contractions. How- be applied in order to give a tension equivalent to 0.9 of
ever, there is no information on the involvement of ROK L13.3kPa, where L13.3kPa is vessel diameter at which the
in constriction of human maternal or placental arteries. active effective pressure is 13.3 kPa. Placental arteries
In human myometrium itself, there are conflicting reports were normalized to a tension equivalent to 0.9 of L5.1kPa,
of the actions of ROK.23-25 Therefore, we investigated the where L5.1kPa is vessel diameter at which the active
contractile responses of arteries isolated from normal effective pressure is 5.1 kPa. The vessels were equilibrated
pregnant woman at term. We investigated arteries dis- for 20 minute (at 37(C for intact artery studies and 25(C
sected from biopsies of omentum, myometrium, and for permeabilized artery protocols).
placenta, as this allowed us to observe the contractile
responses in arteries of the maternal (omental and myo-
metrial) and placental circulations. We assessed if ROKa Intact tissue contractile studies
was expressed in these vessels, and examined whether the
ROK inhibitor Y27632 affected thromboxane-stimulated Control 2-minute exposures to high KC solution (60
contractions of intact arteries. We also determined the mmol/L KCl containing PSS isosmotically substituted
extent of sensitization at submaximal activating Ca2C in for NaCl) were performed to test for contractile viabil-
permeabilized human arteries evoked by thromboxane ity. Arteries were then constricted to the thromboxane
stimulation, or phosphatase inhibition, and investigated receptor agonist U46619 (0.1 nmol/L-1 mmol/L) for up
if Y27632 inhibited each of these phenomena to obtain to 15 minutes and washed in PSS. Vessels were
information on any mechanism of action of ROK. reconstricted to 1 mmol/L U46619 and, after 3 to 4
Wareing et al 817
minutes, exposed to incremental doses of the ROK

inhibitor Y27632 (0.1-10 mmol/L).
Permeabilized arterial contractile studies

In this set of experiments, PSS-equilibrated arteries were
Figure 1 Representation of the method of calculating ago-
exposed to a mock intracellular relaxing solution (com- nist-induced Ca2C sensitization of constriction in permeabi-
position: sodium creatine phosphate 10 mmol/L, lized vessels. A permeabilized omental artery was constricted
Na2ATP 5.2 mmol/L, magnesium methanesulphonate to activating solution pCa4.5 and the amplitude a noted. Upon
7.3 mmol/L, potassium methane sulphonate, 74 mmol/L, relaxation in pCa9.0 solution, the artery was exposed to
K2EGTA 1 mmol/L, buffered to pH 7.1 with 30 mmol/L submaximal activating solution pCa 6.7 plus 10 mmol/L
of the chemical buffer PIPES and addition of KOH) GTP. The difference between this tension and that subse-
before permeabilization with a-toxin. The solution is so quently achieved with 1 mmol/L U46619 addition reflected the
termed because it contains no added CaCl2. The remain- Ca2C sensitization of constriction, denoted as b. This was
ing constituents of the solution mimic an intracellular calculated relative to the constriction to pCa4.5 solution ((b/a)
! 100%). Filled bar for time scale (10 min).
environment. Vessels were then equilibrated for 10 to 15
minutes in external relaxing solution, which was then
aspirated and a 25-mL droplet of activating solution of 6.7 and pCa 4.5 contained 10 mmol/L EGTA, and were
pCa 6.7, where pCa = ÿlog10[Ca2C]. The activating pCa prepared by addition of the appropriate amount of
6.7 solution was supplemented with 500 U/mL a-toxin CaEGTA.27
plus 10 mmol/L A23187 (an ionophore that renders the
intracellular stores permeable to Ca2C) carefully placed SDS-PAGE and Western blotting
over the vessel. Permeabilization with a-toxin was as-
sumed to be complete when the ensuing constriction had Frozen arterial segments were homogenized at 4(C until
reached a plateau or, in cases of little constriction on completion in clean Duall glass tissue grinders in w250
permeabilization, up to 30 minutes. The artery was then mL of homogenization buffer (10% glycerol [v/v], 4%
returned to relaxing solution (pCa 9). The experimental sodium dodecyl sulfate (SDS; w/v), 1% Triton X-100 (v/v),
protocol then proceeded as depicted in the raw tracing of 20 mmol/L MOPS, pH 7.0, 10 mmol/L dithiothreitol,
Figure 1. The vessel was exposed to activating solution 20 mmol/L a-glycerophosphate, 5.5 mmol/L leupeptin,
pCa 4.5, and the ensuing constriction monitored. The 5.5 mmol/L pepstatin, 0.0205 TIU/mL aprotinin, 20
amplitude of this constriction is measured as a in Figure 1. mmol/L PMSF). Each homogenate was centrifuged at
On return to relaxing pCa 9 solution, the vessel relaxed. 13,000g for 5 minutes, the supernatants decanted, and the
Agonist-induced Ca2C sensitization of constriction was protein concentration of each sample estimated using
then examined as follows: vessels were exposed, in turn, Bio-Rad DC Protein Assay reagents (Bio-Rad, Hercules,
to submaximal activating solution pCa 6.7, solution pCa Calif). Following the addition of !2 Laemmli sample
6.7 plus 10 mmol/L GTP and GTP pCa 6.7 solution plus buffer (4% SDS, 20% glycerol, 10% 2-mercaptoethanol,
1 mmol/L U46619. The amplitude of the constriction 0.004% bromphenol blue, 0.125 mol/L Tris-HCl, pH 6.8)
recorded, above that of pCa6.7/GTP solution, upon the in a 1:1 ratio, the samples were heated at 100(C for 10
addition of U46619 was noteddthis is depicted as b in minutes, and subsequently stored at ÿ20(C until use.
Figure 1. Ca2C sensitization of constriction was then Tissue homogenates (10-20 mg) were loaded onto a 10%
calculated as the change in tension observed by addition separating polyacrylamide gel (0.1% SDS [w/v], 0.05%
of U46619 to pCa 6.7/GTP solution as a percentage of the ammonium persulfate, 1:2000 TEMED, 0.375 mol/L
change in tension seen to pCa4.5 solution ((b/a)!100% Tris-HCl, pH 8.8), and electrophoresed in Tris/glycine
in Figure 1). Once the constriction had reached a plateau, running buffer at 40 mA/gel for for 35 to 40 minutes.
the effect of Y27632 (10 mmol/L) on U46619-induced Electrophoretic transfer of proteins from the gels to
sensitization of contraction (to GTP pCa 6.7 solution) PVDF membranes was performed at 50 V/gel for 1.5
was examined. For each patient, any observed effects of hours at room temperature (blotting buffer composition:
Y27632 addition were compared to a simultaneous time 25 mmol/L Tris, 192 mmol/L glycine, 20% methanol, pH
control U44619-constricted vessel from the same patient 8.3). Blocked membranes (5% dried milk in Tris-buffered
run in parallel. In some experiments on permeabilized saline [TBS], 1 hour) were incubated overnight at 4(C in
omental arteries, the Ca2C-sensitizing action of the TBS with 1% dried milk with 1:500 anti-ROKa (BD
phosphatase inhibitor calyculin A (cal A, 2 mmol/L) Transduction Laboratories, Lexington, Ky). Washed
was examined by addition to GTP pCa 6.7 solution and membranes were then incubated with 1:2000 goat anti-
compared to the effect of 1 mmol/L U46619 in a parallel mouse peroxidase conjugate (CN Biosciences, Notting-
artery from the same patient. Activating solutions pCa ham, UK) secondary antibody for 1 hour at room
818 Wareing et al
temperature, and stained with enhanced chemilumines- Y27632 used (10 mmol/L), U46619 constrictions were
cence (ECL, Pierce Super Signal, Pierce, Rockford, Ill). significantly reduced to 61 G 6% (N = 8, Figure 3A),
44 G 8% (N = 4, n = 6, Figure 3B), and 79 G 5.6%
(N = 5, Figure 3C) of maximum in, respectively, omen-
Consumables tal, myometrial, and placental arteries (MANOVA).
Other than those materials mentioned above, chemicals The above experiments suggest that ROK activation
were purchased from Sigma-Aldrich Company, Ltd contributes to the contractions of intact human arteries.
(Poole, Dorset, UK) except the following: NaCl, The possible mechanism of action of ROK was further
CaCl2, NaHCO3, KHPO4, K2EDTA, acetic acid, and investigated in arteries permeabilized with a-toxin,
methanol were purchased from BDH Laboratory Sup- which retains receptor-coupled signalling mechanisms
plies, (Poole, Dorset, UK), acrylamide/bis acrylamide while allowing the concentration of Ca2C of the bathing
and filter paper from Bio-Rad Laboratories, Ltd, PVDF solution surrounding the myofilaments to be clamped.
membrane from Schleicher & Schuell UK Ltd (London, After permeabilization with a-toxin, exposure to pCa
UK), Super Signal West Pico Chemiluminescent Sub- 6.7 solution plus GTP induced submaximal contractions
strate from Pierce, Ltd, A23187, a-toxin, and Y27632 that were 10 G 2% (N = 7, n = 19, omental, Figure 4A
from CN Biosciences. A 10 mmol/L stock solution of and B), 16 G 6% (N = 3, n = 6, myometrial, Figure 4C
U46619 was made by dissolving in a mixture of 100% and D), and 30 G 11% (N = 3, n = 6, placental, Figure
ethanol and 1 mgkg-1 sodium carbonate (1:2). Y27632 4E and F) of those observed to pCa 4.5 solution.
was made as a 10 mmol/L stock dissolved in distilled Subsequent addition of 1 mmol/L U46619 to the pCa
water. a-toxin was dissolved in distilled water as a stock 6.7/GTP solution induced further pronounced tonic
solution of 2000 U/mL. contractions (Figure 4A to F). The magnitude of the
U46619-induced constrictions was not significantly dif-
ferent in omental (active effective pressure 8.8 G 0.9
Statistics kPa), myometrial (10.2 G 2.1 kPa), and placental (6.8 G
0.4 kPa) vascular beds (P ! .05; ANOVA). The
Active effective pressure (kPa) was calculated from the U46619-induced Ca2C sensitization of contraction was
wall tension (mN/mm) divided by the internal radius of 173 G 47% (N = 7, n = 13, omental), 131 G 19%
the vessel (mm). Dose response curves were analyzed by (N = 3, n = 6, myometrial), and 54 G 15% (N = 3,
multifactoral analysis of variance (MANOVA) followed n = 6, placental) of contraction to pCa 4.5 solution.
by Bonferroni post hoc test to indicate significance Ten mmol/L Y27632 significantly reduced this Ca2C
between data points. A comparison of U46619-induced sensitization of contraction in all 3 artery types (Figure
arterial vasoconstrictions was performed using one-way 4B, D, and E). In the presence of Y27632, the U46619-
ANOVA followed by Bonferroni post hoc test. In all induced Ca2C sensitization of contraction was reduced
other instances, differences between data points were to 0.68- G 0.06-fold (N = 7, n = 8, omental, Figure
analyzed by unpaired or paired Student t test as appro- 5A), 0.75- G 0.04-fold (N = 3, myometrial, Figure 5B),
priate. Significance was indicated by P ! .05. Values are and 0.41- G 0.02-fold (N = 3, placental, Figure 5C) of
quoted as mean G standard error of the mean, with that in the absence of ROK inhibitor.
n = number of samples, and N = number of patients. In other smooth muscle tissues, it has been proposed
that ROK activation results in a reduction of MYPT
Results activity. Therefore, we sought to establish if direct inhi-
bition of phosphatase activity in permeabilized omental
The normalized mean diameters of vessels used in this arteries, with application of calyculin A, induced a sig-
study were 363 G 39 mm (N = 15 patients, n = 27 nificant Ca2C sensitization. As illustrated in Figure 6C,
vessels), 265 G 28 mm (N = 8, n = 13), and 216 G 23 application of calyculin A to pCa 6.7/GTP solution in-
mm (N = 8, n = 10) for, respectively, omental, myome- duced substantial Ca2C sensitization of contraction that,
trial, and placental arteries. Western blotting of homo- when compared with U46619-stimulated Ca2C sensitiza-
genates of each arterial type from 3 separate women tions in arteries from the same patient, was of a similar
indicated that ROKa was expressed in omental, myome- magnitude. On average, the U46619-induced Ca2C sensi-
trial, and placental arteries (Figure 2A). tization of contraction was 256 G 95% of pCa 4.5,
U46619 produced maintained constrictions of similar whereas the calyculin A-mediated Ca2C sensitization was
magnitude in arteries from omental (active effective 392 G 217% of pCa 4.5. Furthermore, while 10 mmol/L
pressure 12.7 G 3.1 kPa), myometrial (14.0 G 3.2 kPa), Y27632 significantly reduced the U46619-induced Ca2C
and placental (12.1 G 1.0 kPa) vascular beds (Figure 2B sensitization to 0.57- G 0.08-fold of that in the absence
to D; MANOVA). The ROK inhibitor Y27632 dose- of ROK inhibitor, similar to reported above, it was with-
dependently reduced U46619 constrictions in each vessel out effect on calyculin A-dependent Ca2C sensitization of
type (Figures 2 and 3). At the highest concentration of contraction (N = 3, Figure 6E).
Wareing et al 819
Figure 2 ROK expression in human arteries, and inhibition of contractions by the ROK inhibitor Y27632. Upper panel: ROKa
expressions in each artery type (A). Labels 1 to 3 refer to arterial samples from separate women. Lower panels: Representative
tracings of the tonic constrictions of human omental (B), myometrial (C), and placental (D) arteries to 1 mmol/L U46619 (left traces)
together with the inhibitory effect of cumulative doses (0.1-10 mmol/L) of the ROK inhibitor Y27632 (right traces). Parallel lines
within traces represent a break in the experimental recordings.
820 Wareing et al
significant clinical conditions,20-22 suggest that ROK

stimulation is important in regulating arterial contractil-
ity in humans in both physiologic and pathophysiologic
circumstances, respectively.
ROK stimulation has been suggested to contribute to
agonist-induced Ca2C sensitization of force in a wide
variety of animal smooth muscle tissues.7,8,12,15-17 This
phenomenon can best be investigated in permeabilized
tissues where the intracellular milieu, particularly the
concentration of calcium surrounding the myofilaments,
can be controlled by the experimenters. In addition,
receptor-coupled signaling of these preparations remains
intact due to the small pore size created in the plasma
membrane by a-toxin. In the a-toxin-permeabilized hu-
man omental, myometrial, and placental artery types
studied here, where the activating Ca2C is clamped at a
suprabasal submaximal level, we found that U46619
resulted in pronounced Ca2C sensitizations of force.
Furthermore, Y27632 significantly reduced these ago-
nist-mediated Ca2C sensitizations in all 3 human artery
types when compared to time controls in the absence of
Y27632. As far as we are aware, this is the first direct
demonstration of agonist-mediated Ca2C sensitization of
permeabilized arteries from pregnant women, and that
these are inhibited by Y27632. These data, thus, illustrate
that arteries of the maternal and placental circulations
utilize Ca2C sensitization pathways for constriction and
implicate such mechanisms as a means of governing
arterial constriction in pregnancy.
One possible physiologic mechanism whereby ROK
regulates Ca2C sensitization of contraction is by inhibi-
tion of MYPT activity. In permeabilized human omental
arteries we sought to find out if inhibition of phosphatase
activity with calyculin A affected Ca2C sensitization of
force, and if Y27632 could inhibit any such contraction.
Calyculin A was found to induce substantial Ca2C
sensitizations of force that were of similar magnitude to
that induced in arteries from the same patients by
U46619. This indicates that even at submaximal activat-
Figure 3 Mean data demonstrating the attenuation of ing Ca2C there is substantial phosphatase activity in these
U46619-induced contractions of human omental (A), myome- human arteries, in agreement with that observed previ-
trial (B), and placental (C) arteries by the ROK inhibitor ously in animal smooth muscles.28 Furthermore, Y27632
Y27632. did not affect the calyculin A-dependent constrictions
even though it did reduce by approximately 43% the
Comment U46619-induced Ca2C sensitizations in parallel arteries
from the same patients. These experiments, thus, indicate
In this study we report that ROKa is expressed in human that a likely mechanistic effect of the ROK inhibitor
omental, myometrial, and placental blood vessels iso- Y27632 on U46619-dependent contractions of human
lated from normal pregnant women at term. Y27632, a arteries is an enhancement of myosin phosphatase activ-
pharmacologic inhibitor of ROK, reduces the U46619- ity. Future experiments should elucidate if this is as a
induced contractions of each of these 3 artery types, result of altered MYPT phosphorylation directly, or of an
implicating a role of ROK activation in agonist-induced indirect action on the activity of the phosphatase inhib-
contractions of human adult and fetal arteries of preg- itor protein CPI-17.13
nant women. The data presented here from biopsies of In summary, we have reported that, in maternal and
normal pregnant women, together with that published placental arteries from pregnant women, the ROK in-
previously on other arteries obtained from patients with hibitor Y27632 attenuates agonist-induced contractions
Wareing et al 821
Figure 4 Representative tracings of a-toxin permeabilized human omental (A), myometrial (C), and placental (E) arteries
illustrating Ca2C-induced constrictions on changing from relaxing pCa 9 solution to maximal activating pCa 4.5 solution, and also
the Ca2C-sensitizing constriction of 1 mmol/L U46619 in GTP pCa 6.7 submaximal activating solution. Ten mmol/L Y27632
significantly reduced the U46619-induced Ca2C sensitization in placental (B), myometrial (D), and omental (F) arteries. Filled
bar for time scale (10 min). Ten mmol/L Y27632 significantly reduced the U46619-induced Ca2C-sensitization in omental
(B), myometrial (D), and placental (F) arteries.
822 Wareing et al
Figure 5 Mean data indicating the significant reduction in 1 mmol/L U46619-dependent Ca2C sensitization of contraction
(normalized as 1.0) by 10 mmol/L Y27632 in permeabilized human omental (A), myometrial (B), and placental (C) arteries.
Wareing et al 823
Figure 6 Data exhibiting the effect of phosphatase inhibition with calyculin A (cal A) in permeabilized human omental arteries.
Representative tracings from arteries from the same patient, run in parallel, are shown of pCa 4.5 activating solution contractions
(A–D) and 1 mmol/L U46619 (A, B) or 2 mmol/L cal A-induced (C, D) Ca2C-sensitizing contractions in GTP pCa 6.7 submaximal
activating solution. Ten mmol/L Y27632 significantly reduced the U46619-induced Ca2C sensitization (B, mean data in E), but did
not affect the cal A-dependent Ca2C sensitizations (D, mean data in E). Filled bar in panels A to D for time scale (10 min).
824 Wareing et al
of intact and permeabilized vessels most likely via an 11. Ichikawa K, Ito M, Hartshorne DJ. Phosphorylation of the large
action on myosin phosphatase activity. This places ROK subunit of myosin phosphatase and inhibition of phosphatase
activity. J Biol Chem 1996;271:4733-40.
activation as a key component of receptor-coupled signal 12. Sward K, Dreja K, Susnjar M, Hellstrand P, Hartshorne DJ,
transduction mechanisms regulating human arterial con- Walsh MP. Inhibition of rho-associated kinase blocks agonist-
tractility during pregnancy. Y27632 has been shown to induced Ca2C-sensitisation of myosin phosphorylation and force
inhibit arterial contractility in in vitro and in vivo in guinea-pig ileum. J Physiol 2000;522:33-49.
assessments of arterial function in animal models of 13. Kitazawa T, Eto M, Woodsome TP, Khalequzzaman M. Phos-
phorylation of the myosin phosphatase targeting subunit and CPI-
vascular complications. Therefore, it will be of interest in 17 during Ca2C-sensitisation in rabbit smooth muscle. J Physiol
the future to examine if clinical complications of preg- 2003;546:879-89.
nancy that are associated with elevated vascular resis- 14. Kitazawa T, Gaylinn BD, Denney GH, Somlyo AP. G-protein-
tances, for example preeclampsia and intrauterine growth mediated Ca2C-sensitisation of smooth muscle contraction through
restriction, are associated with increased ROK activation myosin light chain phosphorylation. J Biol Chem 1991;266:1708-15.
15. Ueheta M, Ishizaki T, Satoh H, Ono T, Kawahara T, Morishita T,
and attenuated by administration of Y27632. et al. Calcium sensitisation of smooth muscle mediated by a Rho-
associated protein kinase in hypertension. Nature 1998;389:990-4.
Acknowledgments 16. Fu X, Gong MC, Jia T, Somlyo AV, Somlyo AP. The effects of the
Rho-kinase inhibitor Y27632 on arachidonic acid-, GTPlS- and
This work was supported by Tommy’s, the baby charity. phorbol ester-induced Ca2C-sensitisation of smooth muscle. FEBS
We would like to thank the research midwifes and Lett 1998;440:183-7.
clinical fellows of MFHRC and St Mary’s Hospital, and 17. Ishizaki T, Ueheta M, Tamechika I, Keel J, Nonomura K,
Maekawa M, et al. Pharmacological properties of Y27632, a specific
the patients of St Mary’s Hospital for their cooperation inhibitor of rho-associated kinases. Mol Pharmacol 2000;57:976-83.
in this study. M. W. is a British Heart Foundation 18. Shimokawa H, Morishige K, Miyata K, Kandabashi T, Eto Y,
Intermediate Research Fellow. Ikegaki I, et al. Long-term inhibition of Rho-kinase induces a
regression of arteriosclerotic coronary lesions in a porcine model in
vivo. Cardiovasc Res 2001;51:169-77.
References 19. Morishige K, Shimokawa H, Eto Y, Kandabashi T, Miyata K,
Matsumoto Y, et al. Adenovirus-mediated transfer of dominant-
1. D’Angelo G, Osol G. Regional variation in resistance artery negative Rho-kinase induces a regression of coronary arterioscle-
diameter responses to alpha-adrenergic stimulation during rosis in pigs in vivo. Arterioscler Thromb Vasc Biol 2001;21:
pregnancy. Am J Physiol 1993;264:H78-85. 548-54.
2. Hermsteiner M, Zoltan DR, Kunzel W. The vasoconstrictor 20. Kandabashi T, Shimokawa H, Mukai Y, Matoba T, Kunihiro I,
response of uterine and mesenteric resistance arteries is differen- Morikawa K, et al. Involvement of rho-kinase in agonist-induced
tially altered in the course of pregnancy. Eur J Obstet Gynaecol contractions of arteriosclerotic human arteries. Arterioscler
Reprod Biol 2001;100:29-35. Thromb Vasc Biol 2002;22:243-8.
3. Wilson BJ, Watson MS, Prescott GJ, Sunderland S, Campbell 21. Batchelor TJP, Sadaba JR, Ishola A, Pacaud P, Munsch CM,
DM, Hannaford P, et al. Hypertensive diseases of pregnancy and Beech DJ. Rho-kinase inhibitors prevent agonist-induced vaso-
risk of hypertension and stroke in later life: results from cohort spasm in human internal mammary artery. Brit J Pharmacol
study. BMJ 2003;326:845-51. 2001;132:302-8.
4. Kestenbaum B, Seliger SL, Easterling TR, Gillen DL, Critchlow 22. Martinez MC, Randriamboavonjy V, Ohlmann P, Komas N,
CW, Stehman-Breen CO, et al. Cardiovascular and thromboem- Duarte J, Schneider F, et al. Involvement of protein kinase C,
bolic events following hypertensive pregnancy. Am J Kidney Dis tyrosine kinases and Rho kinase in Ca2C handling of human small
2003;42:982-9. arteries. Am J Physiol 2000;279:H1228-H38.
5. Aalkjaer C, Danielsen H, Johannesen P, Pedersen EB, Rasmussen 23. Kupittayanant S, Burdyga T, Wray S. The effects of inhibiting
A, Mulvany MJ. Abnormal vascular function and morphology in Rho-associated kinase with Y-27632 on force and intracellular
pre-eclampsia: a study of isolated resistance vessels. Clin Sci calcium in human myometrium. Pflugers Arch 2001;443:112-4.
(Lond) 1985;69:477-82. 24. Moran CJ, Friel AM, Smith TJ, Cairns M, Morrison JJ. Expres-
6. Pascoal IF, Lindheimer MD, Nalbantian-Brandt C, Umans JG. sion and modulation of Rho kinase in human pregnant myome-
Pre-eclampsia selectively impairs endothelium-dependent relaxa- trium. Mol Human Reprod 2002;8:196-200.
tion and leads to oscillatory activity in small omental arteries. 25. Woodcock NA, Taylor CW, Thornton S. Effect of an oxytocin
J Clin Invest 1998;101:464-70. receptor antagonist and rho kinase inhibitor on the [CaCC]i
7. Hai CM, Murphy RA. Ca2C, crossbridge phosphorylation and sensitivity of human myometrium. Am J Obstet Gynecol
contraction. Ann Rev Physiol 1989;51:285-98. 2004;190:222-8.
8. Sward K, Mita M, Wilson DP, Deng JT, Susnjar M, Walsh MP. 26. Wareing M, Bhatti M, O’Hara M, Kenny L, Warren AY, Taggart
The role of RhoA and Rho-associated kinase in vascular smooth MJ, et al. Vasoactive effects of neurokinin B on human blood
muscle contraction. Curr Hypertens Rep 2003;5:66-72. vessels. Am J Obstet Gynecol 2003;188:196-202.
9. Somlyo AP, Somlyo AV. Ca2C-sensitivity of smooth muscle and 27. Horiutu K. Mechanism of contracture on cooling of caffeine-
nonmuscle myosin. II. Modulated by g-proteins, kinases and treated frog skeletal muscle fibres. J Physiol 1988;398:131-48.
myosin phosphatase. Physiol Rev 2003;83:1325-58. 28. Gong MC, Cohen P, Kitazawa T, Ikebe M, Masuo M, Somlyo
10. Kimura K, Ito M, Amano M, Chihara K, Fukata Y, Nakafuku M, AP, et al. Myosin light chain phosphatase activities an the effects
et al. Regulation of myosin phosphatase by Rho and Rho- of phosphatase inhibitors in tonic and phasic smooth muscle.
associated kinase. Science 1996;273:245-8. J Biol Chem 1992;267:14662-8.
www.ajog.org
Novel peptides prevent alcohol-induced spatial learning

deficits and proinflammatory cytokine release in a
mouse model of fetal alcohol syndrome
Joy Vink, BA,* Jonathan Auth, MD, Daniel T. Abebe, BS, Douglas E. Brenneman, PhD,
Catherine Y. Spong, MD
Unit on Perinatal and Developmental Neurobiology, National Institute of Child Health and Human Development,
National Institutes of Health, Bethesda, MD
Received for publication October 11, 2004; revised January 2, 2005; accepted February 17, 2005
KEY WORDS Objective: Previously, the novel peptides NAPVSIPQ and SALLRSIPA were shown to prevent
Alcohol alcohol-induced fetal death and growth abnormalities in a mouse model of fetal alcohol
Spatial learning syndrome. This study evaluated whether these peptides could prevent long-term alcohol-induced
Proinflammatory learning abnormalities. In addition, because specific cytokines are known to effect long-term
cytokine potentiation, a model of learning at the molecular level, we studied the effect of these novel
peptides on tumor necrosis factor–a, interleukin-6, and interferon-g levels.
Study design: We used a well-characterized mouse model of fetal alcohol syndrome. Pregnant
mice were injected on day 8 with alcohol (0.03 mL/kg) or placebo. Pretreatment with
NAPVSIPQ C SALLRSIPA (20 mg) or placebo was given 30 minutes before alcohol. Embryos
were removed after 6 hours, at which time cytokine, tumor necrosis factor–a, interleukin-6, and
interferon-g levels were measured with enzyme-linked immunoassays. To test spatial learning,
adult offspring from litters that were treated with alcohol, control, NAPVSIPQ C SALLRSIPA
then alcohol, or NAPVSIPQ C SALLRSIPA alone were evaluated for latency to find a hidden
platform in the Morris water maze.
Results: Alcohol treatment increased tumor necrosis factor–a levels versus control levels (50.0 G
3.5 pg/mL vs 32.7 G 2.4 pg/mL; P ! .001). NAPVSIPQ C SALLRSIPA pretreatment prevented
this increase (39.9 9 G 2.8 pg/mL; P % .01), with levels similar to control (P = .1). Similarly,
alcohol increased interleukin-6 levels versus control levels (22.6 G 1.4 pg/mL vs 17.3 G 0.6 pg/mL;
P ! .001), and NAPVSIPQ C SALLRSIPA prevented this increase (19.1 G 1.0 pg/mL; P % .02),
with levels similar to control levels (P = .2). Interferon-g levels were not different among the 3
groups (alcohol, 14.6 G 4.9 pg/mL; control, 17.9 G 6.6 pg/mL; alcohol C NAPVSIPQ C
SALLRSIPA, 13.6 G 4.9 pg/mL; P = .2). In the Morris water maze, alcohol-treated groups did
not learn over the 7-day trial compared with the control group (P = .001). Groups that were
pretreated with NAPVSIPQ C SALLRSIPA then alcohol learned significantly, which was similar
to the control group. Groups that were treated with only NAPVSIPQ C SALLRSIPA learned
significantly earlier, with the shortest latency once learning commenced.
* Reprint requests: Joy Vink, BA, Unit NICHD, NIH, on Perinatal and Developmental Neurobiology, Building 9, 1W116, MSC 0925, 9000
Rockville Pike, Bethesda, MD 20892.
E-mail: jyv5y@virginia.edu
doi:10.1016/j.ajog.2005.02.101
826 Vink et al
Conclusion: The peptides, NAPVSIPQ C SALLRSIPA, prevented the alcohol-induced spatial

learning deficits and attenuated alcohol-induced proinflammatory cytokine increase in a model of
fetal alcohol syndrome. This study demonstrates the peptides’ significant in vivo efficacy with
long-lasting effects obtained after prenatal administration.
Fetal alcohol syndrome (FAS) is the most common efficacy of peptide treatment in preventing the long-term
nongenetic cause of mental retardation and includes cognitive deficits caused by prenatal alcohol exposure.
prenatal alcohol exposure, growth restriction, distinct
facial features, and neurodevelopmental problems, par- Material and methods
ticularly life-long compromises in learning and memory.1
In the United States, it affects 0.5 to 3 babies per 1000 We used a well-characterized mouse model of FAS.13
births each year1; in the 1990s, the prevalence of mater- C57Bl6/J female mice (Jackson Laboratories, Bar Har-
nal alcohol consumption during pregnancy, including bor, Maine) were kept under a 12-hour light, 12-hour
binge drinking, has increased.2 Additionally, many fe- dark regimen, with food and water available at all times.
tuses are exposed to alcohol but do not meet the full The mice received humane animal care in compliance
criteria for FAS. Depending on a number of factors that with the ‘‘Guideline for Care and Use of Experimental
are not well elucidated (including amount, timing, and Animals.’’ Six-week-old females (21-24 g) were mated
genetic predisposition), these children may also have with C57Bl6/J males for 4 hours. On gestational day 8,
significant sequelae from alcohol exposure, which in- we injected pregnant mice intraperitoneally with 25%
cludes neurodevelopmental and learning difficulties. The ethyl alcohol in saline (vol/vol) or vehicle alone at 0.03
mechanisms by which alcohol induces the characteristic mL/g body weight. Pretreatment with the peptides
physical and cognitive deficits are complex and not well ADNF-9 and NAP (20 mg in 0.2 mL) or placebo was
understood. However, the effects of prenatal alcohol given 30 minutes before the alcohol. NAP was diluted
administration have been paralleled in animal models, in 50 mL DMSO and diluted in filtered Dulbecco’s
which implies that dysfunctional mechanisms or neuro- phosphate-buffered saline solution. ADNF-9 was dissolved
nal plasticity are responsible for the persistent neuro- and diluted in filtered Dulbecco’s phosphate-buffered
behavioral sequelae from prenatal alcohol exposure.3 saline solution. The doses of the peptides were based on
Animal studies that used the Morris water-maze, the protective effects of these peptides in the prevention
which is a widely used test of spatial learning, have of alcohol-induced fetal death, growth restriction, and
shown that alcohol exposure prenatally,4 binge,5 or microcephaly.9 Because the animals that received the
during the brain growth spurt6 results in adult spatial alcohol were incapacitated for approximately 6 hours
learning deficits. Because spatial learning is particularly after the injection, food and water were withheld from
affected in FAS, we used this model, which requires all groups for the initial 6 hours.
spatial ‘‘mapping’’ based on localization of extra-maze Adult male offspring (R35 and not O50 days of age)
cues.7 Previous studies have also shown that proinflam- were then tested in the Morris water-maze to assess
matory cytokines, interleukin-6 (IL-6) and tumor necro- spatial learning. Male offspring from the litters that were
sis factor-alpha (TNF-a) affect long-term potentiation,8 treated with alcohol (n = 9 mice), vehicle (control,
a model of learning at the molecular level. Thus, these n = 15 mice), pretreatment with NAP C ADNF-9 and
cytokines were measured as markers for alcohol-induced alcohol (n = 15 mice), or NAP C ADNF-9 alone
damage, perhaps giving insight into the basis of FAS- (n = 8 mice) were tested for latency to find a hidden
related cognitive deficits. platform in the water maze. Each animal underwent 2
Previously, we demonstrated that treatment with 2 consecutive trials daily for 7 days. Each trial allowed the
novel peptides, SALLRSIPA (ADNF-9) and NAPVSIPQ mouse a maximum of 1 minute to find the hidden
(NAP), prevented alcohol-induced fetal growth restric- platform. The latency for each trial was recorded, and
tion, microcephaly, and oxidative damage in the mouse the average of the 2 trials was calculated for each of the
model of FAS.9 These peptides (which are derived from 7 days. The average latency for each of the 7 days was
the glial proteins, activity-dependent neurotrophic fac- then analyzed by analysis of variance (StatView, version
tor,10 and activity-dependent neuroprotective protein,11 5.0.1; SAS Institute Inc., Cary, NC); a probability value
respectively) have broad neuroprotective action evident of ! .05 was considered significant.
at subfemtomolar levels and a unique pharmacologic To further delineate how the peptides prevented the
makeup because they mimic the activity of their parent alcohol-induced learning deficits, 2 inflammatory cyto-
proteins.12 Given the neuroprotective properties of these kines (TNF-a and IL-6) and an immunosuppressive
peptides, the objective of this study was to evaluate the cytokine (interferon gamma [IFN-g]) as an internal
Vink et al 827
control were measured in the embryo and surrounding

decidua of each mouse fetus. At least 2 embryos from
6 different litters per cytokine were tested. Embryo/
deciduas were dissected 6 hours after treatment with
alcohol, control, or alcohol plus peptide treatment. They
were quick frozen on dry ice and stored at ÿ80(C. All
samples were analyzed for total protein content, which
was not found to be different between the groups. With
the use of an enzyme-linked immunoassay (R&D
Systems, Minneapolis, Minn), samples were assayed in
duplicates, and the mean value was used for analysis.
Each enzyme-linked immunosorbent assay that was
performed was specific for its respective cytokine with-
Figure Pretreatment with NAP C ADNF-9 prevents alcohol-
out cross-reactivity or interaction with other cytokines. induced learning deficits. Male offspring from litters that were
The sensitivities of the assays for all 3 cytokines (TNF-a, treated with alcohol (n = 9), control litters (n = 15), litters
IL-6, and IFN-g) were 1.95 pg/mL. The intra-assay that were pretreated with NAP C ADNF-9 and alcohol
coefficients of variation were !5%. Data were analyzed (n = 15), or litters that were treated with NAP C ADNF-9
by Mann-Whitney U test (StatView 5.0.1); a probability alone (n = 8) were tested twice daily for 7 days for latency to
value of ! .05 was considered significant. find a hidden platform in the Morris water maze. Offspring
from the alcohol litters did not learn in this paradigm, with
latency on day 7 similar to days 1 to 3. Offspring from the
Results
other 3 groups did learn significantly (decreased latency) over
In the learning paradigm, offspring from the alcohol- the trials (P ! .001). Males from the NAP C ADNF-9 alone
group had a significantly earlier onset of learning than the
treated litters demonstrated no evidence of learning over
other groups (P ! .01). The asterisks denote the significant
the 7-day trial (Figure). In contrast, animals from the
differences on days 4, 5, 6, and 7.
control litters decreased their latency 50% by the fifth
day (P!.001). Males from the litters who were pre-
treated with NAP C ADNF-9 and then given alcohol learning deficits in affected offspring. The data also
also significantly learned, with a learning curve not imply that the peptides, NAP C ADNF-9, prevent
different from that of the control at all time points that alcohol-induced increases in proinflammatory cytokines
were tested. The offspring from litters that were treated and protect against alcohol-induced spatial learning
with only NAP C ADNF-9 had a learning curve with 2 deficits. In addition, this study suggests that treatment
distinguishing features: The onset of learning was sig- with NAP C ADNF-9 alone enhanced learning in nor-
nificantly earlier than all other groups (P ! .01), and the mal mice, when compared with control. The data are
latency period was the shortest of all groups after the consistent with previous work in which neonatal mice
onset of learning. that were treated with intraperitoneal NAP alone
In the cytokine paradigm, embryo/decidua TNF-a exhibited increased performance in the Morris water-
levels were elevated significantly in the alcohol-treated maze.14 Cumulatively, these data suggest that peptide
group (50.0 G 3.5 pg/mL) versus control (32.7 G 2.4 treatment that is initiated in utero significantly influen-
pg/mL; P ! .001). NAP C ADNF-9 pretreatment ces spatial learning, possibly through cytokine-mediated
prevented the alcohol-induced increase in TNF-a mechanisms.
(39.9 G 2.8 pg/mL; P %0.01), with levels not different This study confirmed previous studies that showed
from control (P = .1). Similarly, embryo/decidua IL-6 that alcohol administration to pregnant mice produces
levels were elevated in the alcohol-treated group impairment of spatial learning and memory perfor-
(22.6 G 1.4 pg/mL) versus control (17.3 G 0.6 pg/mL; mance15,16 and that prenatal combined peptide treat-
P ! .001), and NAP C ADNF-9 prevented the alcohol- ment effectively prevented alcohol-induced learning and
induced increase in IL-6 (19.1 G 1.0 pg/mL; P % .02), memory deficits in mature mice that were exposed in
with levels similar to control (P = .2). IFN- g levels utero to alcohol. Although these data unveil some clues
were not different among the 3 groups (alcohol, to the mechanism responsible for alcohol-induced loss-
14.6 G 4.9 pg/mL; control, 17.9 G 6.6 pg/mL; alcohol C of-function, the complete picture remains to be delin-
NAP C ADNF-9, 13.6 G 4.9 pg/mL; P = .2). eated. Previous studies suggest the loss of brain neurons
caused by alcohol toxicity may be produced by multi-
Comment factorial mechanisms, which include increases in cyto-
kines,17 decreased glutamate neurotransmission,18
This study showed that alcohol treatment increases interference with neurotrophic factor signaling or
proinflammatory cytokine levels and produces spatial expression,19 and increases in oxidative stress through
828 Vink et al
free radical generation.20 Therefore, previous studies source of ethanol-related free radical generation.26 In
regarding NAP and ADNF-9 action that involve these addition, the generation of 1-hydroxyethyl radicals and
factors are relevant to understanding the effectiveness acetaldehyde from xanthine oxidase metabolism of eth-
of treatment with these peptides. anol has been detected; this pathway could play a role in
Regarding the protective effects of the peptides on toxic effects that are associated with alcohol.27 Both
learning and memory, recent studies of hippocampal NAP and ADNF-9 have been shown to provide pro-
cultures demonstrated that the peptides alter glutamate tective action against oxidative stress that is produced
release and N-methyl-D-aspartate (NMDA) receptors,21 by a number of agents, including ferrous sulfate,28
both of which are important in learning and memory. hydrogen peroxide,29 glutathione,4 and beta amyloid
Particularly, hippocampal neurons that were treated peptide.30 These studies suggest that the peptides have
with ADNF-9 exhibited increased frequency of minia- effects that prevent oxidative damage and therefore
ture postsynaptic currents, which suggests a presynaptic provide an additional mechanism to protect from
site as a target of the peptide-induced signaling system.21 alcohol-induced damage.
Additionally, ADNF-9 treatment controls NR2A and In summary, there is no single mechanism that can
NR2B subunit stability of the NMDA receptor in account for all the toxic effects that are produced by
neurons that have yet to establish efficient synaptic ethanol or the protective action of the peptides. As
connections.21 Although the exact mechanism by which shown in the present study, the breadth and duration of
ADNF-9 influences the synaptogenesis and neurotrans- the protective action that is provided by the peptides
mission used by glutamate remains unknown, in vitro strongly supports broad overlap between the multifac-
studies strongly suggest that the peptide interacts with torial mechanisms that produce damage that is associ-
and regulates the glutaminergic synapses in developing ated with alcohol toxicity and the effective multifactorial
neural systems. In addition, NMDA antagonists are mechanisms that are elicited by NAP and ADNF-9.
known to produce apoptosis in the developing brain,
and NMDA protects cerebellar neurons from alcohol References
toxicity. This protection was correlated with increases
in nitric oxide and cyclic guanosine monophosphate 1. Stratton K, Howe C, Battaglia F, editors. Fetal alcohol syndrome:
levels.22 Thus, peptide-induced enhancement of gluta- diagnosis, epidemiology, prevention and treatment. Washington
mate neurotransmission/synaptogenesis may have a (DC): National Academy Press; 1996.
2. Ebrahim SH, Anderson AL, Floyd RL. Alcohol consumption by
protective effect on alcohol-induced decreases in NMDA reproductive-aged women in the USA: an update on assessment,
receptor stimulation. burden and prevention in the 1990s. Prenatal Neonatal Med
Support for a relationship between alcohol and 1999;4:419-30.
cytokine production was observed in a trophoblast cell 3. Hannigan JH. What research with animals is telling us about
line, which exhibited increases in IL-6, RANTES, and alcohol-related neurodevelopmental disorder. Pharmacol Biochem
Behav 1996;55:489-99.
granulocyte colony-stimulating factor after ethanol 4. Matthews DB, Simson PE. Prenatal exposure to ethanol disrupts
treatment.17 Importantly, astrocyte cultures that were spatial memory: effect of the training-testing delay period. Physiol
prepared from rats that were treated prenatally with Behav 1998;64:63-7.
alcohol were more sensitive to the cytotoxic effects of 5. Girard TA, Xing HC, Ward GR, Wainwright PE. Early postnatal
TNF-a than cultures that were obtained from control ethanol exposure has long-term effects on the performance of male
rats in a delayed matching-to-place task in the Morris water maze.
rats.23 Previous studies have shown that NAP prevented Alcohol Clin Exp Res 2000;24:300-6.
increases in TNF-a after closed head injury and that 6. Goodlett CR, Kelly SJ, West JR. Early postnatal alcohol exposure
NAP treatment prevented TNF-a–induced toxicity in that produces high blood-alcohol levels impairs development of
PC12 cells.24 Thus, the inhibition of proinflammatory spatial navigation learning. Psychobiology 1987;15:64-74.
cytokine production is another plausible mechanism 7. Morris R. Developments of a water-maze procedure for studying
spatial learning in the rat. J Neurosci Meth 1984;11:47-60.
through which the peptides could prevent ethanol- 8. Jankowsky JL, Patterson PH. Cytokine and growth factor involve-
induced toxicity. In our study, the effect of combined ment in long-term potentiation. Mol Cell Neurosci 1999;14:
peptide treatment on cytokine levels is consistent with 273-86.
this action, which accounts for the observed protective 9. Spong CY, Abebe DT, Gozes I, Brenneman DE, Hill JM.
effects. Prevention of fetal demise and growth restriction in a mouse
model of fetal alcohol syndrome. J Pharmacol Exp Ther 2001;
Another attractive mechanism is the link between 297:774-9.
ethanol toxicity and oxidative stress. Although evidence 10. Brenneman DE, Hauser J, Neale E, Rubinraut S, Fridkin M,
for such a relationship in the central nervous system is Davidson A, et al. Activity-dependent neurotrophic factor:
not yet apparent, the generation of free radicals has been structure-activity relationships of femtomolar-acting peptides.
proposed as a likely mechanism for alcohol-induced J Pharmacol Exp Ther 1998;285:619-20.
11. Bassan M, Zamostiano R, Davidson A, Pinhasov A, Giladi E, Perl
cytotoxicity.20 Indeed, antioxidant-mediated protection O, et al. Complete sequence of a novel protein containing
from ethanol-induced toxicity has been observed.25 femtomolar-activity-dependent neuroprotective peptide. J Neuro-
Recent studies have implicated NADPH oxidase as a chem 1999;72:1283-93.
Vink et al 829
12. Brenneman DE, Spong CY, Gozes I. Protective peptides derived 22. Pantazis NJ, West JR, Dai D. The nitric oxide-cyclic GMP
from novel glial proteins. Biochem Soc Trans 2000;28:452-5. pathway plays an essential role in both promoting cell survival
13. Webster WS, Walsh DA, Lipson AH, McEwen SE. Teratogenesis of cerebellar granule cells in culture and protecting the cells against
after acute alcohol exposure in inbred and outbred mice. ethanol neurotoxicity. J Neurochem 1998;70:1826-38.
Neurotoxicol Teratol 1980;2:227-34. 23. De Vito WJ, Xhaja K, Stone S. Prenatal alcohol exposure
14. Giladi E, Davidson A, Bachar M, Bardea A, Rubinraut S, Fridkin increases TNF-a-induced cytotoxicity in primary astrocytes.
M, et al. Protection against developmental and learning impair- Alcohol 2000;21:63-71.
ments in apolipoprotein E-deficient mice by activity-dependent 24. Beni-Adani L, Gozes I, Cohen Y, Assaf Y, Steingart R,
femtomolar-acting peptides. Neurosci Lett 1997;48(suppl):19. Brenneman DE, et al. A peptide derived from activity-dependent
15. Sutherland RJ, McDonald RJ, Savage DD. Prenatal exposure neuroprotective protein (ADNP) ameliorates injury response in
to moderate levels of ethanol can have long-lasting effects on closed head injury in mice. J Pharmacol Exp Ther 2001;296:
learning and memory in adult offspring. Psychobiology 2000; 57-63.
28:532-9. 25. Mitchell JJ, Paiva M, Heaton MB. The antioxidants vitamin E and
16. Kim CK, Kalynchuk LE, Kornecook TJ, Mumby DG, Dadgar beta-carotene protect against ethanol-induced neurotoxicity in
NA, Pinel JP, et al. Object-recognition and spatial learning and embryonic rat hippocampal cultures. Alcohol 1999;17:163-8.
memory in rats prenatally exposed to ethanol. Behav Neurosci 26. Kono H, Rusyn I, Ming Y, Gabele E, Yamashina S, Dikalova A,
1997;111:985-95. et al. NADPH oxidase-derived free radicals are key oxidants in
17. Svinarich DM, DiCerbo JA, Zaher FM, Yelian FD, Gonik B. alcohol-induced liver disease. J Clin Invest 2000;106:867-72.
Ethanol-induced expression of cytokines in a first-trimester tropho- 27. Castro GD, Delagdo de Layno AM, Costantini MH, Castro JA.
blast cell line. Am J Obstet Gynecol 1998;179:470-5. Rat ventral prostate xanthine oxidase bioactivation of ethanol to
18. Ikonomidou C, Bittigau P, Ishimaru MJ, Wozniak DF, Koch C, acetaldehyde and 1-hydroxyethyl free radicals: analysis of its
Genz K, et al. Ethanol-induced apoptotic neurodegeneration and potential role in heavy alcohol drinking tumor-promoting effects.
fetal alcohol syndrome. Science 2000;287:1156-60. Teratogen Carcin Mut 2001;21:109-19.
19. McAlhany RE, Miranda RC, Finnell RH, West JR. Ethanol 28. Glazner GW, Boland A, Dresse AE, Brenneman DE, Gozes I,
decreases glial derived neurotrophic factor (GDNF) protein release Mattson MP. Activity-dependent neurotrophic factor peptide
but not mRNA expression and increases GDNF-stimulated shc (ADNF9) protects neurons against oxidative stress-induced
phosphorylation in the developing cerebellum. Alcohol Clin Exp death. J Neurochem 1999;73:2341-7.
Res 1999;23:1691-7. 29. Steingart RA, Solomon B, Brenneman DE, Fridkin M, Gozes I.
20. Zima T, Fialova L, Mestek O, Janebova M, Crkovska J, VIP and peptides related to activity-dependent neurotrophic factor
Malbohan I, et al. Oxidative stress, metabolism of ethanol and protect PC12 cells against oxidative stress. J Mol Neurosci
alcohol-related diseases. J Biomed Sci 2001;8:59-70. 2000;15:137-45.
21. Blondel O, Collin C, McCarran WJ, Shiaoping Z, Zamostiano R, 30. Zemlyak I, Furman S, Brenneman DE, Gozes I. A novel peptide
Gozes I, et al. A glia-derived signal regulating neuronal differen- prevents death in enriched neuronal cultures. Regul Peptides
tiation. J Neurosci 2000;20:8012-20. 2000;96:39-43.
www.ajog.org
Neutrophils from pregnant women produce thromboxane

and tumor necrosis factor-a in response to linoleic acid
and oxidative stress
John E. Vaughan, PhD, Scott W. Walsh, PhD*
From the Departments of Obstetrics and Gynecology and Physiology, Virginia Commonwealth University Medical
Center, Richmond, VA
KEY WORDS Objective: Preeclampsia is associated with oxidative stress, neutrophil activation, neutrophil
Preeclampsia infiltration into systemic vasculature, and elevated plasma levels of linoleic acid, the fatty acid
Neutrophils precursor to arachidonic acid and its metabolite, thromboxane. In this study we evaluated
Oxidative stress whether linoleic acid under conditions of oxidative stress would stimulate neutrophil production
Linoleic acid of thromboxane and tumor necrosis factor-a.
Thromboxane Study design: Neutrophils were isolated from 14 normal pregnant women. Western blot
Tumor necrosis factor demonstrated cyclooxygenase-2 expression at 18 hours of incubation, so this incubation time was
alpha used for experiments. Neutrophils (2 ! 106 cells/mL) were incubated in Dulbecco’s modified
Lipid peroxides Eagle’s medium/F-12 with: (1) linoleic acid control; (2) an oxidizing solution enriched with
linoleic acid; (3) oxidizing solution enriched with linoleic acid plus indomethacin; (4) oxidizing
solution enriched with linoleic acid plus aspirin; (5) oxidizing solution enriched with linoleic acid
plus NS-398, a specific inhibitor of cyclooxygenase-2; or (6) oxidizing solution enriched with
linoleic acid plus pinane thromboxane, a thromboxane synthase inhibitor and receptor blocker.
Results: Oxidizing solution enriched with linoleic acid significantly increased oxidative stress in
neutrophils. Compared with linoleic acid, oxidizing solution enriched with linoleic acid
significantly increased neutrophil production of thromboxane and tumor necrosis factor-a.
Indomethacin and aspirin inhibited oxidizing solution enriched with linoleic acid stimulation of
thromboxane, but NS-398 was equally effective implicating cyclooxygenase-2 in the thromboxane
response. Indomethacin inhibited oxidizing solution enriched with linoleic acid stimulation of
tumor necrosis factor-a, but so did pinane thromboxane implicating thromboxane in the tumor
necrosis factor-a response.
Conclusions: These data demonstrate that exposure of neutrophils from normal pregnant women
to conditions present in preeclamptic women results in neutrophil activation with release of
thromboxane and tumor necrosis factor-a. Newly synthesized thromboxane is cyclooxygenase-2
dependent and plays a role in the tumor necrosis factor-a response. Our data suggest a mechanism
for maternal vasoconstriction and vascular inflammation in preeclampsia because activated,
Supported by grants (to S.W.W.) from the National Institutes of Health (HD 20973 and HL069851).
* Reprint requests: Scott W. Walsh, PhD, Virginia Commonwealth University Medical Center, Department of Obstetrics and Gynecology, P.O.
Box 980034, Richmond, VA 23298-0034.
E-mail: swwalsh@vcu.edu
doi:10.1016/j.ajog.2005.01.057
Vaughan and Walsh 831
thromboxane-secreting neutrophils migrate across endothelium into the microenvironment of the

vasculature in which they could promote vasoconstriction, whereas release of tumor necrosis
factor-a could cause vascular inflammation.
Preeclampsia is associated with neutrophil activation1; gravid and multigravid patients between the ages of 19
oxidative stress2; and elevated plasma levels of linoleic and 40 years. The racial/ethnic composition was 8 black
acid (LA),3 the fatty acid precursor to arachidonic acid women, 4 white women, 1 Hispanic woman, and 1 Asian
and its inflammatory, vasoconstrictor metabolite, throm- woman. This study was approved by the Office of
boxane. To date, no satisfactory explanation for maternal Research Subjects Protection, Virginia Commonwealth
hypertension or vascular dysfunction has been offered University. The blood was processed in the laboratory
and substantiated. Recently we demonstrated inflamma- within an hour following sample collection. Neutrophils
tion and infiltration of neutrophils in systemic vascular were isolated by dual-histopaque density gradient cen-
tissue in women with preeclampsia.4 Neutrophils that trifugation (Sigma Chemical Company, St Louis, MO).
have infiltrated the vasculature could cause endothelial Neutrophils were aspirated from the histopaque-1077/
and vascular smooth muscle dysfunction by release of histopaque-1119 interface and washed with phosphate-
inflammatory and vasoconstrictive substances, such as buffered saline, followed by centrifugation for 10
tumor necrosis factor-a (TNFa) and thromboxane. minutes at 200 ! g. To lyse contaminating red blood
Neutrophil activation may be initiated in the inter- cells, the cell pellet was suspended in ice-cold double-
villous space by increased secretion of lipid peroxides by distilled water (3 mL) for 30 seconds with agitation on
the placenta. The placental production rate of lipid a vortex mixer. Ice-cold 0.6 M potassium chloride
peroxides is abnormally increased in preeclampsia, and (1 mL) was then added to restore tonicity. The cells
placental lipid peroxides are primarily secreted into the were pelleted, the supernatant was discarded, and the cells
intervillous space.5,6 Studies on isolated perfused pla- were resuspended for counting in 1 mL of Dulbecco’s
cental cotyledons have demonstrated that peroxides modified Eagles’s medium (DMEM)/F-12 (Gibco, In-
stimulate cyclooxygenase (COX) to increase placental vitrogen Corp, Grand Island, NY) supplemented with
thromboxane production,6,7 which is abnormally 10% heat-inactivated fetal bovine serum. Cell viability
increased in preeclampsia.8 Monocytes also produce was greater than 90% as assessed by trypan blue
increased amounts of thromboxane in preeclampsia.9 exclusion staining.
Lipid peroxides have been shown to stimulate mono-
cyte/macrophage release of TNFa,10 and thromboxane Western blot analysis
has been implicated in monocyte TNFa release.11
Elevated levels of lipid peroxides in preeclampsia may To determine the time course for COX-2 expression,
similarly activate neutrophils to release thromboxane neutrophils were divided into 15 ! 100 mm nonadherent
and TNFa. Teflon tubes (Minisorp, Nunc, Rochester, NY) (2 ! 106
We hypothesized that exposure of neutrophils to cells/tube). COX-2 expression was assessed at baseline (0
conditions present in preeclamptic women would result hours) and after 2, 4, 6, and 18 hours of incubation in
in increased production of thromboxane and TNFa and a final volume of 1 mL with either LA (90 mM) or an
that these effects would be mediated by increased oxidizing solution containing LA (OxLA). The oxidizing
activity of the inducible form of COX-2. To test this, solution was composed of hypoxanthine (1.8 mM) C
neutrophils were isolated from the blood of normal xanthine oxidase (0.005 units/mL) C ferrous sulfate (50
pregnant women and incubated with an oxidizing mM) plus LA (90 mM). After incubation, cells were rinsed
solution enriched with linoleic acid in the presence and with phosphate-buffered saline and then lysed with buffer
absence of inhibitors of various steps in the arachidonic (100 mL) consisting of 50 mM TRIS (pH 7.5), 1% Nonidet
acid pathway to thromboxane. P-40, 100 mM NaCl, 1 mM ethyleneglycol-bis-(-b-amino-
ethyl ether)-N,N,N,N-tetra-acetic acid, 1mM EDTA, 1
mM aminoethyl benzene sulfonyl fluoride hydrochloride,
5 mM NaF, and 10% glycerol. Cell lysates were diluted
with buffer (0.5 mol/L TRIS-HCl, 0.1% glycerol, 10%
Neutrophil isolation sodium dodecyl sulfate, 0.05% 2-methanol) to equalize
protein concentrations and loaded on a 9% polyacryl-
Blood samples were obtained from 14 normal pregnant amide gel for electrophoresis. Proteins were electropho-
women during routine blood draws between 26 and 30 retically transferred to a nitrocellulose membrane for 1
weeks of gestation by vein puncture into sodium heparin hour at 65 V. After overnight blocking with a 5% solution
tubes. The study population consisted of both primi- of dry milk in Tris-buffered saline, the membrane was
832 Vaughan and Walsh
Figure 1 Neutrophils were isolated from heparinized blood

collected from pregnant women. Neutrophils (2 ! 106 cells/
mL) were assessed at baseline (0 hours) and after incubation
for 2, 4, 6, and 18 hours with LA or OxLA (Ox) to establish
a time course for the expression of COX-2 protein. There was
slight expression of COX-2 at baseline and at 2, 4, and 6 hours,
probably due to isolation procedures. By 18 hours, cells
exposed to LA showed some expression, but cells exposed to
OxLA showed marked expression of COX-2 protein. An 18-
hour incubation period with 2 ! 106 cells/mL per tube was
chosen for subsequent experiments. The immunoblot is repre- Figure 2 TBARS were measured as an indicator of oxidative
sentative of 3 experiments. stress. OxLA significantly increased media TBARS levels as
compared with LA after 18 hours of incubation of neutrophils
obtained from pregnant women. Data represent mean G SE,
incubated for 2 hours with mouse COX-2 monoclonal n = 9. Double asterisk, P ! .01.
antibody (1:1000, Cayman Chemical, Ann Arbor, MI).
Blots were then incubated for 1 hour with secondary
mixer (VXR-510, Teckmar, West Germany). The as-
antibody, a goat monoclonal antimouse IgG antibody
sembly was placed in an incubator, and the tubes were
conjugated to horseradish peroxidase (1:5000, Santa
incubated overnight with agitation at 37(C in an
Cruz Biotechnology, Santa Cruz, CA). Western Light-
incubator gassed with 5% CO2. After 18 hours the
ning Chemiluminescence Reagent Plus enhanced luminol
tube contents were removed and centrifuged at 200 ! g
(PerkinElmer Life Sciences, Boston, MA) was used
for 5 minutes. The supernatant was aliquoted and stored
to develop horseradish peroxidase using Kodak film
at –20(C until analysis.
(Hyperfilm MP, Amersham Pharmacia, Piscataway, NJ).
Assays
Experimental conditions
Lipid peroxide assay
To assess neutrophil production of thromboxane and
Lipid peroxides were estimated by an improved analysis
TNFa in response to oxidative stress, neutrophils were
of thiobarbituric acid reactive substances (TBARS),
diluted in DMEM/F-12 with 10% heat-inactivated fetal
which primarily reflects malondialdehyde, a breakdown
bovine serum to a final concentration of 8 ! 106 cells/mL.
product of lipid peroxides.12 Tetramethoxypropane was
Aliquots of the cell suspension (250 mL) were added to
used to generate malondialdehyde for the standard
15 ! 100 mm nonadherent Teflon tubes (Minisorp, Nunc,
curve. Butylated hydroxytoluene was added to prevent
Rochester, NY) containing 750 mL of DMEM/F-12
oxidation during the heating step with thiobarbituric
media with experimental treatments. The number of
acid. Freshly thawed samples were analyzed. Serially
neutrophils isolated from each blood draw limited the
diluted samples were parallel to the standard curve and
number of treatments that could be done, so experiments
within- and between-assay variations were ! 10%. We
were divided as follows. For assessment of oxidative
have previously shown that TBARS concentrations
stress, treatments were LA control and OxLA. For
highly correlate with 8-isoprostane concentrations for
assessment of thromboxane B2 (TXB2) production, treat-
media samples and accurately reflect lipid peroxidation.5
ments were: (1) LA control; (2) OxLA; (3) OxLA plus
indomethacin at a dose sufficient to inhibit phospholipase Thromboxane assay
A2 (100 mM); (4) OxLA plus aspirin (100 mM) to inhibit Thromboxane concentrations were estimated by specific
COX-1 and COX-2; and (5) OxLA plus NS-398 (300 nM), radioimmunoassay of its stable metabolite, TXB2. TXB2
a specific COX-2 inhibitor. For assessment of TNFa standard was purchased from PerSeptive Diagnostics,
production, treatments were: (1) LA control; (2) OxLA; Inc (Cambridge, MA), and TXB2 antibody was pur-
(3) OxLA plus indomethacin; and (4) OxLA plus pinane chased from Oxford Biomedical Research, Inc (Oxford,
thromboxane (10 mM), a thromboxane synthase inhibitor MI). Tritiated TXB2 was purchased from New England
and thromboxane receptor blocker. Nuclear (Dupont Research, Wilmington, DE). Dextran-
LA, indomethacin, and aspirin were purchased from coated charcoal was used to separate the bound from
Sigma. Pinane thromboxane was purchased from Cay- the free fraction. Serial sample dilutions were parallel to
man. Treatments were run in duplicate. The tubes were the standard curve. Assay of media alone resulted in
placed in a slant rack that was then secured to an orbital a zero dose response. Addition of increasing sample
Figure 3 Release of TXB2 into the media after 18 hours of

Figure 4 Release of TNFa into the media after 18 hours of
incubation of neutrophils obtained from pregnant women with
incubation of neutrophils obtained from pregnant women with
LA, OxLA, or OxLA plus inhibitors. OxLA significantly
LA, OxLA, or OxLA plus inhibitors. OxLA significantly
increased neutrophil TXB2 production as compared with
increased neutrophil TNFa production as compared with LA.
LA. The ability of OxLA to stimulate neutrophil TXB2
Indomethacin (Indo) abolished the ability of OxLA to stimulate
production was prevented by indomethacin (Indo) and aspirin
TNFa. Pinane thromboxane, a thromboxane synthase inhibitor
(ASA). NS-398, a specific inhibitor of COX-2, was equally
and receptor blocker, significantly inhibited OxLA-induced
effective in inhibiting TXB2 production. Data represent mean
TNFa production. Data represent mean G SE, n = 11. Double
G SE, n = 8. Asterisk, P ! .05, compared with other
asterisk, P ! .01, compared with other treatments.
treatments.
volumes resulted in a linear dose response. Within-assay for cells exposed to OxLA. An 18-hour incubation time
variation was 4.5% and between-assay variation was period was used for the TXB2 and TNFa experiments.
14.3%. To verify that OxLA induced oxidative stress, we
incubated neutrophils with LA or OxLA for 18 hours
TNF-a assay
and assayed the media for TBARS. TBARS were
TNFa concentrations were determined by enzyme-linked
significantly higher in the media of cells exposed to
immunosorbent assay using a commercially available
OxLA (529 G 85 nM per 1 million cells) as compared
reagent kit (OptEIA Human TNFa Set, PharMingen,
with those exposed to LA (233 G 50 nM per 1 million
San Diego, CA). Assay of varying volumes of cell
cells) (Figure 2), indicating that treatment with OxLA
homogenate resulted in a linear response, Y = 0.739x –
induced oxidative stress.
4.225, r2 = 0.991. Within-assay variation was 2.5%.
OxLA significantly increased neutrophil production
Statistical analyses of TXB2 as compared with LA (542 G 23 versus 421 G
29 pg per 1 million cells, respectively) (Fig. 3). Both indo-
Experimental data were analyzed by 1-way analysis of methacin and aspirin inhibited the ability of OxLA to
variance with Newman-Keul’s post hoc test. A statistical stimulate TXB2 (443 G 29 and 413 G 22 pg per 1 million
computer software program was used for analysis cells, respectively). NS-398 was equally as effective as
(GraphPad Prism 4.0 for Macintosh, GraphPad Soft- indomethacin and aspirin at inhibiting TXB2 production
ware, Inc, San Diego, CA, www.graphpad.com). A in the presence of OxLA (435 G 35 pg per 1 million cells).
probability level of P ! .05 was considered to be OxLA significantly increased the production of TNFa
statistically significant. Data are presented as mean G SE. by neutrophils as compared with LA (99 G 26 versus
20 G 8 pg per 1 million cells, respectively) (Figure 4).
Indomethacin abolished the production of TNFa in-
Results duced by OxLA (2.3 G 0.5 pg per 1 million cells). When
neutrophils were exposed to OxLA plus pinane throm-
Western blot analysis was done to assess the time course
boxane, neutrophil TNFa production was significantly
for COX-2 protein expression when neutrophils were
inhibited (23 G 8 pg per 1 million cells).
exposed to oxidative stress. Figure 1 shows that there
was very slight expression of COX-2 protein at baseline
(0 hours) and after 2, 4, and 6 hours of incubation, Comment
probably resulting from the isolation procedure. By 18
hours, there was some expression of COX-2 protein for To date, no satisfactory explanation for maternal
cells exposed to LA but marked expression of COX-2 hypertension or vascular cell dysfunction in women
834 Vaughan and Walsh
with preeclampsia has been offered and substantiated. produce a long-lasting activation of monocytes in non-
Recently we demonstrated inflammation and infiltration pregnant subjects. Our finding that OxLA stimulates
of activated neutrophils in the maternal systemic vascu- neutrophils from pregnant women to release TNFa
lature in preeclamptic women.4 Therefore, one possibil- suggests a critical link between oxidative stress and
ity for hypertension and vascular cell dysfunction is immune system dysfunction in preeclampsia.
release of thromboxane and inflammatory mediators, To investigate whether the OxLA-induced increase in
such as TNFa, by neutrophils that have infiltrated the neutrophil TNFa required thromboxane, we used in-
vasculature. Because oxidative stress is a prominent hibitors of enzymes that lead to thromboxane synthesis.
feature in preeclampsia, we investigated whether oxida- Indomethacin completely abolished TNFa production,
tive stress stimulates neutrophils to produce thrombox- and pinane thromboxane, a thromboxane receptor
ane and TNFa. We used OxLA because the plasma antagonist and thromboxane synthase inhibitor, signif-
concentrations of LA in preeclampsia are significantly icantly reduced OxLA-induced neutrophil TNFa pro-
elevated as compared with normal pregnancy,3 and LA duction. Thromboxane may mediate increased TNFa by
is the dietary precursor for arachidonic acid, which is enhancing the effect of nuclear factor-kB.19
metabolized to thromboxane. Numerous reports indicate that leukocytes are acti-
OxLA induced expression of COX-2 protein and vated in preeclampsia1. Activated neutrophils produce
caused a significant increase in neutrophil thromboxane TNFa,20 and the activities of this cytokine may have
production, compared with the LA control. Inhibiting important effects on the progression of preeclampsia.
phospholipase A2 abolished the OxLA-induced increase TNFa interacts with fatty acids to induce oxidative
in thromboxane. These results suggest that neutrophils stress and dysfunction in endothelial cells.21 TNFa
contribute to increased thromboxane levels in disorders could also contribute to the pathophysiology of pre-
marked by hyperlipidemia and oxidative stress, such as eclampsia by stimulating expression of endothelial intra-
preeclampsia. cellular cell adhesion molecule-1, causing adherence of
To resolve the mechanism whereby OxLA increased neutrophils to endothelial cells.22,23 We recently demons-
neutrophil thromboxane production, we focused on the trated increased endothelial expression of intracellular
activity of COX because the level of peroxide tone cell adhesion molecule-1 coincident with neutrophil
modulates this enzyme’s activity. COX exists in two adherence to the endothelium in preeclamptic women.4
main isoforms: COX-1, a constitutively active form, and The results of this study show that exposing neutro-
COX-2, an inducible form during inflammation. To phils from normal pregnant women to conditions
examine the potential roles of these enzymes, we used present in preeclamptic women results in expression of
aspirin, an inhibitor of COX-1 and COX-2, and NS-398, COX-2 and increased thromboxane production. The
a specific COX-2 inhibitor. Aspirin completely blocked newly synthesized thromboxane plays a role in mediat-
the OxLA-induced increase in thromboxane. NS-398 ing neutrophil TNFa production. These results, com-
was equally as effective as aspirin in abolishing throm- bined with our recent finding that activated neutrophils
boxane production. These results suggest that of the 2 infiltrate the maternal systemic vasculature in pre-
COX isoforms, COX-2 plays the predominant role in eclampsia, link oxidative stress and immune dysfunction
thromboxane production induced by OxLA. Our find- in this disorder and offer mechanistic explanations for
ings are in agreement with recent studies showing maternal vasoconstriction and vascular dysfunction.
selective COX-2 inhibitors can attenuate prostaglandin
E2 and TNFa release by activated neutrophils13 and
block thromboxane production stimulated by various References
neutrophil activators (fMet-Leu-Phe, phorbol myristic
acid, lipopolysaccharide, TNFa, opsonized zymosan).14 1. Clark P, Boswell F, Greer IA. The neutrophil and preeclampsia.
Semin Reprod Endocrinol 1998;16:57-64.
A study by Caughey et al11 implicated thromboxane 2. Walsh SW. Maternal-placental interactions of oxidative stress and
in TNFa production by human monocytes from non- antioxidants in preeclampsia. Semin Reprod Endocrinol 1998;
pregnant patients. In preeclampsia TNFa is abnormally 16:93-104.
increased in the maternal circulation15,16 and placenta.17 3. Lorentzen B, Drevon CA, Endresen MJ, Henriksen T. Fatty acid
We hypothesized that OxLA would stimulate neutrophil pattern of esterified and free fatty acids in sera of women with
normal and pre-eclamptic pregnancy. Br J Obstet Gynaecol
TNFa release through a thromboxane-dependent mech- 1995;102:530-7.
anism. We found that OxLA markedly stimulated 4. Leik CE, Walsh SW. Neutrophils infiltrate resistance-sized vessels
TNFa production as compared with LA in neutrophils of subcutaneous fat in women with preeclampsia. Hypertension
isolated from pregnant women. In nonpregnant pa- 2004;44:72-7.
tients, Jovinge et al10 showed that oxidized low-density 5. Walsh SW, Vaughan JE, Wang Y, Roberts LJ II. Placental
isoprostane is significantly increased in preeclampsia. FASEB J
lipoprotein increased the expression of TNFa messenger 2000;14:1289-96.
ribonucleic acid and stimulated release of TNFa in 6. Walsh SW, Wang Y. Secretion of lipid peroxides by the human
human monocytes. Gorog18 found that oxidized lipids placenta. Am J Obstet Gynecol 1993;169:1462-6.
7. Walsh SW, Wang Y, Jesse R. Peroxide induces vasoconstriction in fluid of patients with severe preeclampsia. Am J Obstet Gynecol
the human placenta by stimulating thromboxane. Am J Obstet 1994;170:1752-9.
Gynecol 1993;169:1007-12. 16. Williams MA, Farrand A, Mittendorf R, Sorensen TK, Zingheim
8. Walsh SW. Preeclampsia: an imbalance in placental prostacyclin RW, O’Reilly GC, et al. Maternal second trimester serum tumor
and thromboxane production. Am J Obstet Gynecol 1985;152: necrosis factor-alpha-soluble receptor p55 (sTNFp55) and sub-
335-40. sequent risk of preeclampsia. Am J Epidemiol 1999;149:323-9.
9. Chen G, Wilson R, Cumming G, Walker JJ, McKillop JH. 17. Wang Y, Walsh SW. TNFa concentrations and mRNA expression
Production of prostacyclin and thromboxane A2 in mononuclear are increased in preeclamptic placentas. J Reprod Immunol 1996;
cells from preeclamptic women. Am J Obstet Gynecol 1993; 32:157-69.
169:1106-11. 18. Gorog P. Activation of human blood monocytes by oxidized
10. Jovinge S, Ares MP, Kallin B, Nilsson J. Human monocytes/ polyunsaturated fatty acids: a possible mechanism for the gener-
macrophages release TNF-alpha in response to Ox-LDL. Arte- ation of lipid peroxides in the circulation. Int J Exp Pathol
rioscler Thromb Vasc Biol 1996;16:1573-9. 1991;72:227-37.
11. Caughey GE, Pouliot M, Cleland LG, James MJ. Regulation of 19. Ishizuka T, Kawakami M, Hidaka T, Matsuki Y, Takamizawa M,
tumor necrosis factor-alpha and IL-1 beta synthesis by thrombox- Suzuki K, et al. Stimulation with thromboxane A2 (TXA2) receptor
ane A2 in nonadherent human monocytes. J Immunol 1997;158: agonist enhances ICAM-1, VCAM-1 or ELAM-1 expression by
351-8. human vascular endothelial cells. Clin Exp Immunol 1998;112:464-70.
12. Jentzsch AM, Bachmann H, Furst P, Biesalski HK. Improved 20. Edwards SW. Biochemistry and physiology of the neutrophil. New
analysis of malondialdehyde in human body fluids. Free Radic York (NY): Cambridge University Press; 1994.
Biol Med 1996;20:251-6. 21. Toborek M, Barger SW, Mattson MP, Barve S, McClain CJ,
13. Kimura T, Iwase M, Kondo G, Watanabe H, Ohashi M, Ito D, Hennig B. Linoleic acid and TNF-a cross-amplify oxidative injury
et al. Suppressive effect of selective cyclooxygenase-2 inhibitor on and dysfunction of endothelial cells. J Lipid Res 1996;37:123-35.
cytokine release in human neutrophils. Int Immunopharmacol 22. d’Alessio P, Moutet M, Coudrier E, Darquenne S, Chaudiere J.
2003;3:1519-28. ICAM-1 and VCAM-1 expression induced by TNF-alpha are
14. Pouliot M, Gilbert C, Borgeat P, Poubelle PE, Bourgoin S, inhibited by a glutathione peroxidase mimic. Free Radic Biol Med
Creminon C, et al. Expression and activity of prostaglandin 1998;24:979-87.
endoperoxide synthase-2 in agonist activated human neutrophils. 23. Gamble JR, Harlan JM, Klebanoff SJ, Vadas MA. Stimulation of
FASEB J 1998;12:1109-23. the adherence of neutrophils to umbilical vein endothelium by
15. Kupferminc MJ, Peaceman AM, Wigton TR, Rehnberg KA, Socol human recombinant tumor necrosis factor. Proc Natl Acad Sci
ML. Tumor necrosis factor-a is elevated in plasma and amniotic USA 1985;82:8667-71.
www.ajog.org
Heat shock protein-70 and 4-hydroxy-2-nonenal adducts

in human placental villous tissue of normotensive,
preeclamptic and intrauterine growth restricted
pregnancies
Michael D. Hnat, DO,a Juliana W. Meadows, PhD,a Diane E. Brockman, MS,a
Brad Pitzer, BS,a Fiona Lyall, MD,c Leslie Myatt, PhDa,b
Departments of Obstetrics and Gynecology,a Pathology and Laboratory Medicine,b University of Cincinnati, College
of Medicine, Cincinnati, OH; Department of Obstetrics and Gynaecology, University of Glasgow, Glasgow Royal
Infirmary, Glasgow, UK c
Received for publication August 23, 2004; revised January 5, 2005; accepted January 25, 2005
KEY WORDS Objective: The purpose of this study was to compare immunohistochemical expression of heat
Placenta shock protein-70 (hsp70), a marker for oxidative stress, and 4-hydroxy-2-nonenal adducts (HNE),
Preeclampsia a marker for lipid peroxidation, in placental villous tissue of normotensive, preeclampsia, and
Intrauterine growth intrauterine growth restricted (IUGR) pregnancies.
restriction Study design: Placentas were collected and flash frozen in liquid nitrogen after delivery from
Heat shock protein normotensive pregnancies (n = 5), and pregnancies complicated by preeclampsia (n = 5), IUGR
(hsp) (n = 5), and preeclampsia plus IUGR (n = 4). Cryosections were cut and immunostained with
Hydroxynonenal polyclonal anti-hsp70 and monoclonal anti-HNE antibodies using Vectastain Elite ABC kit.
(HNE) Normal rabbit serum or mouse IgG were used as negative controls. Three independent observers,
Oxidative stress blinded to identity of tissue, examined each slide to identify cellular localization and intensity of
Lipid peroxidation the immunostaining. Western blot analysis and scanning densitometry were used to quantify and
compare the amount of hsp70 and HNE adducts present in tissue homogenates.
Results: Positive immunostaining for both antibodies was observed in cytoplasm of syncytio-
trophoblasts, extravillous trophoblasts, vascular smooth muscle, and endothelial cells for all
groups. Expression of hsp70 and HNE adducts was reported as observers’ mean stained intensity.
Overall, kappa showed good agreement between observers. Immunostaining intensity was similar
in all tissue types for each group with the exception that immunostaining was significantly more
intense in the vascular endothelium of the preeclamptic group for HNE adducts (P = .02) and
significantly less intense in the IUGR group for hsp70 (P = .013). Scanning densitometric
analysis of the Western blots showed no significant difference in total hsp70 and HNE adducts
expression in all 4 tissue groups.
Presented at the 22nd Annual Meeting of the Society for Maternal-Fetal Medicine, January 14-19, 2002, New Orleans, La.
Reprints not available from the authors.
doi:10.1016/j.ajog.2005.01.059
Hnat et al 837
Conclusion: Immunohistochemistry showed local changes for oxidative stress and lipid
peroxidation in the vascular endothelium from placentas of preeclamptic and IUGR pregnancies.
However, these changes were masked when studying tissue homogenates.
Preeclampsia remains one of the unsolved problems Canada) or monoclonal anti-HNE antibodies (Genox
in obstetrics. Evidence is accumulating that oxidative Corp, Baltimore, Md). Tissues were fixed in acetone at
stress and lipid peroxidation play a role in the patho- ÿ20(C for 10 minutes and then washed with phosphate-
genesis of preeclampsia and may account for its clinical buffered saline (PBS). For hsp70, the tissue sections were
manifestations. Articles dealing with a specific marker of blocked with 5% goat serum diluted in 0.2% Tween-20
oxidative stress or reviewing several ones are evident in in phosphate buffered saline (PBS-T) at room temper-
the literature.1-12 Maternal serum and placental levels of ature for 1 hour, whereas, 5% horse serum diluted in
lipid peroxides are noted to be increased in preeclampsia 1% saponin was used as a blocking agent for HNE
when compared to normal pregnancies.2,3,8,11,12 adduct samples. The tissues were then incubated in
Various substances are linked to oxidative stress primary antibody (anti-hsp70, 1:500; anti-HNE, 1:5) at
and lipid peroxidation. Inducible heat shock protein 4(C overnight. Tissue sections were washed and in-
70 (hsp70) has been shown to be up-regulated during cubated with respective secondary antibody (1:1000) for
oxidative stress.13-15 In addition, 4-hydroxy-2-nonenal 30 minutes at room temperature. Sections were blocked
(HNE), an aldehydic end product of lipid peroxidation, for endogenous peroxidase for 10 minutes using 3%
has been observed to be a mediator of toxic effects H2O2 in deionized water. To achieve better staining
elicited by oxidative stress.16 Both substances have been results for HNE adducts, 1% saponin was included in
shown to play a role in the pathophysiology of various all washes and antibody solutions up to and including
human diseases. Their role in preeclampsia and IUGR the secondary antibody incubation, after which it was
has not been completely defined. Therefore, we pro- excluded from all reagent mixtures. After washing, slides
posed that expression of hsp70 and HNE adducts is were incubated in the ABC complex for 30 minutes at
increased in placental villous tissue of preeclampsia and room temperature, and then stained with 3-amino-9-
IUGR pregnancies compared with those from normo- ethyl carbazole (AEC) until optimal staining was
tensive and uncomplicated pregnancies. obtained (12 minutes for hsp70 and 17 minutes for
HNE). Sections were counterstained with hematoxylin
Material and methods (Biomeda Corp, Foster City, Calif) and mounted in
PBS/glycerol (1:9). Preimmune rabbit serum and mouse
All procedures were performed under protocols ap- IgG were used as negative controls for hsp70 and
proved by the Institutional Review Board of the HNE adduct antibodies, respectively. Cellular localiza-
University of Cincinnati Medical Center and the Ethical tion and intensity of the immunostaining (none = 0,
Committee of Glasgow Royal Infirmary. Placental minimal = 1, moderate = 2, or heavy = 3) of all tissue
villous tissues (1!1!1cm) were collected at the Glas- sections were identified using 3 observers who were
gow Royal Infirmary immediately after delivery from blinded to tissue group identity.
normotensive pregnancies (n = 5) and from pregnancies For Western blot analysis, placenta samples were
complicated by either preeclampsia (n = 5), intrauterine homogenized 3 times for 5 seconds with a Tissue Tearor
growth restriction (IUGR) (n = 5) or preeclampsia plus (Glen Mills, Clifton, NJ) (initial speed of 14 and final of
IUGR (n = 4). Villous tissue was immediately flash- 22) at 4(C in homogenizing buffer consisting of 250
frozen in liquid nitrogen and stored at ÿ80(C until mmol/L sucrose and 50 mmol/L HEPES, pH 7.4,
processed. Preeclampsia was defined as a blood pressure containing the following protease inhibitors: 0.7 mg/
of 140/90 mm Hg on at least 2 occasions at least 6 hours mL pepstatin, 10 mg/mL leupeptin, 100 mmol/L 4-(2-
apart occurring after 20 weeks’ gestation and accompa- aminoethyl) benzenesulfonyl fluoride (AEBSF), 100
nied by proteinuria (O300 mg/L in a 24-hr urine mmol/L Na-p-tosyl-L-lysine-chloromethyl ketone
collection). IUGR was defined as a fetal weight less (TLCK), and 200 mmol/L sodium orthovanadate (Cal-
than the 5th percentile using standardized Scottish birth biochem, San Diego, Calif). The supernatant was
weight table. collected after centrifugation at 1000g for 10 minutes.
Cryosections (7 mm) of villous tissue were cut just Total protein quantification was performed using BCA
before staining. Serial sections were immunostained method.17 Tissue samples were diluted in 2! sample
using the Vectastain ABC Elite Kit (Vector Laborato- buffer containing 0.25 mol/L TRIS (pH 6.8), 20%
ries; Burlingame, Calif) with either polyclonal anti- glycerol, 2% SDS, 5% b-mercaptoethanol, and 0.02%
hsp70 (StressGen Biotechnologies Corp, Victoria, BC, bromophenol blue, and heated to 100(C for 5 minutes.
838 Hnat et al
Table Location and immunostaining observation intensity

of Hsp 70 and HNE adducts
Hsp 70 (mean G SD)
Group SYN EVT VSM VE
Norm 1.6 G 0.5 1.6 G 0.5 1.8 G 0.4 0.8 G 0.4
PE 1.6 G 0.5 1.6 G 0.5 1.8 G 0.4 1.0 G 0.0
IUGR 1.4 G 0.5 1.0 G 0.7 1.4 G 0.5 0.2 G 0.4*
PE and 1.5 G 0.6 1.0 G 0.8 2.3 G 0.5 1.0 G 0.0
IUGR
HNE adducts (mean G SD)
Group SYN EVT VSM VE
Norm 1.2 G 0.4 0.6 G 0.5 1.2 G 0.4 1.2 G 0.4
PE 1.8 G 0.4 1.2 G 0.4 1.8 G 0.8 2.4 G 0.5*
IUGR 1.2 G 0.4 0.6 G 0.5 1.0 G 0.0 1.6 G 0.5
Figure 1 Immunostaining for hsp70 and HNE adducts with
PE and 2.0 G 1.2 0.8 G 0.5 1.0 G 0.0 1.5 G 0.6
controls in normal placentas. (A) hsp70. (B) Preimmune rabbit
IUGR
serum. (C) HNE adducts. (D) Mouse IgG. (A, 40!, B, 100!,
and C, 200!). * P ! .05.
For SDS-PAGE gel electrophoresis, 20 mg of each syncytiotrophoblast, extravillous trophoblast, vascular

protein sample was loaded in a 12% TRIS-glycine gel smooth muscle, and endothelial cells in normal placen-
(BioRad, Hercules, Calif), and run at 40 mA/gel. Follow- tas. Figure 1 represents a sample of the immunostaining
ing electrophoresis, proteins were transferred onto a ni- of hsp70 and HNE adducts for the normotensive group.
trocellulose membrane. The membrane was blocked in All staining was cytoplasmic. No immunostaining was
5% skim milk in TBS-T for 1 hour, and then probed with apparent in the negative controls (Figure 1B and D).
rabbit anti-hsp70 polyclonal antibody (1:1000) at 4(C Expression and localization of the hsp70 and HNE
overnight. The membrane was washed and incubated with adducts are reported in the Table as the observers’ mean
donkey antirabbit secondary antibody (1:1000). ECL stained intensity. Kappa statistic showed good agree-
Western Lightning Chemiluminescence Reagent (Amer- ment. As observed in the normotensive group, expres-
sham) was applied to visualize the hsp70 band on sion of hsp70 and HNE adducts was noted in various
radiographic film. Presence of hsp70 was confirmed at cell types, with the expression all tissue types of hsp70
the 70 kD using a prestained low molecular weight marker and HNE adducts immunostaining in villous tissue
(BioRad). Band density was normalized and relative being similar between the groups except in the following
intensity was determined using scanning densitometry incidences. In the preeclampsia group, HNE adduct
(Alpha Imager 5.0, Alpha Innotech Corp, San Leandro, staining was significantly more intense in the vascular
Calif). The membranes were then stripped and reprobed endothelial cells when compared to the other groups
with anti-HNE monoclonal antibody (15 mg/mL) using (P = .02); immunostaining for hsp70 in the IUGR
sheep antimouse secondary antibody (1:1000). group was significantly less intense in the vascular
Statistical analyses include the following: Fisher exact endothelium than in the other groups (P = .031).
test was used to compare the mean intensity of observed For hsp70, Western blot analysis showed a band at 70
immunostaining between the groups. The presence of kD (Figure 2, a). For HNE, Western blot analysis
IUGR and preeclampsia was compared using logistic revealed 2 adducts at 56 and 61 kD (Figure 2, b).
regression. The degree of classification accuracy re- Analysis of variance of the spot intensity showed total
garding immunostaining intensity between observers hsp70 or HNE adducts present in villous tissue from
was evaluated by kappa statistic. The relative intensity pregnancies with preeclampsia, IUGR; preeclampsia
of the protein bands was compared using analysis of and IUGR were not significantly different than those
variance and Newman-Keuls post-hoc test. Statistical without complications (P = .91).
significance was set at P ! .05.
Comment
Results
Like other studies, our study does not conclusively
Similar patterns of hsp70 and HNE adduct immunos- define the role of oxidative stress and various markers
taining were observed in normal, IUGR, and PE in preeclampsia and IUGR.1,3-7,18-21 It is believed that
placental tissues. Positive staining for hsp70 and HNE endothelial dysfunction plays a role in preeclampsia,
adducts was observed in the following cell types: and may account for the maternal manifestation of the
Hnat et al 839
disease. Immunohistochemistry revealed a significant

increase in staining intensity of HNE adducts only in
the vascular endothelium of preeclamptic placentas
when compared to the other groups, supporting a role
of increased lipid peroxidation and endothelial involve-
ment in the underlying pathogenesis of preeclampsia.
However, the decrease in staining intensity of hsp70 seen
in placental dysfunction in IUGR placentas, represent-
ing a decrease in oxidative stress, would not support
a role for oxidative stress in the placental manifestations
of IUGR.
Western blot analysis did not show a difference in
hsp70 or HNE adduct present in total tissue homoge-
nates the groups. Immunostaining intensity of hsp70
and HNE adducts in the various groups reflects local
changes in the tissue, whereas Western blot analysis
reflects total protein changes in the tissue homogenate,
which would obliterate subtle cellular changes. This can
explain why the immunohistochemical staining and
Western blot results did not concur. The local increases
in protein expression represented as immunohistochem-
ical changes may not have been significant enough to be
reflected as changes in the total amounts of protein or Figure 2 Western blot analysis for hsp70 and HNE adducts.
adduct seen in the homogenate used for the Western (a) hsp70 with bands at 70 kD. (b) HNE adduct bands above
and below 55 kD. (A, normotensive, B, IUGR, C, preeclamp-
blot analysis.
sia plus IUGR, D, preeclampsia).
For the Western blot analysis, a 70 kD band was
observed in all samples probed with anti-hsp70, which is
consistent with other literature reports.14,15 Two bands,
at 56 and 61 kD, were noted on the HNE Western blot, strated in this study. Various explanations may account
again consistent with other investigators’ findings.22-24 for this. First, HNE adducts may not be a relative
The monoclonal antibody used for 4-hydroxy-2-nonenal marker for lipid peroxidation. Placental HNE produc-
immunostaining recognizes histidine and sulfhydryl tion, breakdown, and clearance may be rapid. Second,
adducts of 4-hydroxy-2-nonenal modified proteins.23 technical or system error could have resulted in the
In some cases, 10 or more proteins have been detected absence of significant differences between the groups. In
by Western blot analysis using the 4-hydroxy-2-nonenal addition, techniques for immunohistochemistry along
monoclonal antibody.23-25 with our antisera may not be sensitive enough to show
Abnormal placentation and inadequate maternal a difference between the groups. Tissue collection and
vascular response may also complicate pregnancies preparation may have confounded the findings for both
with preeclampsia or IUGR. Because changes were hsp70 and HNE. For example, hypoxia that occurs in
only seen in the vascular endothelium in this study the uterus after separation of the placenta may induce
and not generally throughout the placental tissue, expression of hsp70 and HNE adducts. However, this
a selective effect on the vascular reactivity of the was not observed in studies performed in our lab with
placenta is a possibility. However, we cannot prove hsp70. We have observed immunostaining intensity for
a cause and effect relationship that HNE adducts reflect hsp70 was consistent among random sampling in the
changes in vascular reactivity. This can only be con- same placenta from pathologic pregnancies similar to
firmed by placental perfusion studies. In addition, these groups, and was not affected by increasing time
abnormal placentation and vasculature changes are from delivery to collection of placenta.
mostly seen in placental bed biopsies in preeclampsia. Our study, consistent with other studies previously
Unfortunately, placental bed biopsies were not obtained mentioned, indicates that hsp70 and HNE adducts do
in this study. This is a limitation of this study. Further not have a vital role in preeclampsia with or without
investigation using placenta bed biopsies for hsp70 and IUGR, confirming that these markers should not be
HNE adduct concentration are needed to determine used in the analysis of oxidative stress in placentas from
their role in preeclampsia. pregnancies complicated by these pathologic processes.
The essential role of lipid peroxidation as the un- However, before completely discarding these 2 markers,
derlying pathogenesis of abnormal placentation leading further studies should be undertaken on placental bed
to preeclampsia and IUGR was not completely demon- biopsies. In addition, other markers may be identified
840 Hnat et al
in the future that aid in the diagnosis of preeclampsia 12. Madazli R, Benian A, Aydin S, Uzun H, Tolun N. The plasma and
with or without IUGR. In closing, preeclampsia is a placental levels of malondialdehyde, glutathione and superoxide
dismutase in pre-eclampsia. J Obstet Gynaecol 2002;22:477-80.
complicated pregnancy-related disease, and has eluded 13. Freeman ML, Borrelli MJ, Meredith MJ, Lepock JR. On the path
researchers as to its origin and cause, and will contin- to the heat shock response: destabilization and formation of
ually do so until we have better tools to aid in the partially folded protein intermediates, a consequence of protein
understanding of the disease. thiol modification. Free Radic Biol Med 1999;26:737-45.
14. Macario AJL. Heat-shock proteins and molecular chaperones:
implications for pathogenesis, diagnostics, and therapeutics. Int J
Clin Lab Res 1995;25:59-70.
References 15. Su CY, Chong KY, Owen OE, Dillmann WH, Chang C, Lai CC.
Constitutive and inducible hsp70s are involved in oxidative
1. Hubel CA. Oxidative stress in the pathogenesis of preeclampsia. resistance evoked by heat shock or ethanol. J Mol Cell Cardiol
PSEMB 1999;222:222-35. 1998;30:587-98.
2. Hubel CA, Kagan VE, Disin ER, McLaughlin MK, Roberts JM. 16. Parola M, Bellomo G, Robino G, Giuseppina B, Dianzani MU.
Increased ascorbate radical formation and ascorbate depletion in 4-Hydroxynonenal as a biological signal: Molecular basis and
plasma from women with preeclampsia: implications for oxidative pathophysiological implications. Antiox Redox Signal 1999;
stress. Free Radic Biol Med 1997;23:597-609. 1:255-84.
3. Shibata E, Ejima K, Nanri H, Toki N, Koyama C, Ikeda M, et al. 17. Brown R, Jarvis K, Hyland K. Protein measurement using
Enhanced protein levels of protein thiol/disulphide oxidoreduc- bicinchoninic acid: elimination of interfering substances. Anal
tases in placentae from pre-eclamptic subjects. Placenta 2001; Biochem 1989;180:136-9.
22:566-72. 18. Regan CL, Levine RJ, Barid DD, Ewell MG, Martz KL, Sibai
4. Taylor RN, deGroot CJM, Cho YK, Lim KH. Circulating factors BM, et al. No evidence for lipid peroxidation in severe pre-
as markers and mediators of endothelial cell dysfunction in eclampsia. Am J Obstet Gynecol 2001;185:572-8.
preeclampsia. Semin Reprod Endocrin 1998;16:17-31. 19. Roggensack AM, Zhang Y, Davidge ST. Evidence for peroxyni-
5. Walsh SW. Maternal-placental interactions of oxidative stress trite formation in the vasculature of women with preeclampsia.
and antioxidants in preeclampsia. Semin Reprod Endocrin 1998; Hypertension 1999;33:83-9.
16:93-104. 20. Santoso DIS, Rogers P, Wallace EM, Manuelpillai U, Walker D,
6. Poranen AK, Ekblad U, Uotila P, Ahotupa M. Lipid peroxidation Subakir SB. Localization of indoleamine 2,3-dioxygenase and
and antioxidants in normal and pre-eclamptic pregnancies. 4-hydroxynonenal in normal and pre-eclamptic placentae.
Placenta 1996;17:401-5. Placenta 2002;23:373-9.
7. Takagi Y, Nikaido T, Toki T, Kita N, Kanai M, Ashida T, et al. 21. Hung TH, Skepper JN, Burton GJ. In vitro ischemia-reperfusion
Levels of oxidative stress and redox-related molecules in the injury in term human placenta as a model for oxidative stress in
placenta in preeclampsia and fetal growth restriction. Virchows pathological pregnancies. Am J Pathol 2001;159:1031-43.
Arch 2003;22:1-14. 22. Uchida K, Szweda LI, Chae HZ, Stadtman ER. Immunochemical
8. Wang Y, Walsh SW, Kay HH. Placental lipid peroxides and detection of 4-hydroxynonenal protein adducts in oxidized hep-
thromboxane are increased and prostacyclin is decreased in women atocytes. Proc Natl Acad Sci 1993;90:8742-6.
with preeclampsia. Am J Obstet Gynecol 1992;167:946-9. 23. Hattori Y, Nishigori C, Tanaka T, Uchida K, Nikaido O, Osawa
9. Myatt L, Kossenjan W, Sahay R, Eis A, Brockman D. Oxidative T, et al. 8-hydroxy-2’-deoxyguanosine is increased in epidermal
stress causes vascular dysfunction in the placenta. J Matern Fetal cells of hairless mice after chronic ultraviolet B exposure. J Invest
Med 2000;9:79-82. Dermatol 1996;107:733-7.
10. Orhan H, Önderoglu L, Yücel A, Sahin G. Circulating biomarkers 24. Kondo S, Toyokuni S, Iwasa Y, Tanaka T, Onodera H, Hiai H,
of oxidative stress in complicated pregnancies. Arch Gynecol et al. Persistent oxidative stress in human colorectal carcinoma,
Obstet 2003;267:189-95. but not in adenoma. Free Radic Biol Med 1999;27:401-10.
11. Ilhan N, Ilhan N, Simsek M. The changes of trace elements, 25. Moreau R, Heath SHD, Doneanu CE, Lindsay JG, Hagen TM.
malondialdehyde levels and superoxide dismutase activities in Age-related increase in 4-hydroxynonenal adduction to rat heart
pregnancy with or without preeclampsia. Clin Biochem 2002; alpha-ketoglutarate dehydrogenase does not cause loss of its
35:393-7. catalytic activity. Antioxid Redox Signal 2003;5:515-27.
www.ajog.org
Cloning and cellular expression of aquaporin 9

in ovine fetal membranes
Shengbiao Wang,a,b,* Jiexiong Chen,a Bing Huang,c Michael G. Rossa
Department of Obstetrics and Gynecology, Los Angeles Biomedical Research Institute at Harbor-UCLA
Medical Center, Torrance, CAa; Quest Diagnostics Nichols Institute, San Juan Capistrano, CAb; and
Genzyme Genetics, Orange, CA c
Received for publication August 30, 2004; revised December 30, 2004; accepted January 11, 2005
KEY WORDS Objective: Amniotic fluid (AF) absorption across fetal membranes is essential for AF volume
Aquaporin 9 homeostasis, balancing fetal swallowing, urine flow, and lung liquid production. In sheep, AF is
Amniotic fluid absorbed primarily across the amniotic membrane into fetal vasculature situated between the
regulation amnion and chorion. Aquaporins (AQPs) are cell membrane proteins that serve as water
Intramembranous channels. Recent studies have demonstrated the expression of AQP 1, 3, 8, and 9 in human
pathway chorioamniotic membranes and placenta. As AF dynamics continued to be explored primarily in
Amnion the ovine model, we sought to clone and characterize the expression of ovine AQP9 in fetal
Sheep membranes.
Methods: Ovine AQP9 gene was cloned with the use of homology reverse transcriptase-
polymerase chain reaction (RT-PCR). RT-PCR and Northern analysis were used to determine
AQP9 gene expression, and immunohistochemistry (IHC) used to localize AQP9 protein
expression in ovine fetal membranes.
Results: A 2085-base pair (bp) full-length complementary DNA (cDNA) sequence of ovine
AQP9 was cloned. The ovine AQP9 cDNA is 86%, 82%, and 82%, and the predicted amino acid
sequence (295 amino acids) is 77%, 71%, and 69% identical to human, rat, and mouse AQP9,
respectively. RT-PCR and Northern analysis detected AQP9 messenger RNA expression in ovine
amnion and allantois, but not in placenta, chorion, or umbilical cord. Immunohistochemistry
localized AQP9 protein in epithelia of amnion and allantois.
Conclusion: The presence of significant AQP9 messenger RNA and protein expression in ovine
fetal membranes suggests that AQP9 may be a major water channel for intramembranous AF
resorption in sheep. The cloning of ovine AQP9 and the demonstration of AQP9 expression in
amnion and allantois significantly enhances our understanding of ovine AF regulation and offers
the potential for therapeutic approaches for the treatment of oligohydramnios and polyhy-
dramnios.
Supported by National Institutes of Health grants R03 HD 044482, R01 HL 40899, and the March of Dimes Birth Defects Foundation.
* Reprint requests: Shengbiao Wang, MD, Department of Obstetrics and Gynecology, Harbor-University of California- Los Angeles Medical
Center, 1000 W Carson St, Box 3, Torrance, CA 90502.
E-mail: sheng@labiomed.org
doi:10.1016/j.ajog.2005.01.019
842 Wang et al
Aquaporins (AQPs) are cell membrane water channel from AF has been postulated as a regulatory factor for
proteins that have 6 transmembrane domains and 2 AF volume changes with gestation.19 Therefore, AQP9
asparagine-proline-alanine (NPA) motifs.1 As demon- may be a critical water channel in fetal membrane fluid
strated by studies of AQP1, AQP proteins organize in absorption.
membranes as tetramers; however, each monomer forms To date, our knowledge of intramembranous AF
a hydrophilic pore in its center and functions indepen- regulation has been primarily obtained from the in vivo
dently as water channel.1 To date, at least 11 types of study of sheep. In sheep, though not humans, an
mammalian AQPs have been identified with different extensive fetal vasculature network is situated between
AQP types expressed in different tissues.2 In vitro the amnion and chorion, potentially facilitating AF
experimental models have demonstrated that AQPs absorption across the amniotic membrane. In the
substantially increase water permeability across the current study, we sought to clone the full-length ovine
membranes in which they are expressed.3 Select AQPs AQP9 complementary DNA (cDNA) and examine the
(AQP 3, 7, and 9) are also permeable to neutral solutes location of AQP9 messenger RNA (mRNA) and protein
such as glycerol and urea, in addition to water.4,5 expression in ovine fetal membranes and placenta.
Recent advances have provided significant insight Cloning and characterization of AQP9 expression in
into the role of AQPs in the cause and pathophysiology ovine fetal membrane will establish an ideal animal
of human diseases. For example, AQP2 mutation causes model for the study of AQP9 in AF regulation and
nephrogenic diabetes insipidus,6 whereas mutation of further explore the possibility of AQP9 as a therapeutic
AQP0, the AQP expressed in epithelia of the lens, causes target for the treatment of oligohydramnios and poly-
familial cataracts.7 More recent studies report that hydramnios.
defective cellular trafficking of AQP5 in lacrimal8 and
salivary glands9 explains the dry eyes and mouth in
Sjogren’s syndrome. Studies in animal models have Material and methods
provided critical information regarding the essential
functions of AQPs. AQP1 knockout mice demonstrate Ovine fetal membranes and other tissues
defective renal concentration ability,10 and transgenic
mice lacking AQP3 develop diabetes insipidus,11 Animal study protocol was approved by the Animal
whereas AQP4 knockout mice exhibit defective gastric Care and Use Committee at the Los Angeles Biomedical
acid secretion,12 colon water transport,13 and altered Research Institute at Harbor-UCLA Medical Center.
skeletal muscle water permeability.14 These recent trans- Five ewes at 135 G 2 days gestation (term = 145-150
genic animal studies further highlight the important days) were euthanized and amnion, chorion, allantois,
physiologic role of aquaporins in transmembranous and placenta were dissected and quick frozen in liquid
fluid transport in various biologic systems. nitrogen. Tissues were stored at 70(C until further
Amniotic fluid (AF) volume is critical for normal fetal analyses. Liver and kidney were obtained from one
movement, growth, and development. Abnormalities of 1-month-old neonatal sheep as control tissues for the
AF volume (ie, oligohydramnios, polyhydramnios) are characterization of ovine AQP9 gene expression. Total
common in human pregnancy. Polyhydramnios occurs in RNA was isolated from these tissues with the use of
0.2% to 1.6% of pregnancies, whereas oligohydramnios Trizol reagent (Invitrogen, Carlsbad, Calif) for reverse
occurs in 8% to 38% of all pregnancies. Both poly- and transcriptase polymerase chain reaction (RT-PCR) and
oligohydramnios are associated with significant perinatal Northern analysis.
morbidity and mortality. AF is produced by fetal urine Homology RT-PCR cloning of ovine AQP9 cDNA
and lung liquid, and reabsorbed in part by fetal swallow-
ing. The intramembranous pathway (ie, AF absorption To clone ovine AQP9, we used the homology RT-PCR
across amniotic membranes) has been recognized as with primers derived from conserved sequences among
a critical regulatory path for AF volume homeostasis,15 human, mouse, and rat AQP9. The sequences of the sense
as it balances fetal fluid production and resorption, and antisense primers were ‘‘ATCTTGATTGTCCTTG-
achieving volume homeostasis. However, the molecular GATGT’’ and ‘‘ACAGGAATCCACCAGAAGTT,’’ re-
and cellular mechanisms for water and solute absorption spectively. RT-PCR was performed with RNA isolated
across the fetal membranes remain to be elucidated. from ovine liver, as liver has been shown to express most
Recent studies have demonstrated the expression of abundant AQP9 in other species.20 RT-PCR methods
AQP 1, 3, and 8 in human chorioamniotic membranes were as detailed later, with the exception that only these 2
and placenta.16,17 Our recent study has demonstrated that primers (10 pmols/reaction) were included. RT-PCR
among 11 AQPs studied, AQP9 is expressed strongest in fragments were gel purified by using Qiagen gel purifica-
fetal membranes.18 In addition to water, AQP9 is also tion kit (Qiagen Inc, Valencia, Calif) and the fragments
permeable to neutral solutes, including urea that are were sequenced by dye terminator cycle sequencing with
present in AF. Notably, the regulation of urea absorption both forward and reversed primers in ABI 3100 DNA
Wang et al 843
Analyzer (Applied Biosystems, Inc, Foster City, Calif). Immunohistochemical staining

Sequence comparisons were performed with the NCBI
BlastN program. Polyclonal primary antibody for AQP9 was purchased
from Alpha Diagnostics (San Antonio, Tex). The AQP9
antibody was raised in rabbits immunized with a poly-
Rapid amplification of cDNA end (RACE) peptide of 16 amino acids derided from carboxyl
On cloning of partial ovine AQP9 cDNA by homology terminal of rat AQP9 that has 60% homology to ovine
RT-PCR, we cloned full-length ovine AQP9 cDNA by AQP9. An immunohistochemical staining kit (Vector
RACE procedures using the SMART RACE cDNA Laboratories, Burlingame, Calif) was used as previously
Amplification Kit (Clontech, Palo Alto, Calif) with minor described.18 After incubation of the primary antibody,
modifications of the manufacturer’s suggested protocol. the tissue was washed and incubated with the biotin-
To increase the success rate for RACE, total RNA labeled secondary antibody. This was followed by
isolated from sheep liver was used as template to syn- a wash and then visualized with an avidin-biotin
thesize the first strand 50 and 30 - RACE cDNA. The se- complex immunoperoxidase system (Vector Laborato-
quences of the gene specific primers for 50 and 30 - RACE ries) by using 0.03% diamino-benzidine as the chroma-
PCR were ‘‘GAAACAGCTGGGTTGATGTGG’’ and gen and hematoxylin as the counterstain. Protein
‘‘AGAGCCTGTTGTCATTGGC,’’ respectively. The expression signal was studied under light microscope.
RACE PCR fragments were gel purified by using Qiagen Negative controls were performed in parallel by using
gel purification kit (Qiagen Inc), and the fragments were the primary antibody that was preabsorbed with the
sequenced by dye terminator cycle sequencing in ABI polypeptide used to immunize the rabbit for the pro-
DNA sequencer 3100 (Applied Biosystems, Inc). Nucle- duction of the anti-AQP9 antibody.
otide and deduced amino acid sequences comparison were
performed with NCBI BlastN and BlastP program, Results
respectively.
Cloning of ovine AQP9
RT-PCR We cloned a full-length 2085-base ovine AQP9 cDNA.
Primers for AQP9 were ‘‘GGAGGGGTCATCACTAT- The ovine AQP9 cDNA encodes for a protein with 295
CAAT’’ and ‘‘ACAGGAATCCACCAGAAGTT,’’ and amino acids. The nucleotide and the deduced amino acid
‘‘ATCGTGATGGACTCCGGTGAC’’ and ‘‘GCTGA- sequences are shown in Figure 1. As expected, the
TCCACATCTGCTGGA’’ for b-actin, respectively. deduced protein contains 2 NPA motifs, a characteristic
These primers are located in different exons, thus able of AQP protein. We performed sequence similarity
to discriminate the PCR products from genomic DNA analysis between this gene with known AQP9 cDNA of
versus cDNA. These primers’ sequences are conserved other species. The nucleotide sequence of this fragment is
in human, mouse, rat, and sheep genes. Thus, this mul- 86% identical to human and 82% identical to mouse and
tiplex RT-PCR can be used to study AQP9 gene rat AQP9 cDNA (data not shown). The predicted amino
expression in these species. The RT-PCR fragment size acid sequence of this cloned ovine gene is 78%, 73%, and
of sheep b-actin mRNA and AQP9 is 831 base pair (bp) 72% homologous to human, mouse, and rat AQP9
and 581 bp, respectively. RT-PCR was performed as protein (Figure 2), respectively. The extensive homology
previously described.18 to AQP9 of other species is consistent with our conclu-
sion that this 2085-base cDNA is ovine AQP9.
Northern analysis RT-PCR
For Northern analysis, biotin-labeled ovine AQP9 To determine AQP9 gene expression in ovine fetal
cDNA probe was prepared with the use of gel-purified membranes, we carried out RT-PCR analysis of RNA
RT-PCR-amplified ovine AQP9 fragment as a template isolated from tissues of 135-day-old ovine fetuses. Liver
for in vitro random-primed synthesis by Bioprime DNA and kidney RNA from 1-month-old lamb were also
Labeling Systems (Invitrogen). Fifteen micrograms of studied as positive and negative controls, respectively.
total RNA per sample were subjected to Northern As expected, RT-PCR detected AQP9 in liver. In all
analysis according to the standard protocol.18 Signals fetal membrane tissues of the 5 fetuses studied, AQP9
were analyzed with Bio-Rad MultiImager and Quantity mRNA was detected in amnion and allantois, but not in
One software (Bio-Rad Laboratories, Hercules, Calif). chorion, placenta, umbilical cord, or kidney (Figure 3).
Membranes were stripped of AQP9 probes in solution Northern analysis
containing 0.02 ! SSC and 0.01% SDS at 95(C for 15
minutes and rehybridized with biotin labeled b-actin We further studied the AQP9 mRNA expression
probe to normalize AQP9 mRNA level. in ovine fetal membranes using Northern analysis.
844 Wang et al
Figure 1 Nucleotide sequence and deduced amino acid sequence of the cloned ovine AQP9. A polyadenylation signal consensus
sequence is underlined. The conserved NPA motifs are in bold face.
Wang et al 845
Figure 2 Comparison of amino acid sequences of sheep, human, mouse, and rat AQP9.
Consistent with the RT-PCR results, AQP9 mRNA specifically the AQP9 protein. Consistent with the
expression was detected in amnion, allantois, and liver, Northern analysis, the AQP9 protein expression level
but not in chorion, placenta, umbilical cord, or kidney was similar between amnion and allantois, and consis-
(Figure 3). As expected, AQP9 mRNA expression was tent among all 5 fetal tissues studied.
much stronger in liver, with lesser though similar AQP9
gene expression levels detected in amnion and allantois.
There was no alternative spliced AQP9 transcript Comment
detected in the Northern analysis. The AQP9 mRNA
expression in amnion and allantois detected by North- AF absorption across chorioamniotic membranes and/
ern analysis was consistent among the 5 fetuses studied. or the placenta surface plays a critical role in AF volume
regulation. In sheep, the chorioamniotic membranes
Immunohistochemistry are highly permeable to water as well as urea.21 The
microscopic structure of ovine amnion and allantois
To determine the cellular expression of AQP9 protein, reveal typical membrane characteristics of transporting
immunohistochemical staining of ovine amnion and epithelia.22 In humans, water transport across the
allantois with anti-AQP9 antibody were performed. As chorionic plate occurs at a much higher rate than
shown in Figure 4, immunohistochemical staining with glucose.23 However, the molecular mechanism(s) for
anti-AQP9 antibody detected positive signals in both such elevated fetal membrane permeability to water is
apical and basal membranes of epithelial cells of amnion poorly understood. Over the past decade, it has been
and allantois. In addition, slight intracellular staining in increasingly recognized that aquaporins are important
amniotic and allantoic epithelia was observed. To the water channels that permit high-cellular membrane
contrary, negative controls performed with AQP9 anti- permeability to water. AQP9 is highly permeable to
body preabsorbed with the AQP9 polypeptide showed water and urea. We previously reported that among 11
no staining, demonstrating that the antibody recognizes AQPs studied, the AQP9 is the 1 with highest level of
846 Wang et al
days) ovine fetal membranes. In this study, the RT-PCR

and Northern analysis both detected AQP9 mRNA
expression in near-term ovine amnion and allantois.
We further determined that AQP9 protein is expressed
in epithelial cells of amnion and allantois by immuno-
histochemistry. In our most recent study, AQP9 gene
expression was detected in human amnion, chorion, and
placenta.18 The findings of only amnion and allantois
AQP9 expression in sheep may represent the species
variation. In humans, the amniotic membrane is avas-
cular with no significant vascularity between the amnion
and chorion. Thus, AF water and solutes must cross
both amnion and chorion (or chorionic plate of pla-
centa) to be absorbed into the circulation. Conversely, in
sheep there is an extensive fetal microvessel network
underneath the amniotic membrane.29 Therefore, water
and solutes may be absorbed into fetal vaculature after
crossing the ovine amnion epithelial cells. In sheep, fetal
urine is excreted into both amniotic and allantoic sacs.
As there is no other fetal route of fluid resorption, AQP
Figure 3 RT-PCR (A) and Northern analysis (B) analysis of channels in the allantoic membrane likely contribute to
AQP9 gene expression in ovine amnion, chorion, allantois,
resorption of water and urea from the allantois. The
placenta, and umbilical cord.
observations that AQP9 is expressed in ovine and
human amnion, ovine allantois, and human chorion
expression in human fetal membranes and placenta, and placenta, are consistent with the notion that AQP9
suggesting AQP9 may be an important water channel is important for both AF and allantoic fluid water and
mediating intramembranous human AF resorption.18 solute absorption across fetal membranes.
In the current study, we aimed at cloning ovine AQP9 Despite the critical importance of intramembranous
cDNA and characterizing its expression in fetal mem- pathway in AF volume homeostasis, there have been
branes. Because human, mouse, and rat AQP9 cDNA limited reports delineating the molecular mechanism for
have been cloned and found to be highly conserved water resorption through fetal membranes. Previously,
among these species, we first took the homology RT- AQP3 expression was demonstrated in cytotrophoblasts
PCR approach to clone a partial ovine AQP9 cDNA of ovine chorion and placenta, as well as fibroblasts of
sequence. We then used 50 and 30 RACE strategies to amnion and allantois.26 These authors also reported
attain and sequence the 50 and 30 end of the ovine AQP9 AQP1 in vascular endothelia of chorion and placenta.
cDNA. By using these highly efficient methods, we have Whereas AQP1 expression was detected in human
cloned a full-length 2085-base ovine AQP9 cDNA that amnion and chorion, AQP3 was only found by RT-
is predicted to encode for a protein of 295 amino acids. PCR and Western blotting analysis of chorion, but not
The predicted ovine AQP9 protein contains 2 NPA in amnion,17 suggesting lower levels of AQP3 expres-
motifs, a characteristic of AQP protein. Furthermore, sion. Our recent study demonstrated AQP9 expression
the nucleotide sequence and the predicted amino acid in human chorioamniotic membranes.18 The consistency
sequences of the cloned ovine sequence are highly of AQP9 gene expression in both human and ovine fetal
homologous to AQP9 of other species. On the other membranes suggests that AQP9 may be a critical water
hand, the nucleotide sequence of the cloned fragment is channel regulating AF resorption through the intra-
only 30% to 50% identical to the 6 known ovine AQPs, membranous pathway.
AQP 1 through 5 and 824-28 (data not shown). There- Previous reports have suggested that AF volume
fore, the 2085-base cDNA we cloned is indeed the ovine changes during gestation may be dependent on alter-
AQP9. ations in solute as well as water permeability. The
AF volume changes dramatically during human permeability of ovine amnion to urea decreases mark-
pregnancy. Average AF volume increases progressively edly with advancing gestation, correlating inversely with
from approximately 20 mL at 10 weeks to a peak the increased AF urea concentration.21 As AQP9
volume of 750 to 1000 mL at 30 to 37 weeks’ gestation functions as both a urea and water channel, one may
and decrease sharply postterm. Recognizing that the AF postulate that amnion AQP9 expression is reduced near
volume is relatively stable during the near-term period term, resulting increased AF urea concentration;
of human pregnancy, we selected to study AQP9 whereas increased expression of alternative AQPs may
expression at approximately 0.9 gestation (135 G 2 enhanced intramembranous water flow and contribute
Wang et al 847
Figure 4 Immunohistochemical analysis of AQP9 protein expression with anti-AQP9 antibody (left panel) and negative control
(right panel) in ovine amnion and allantois. Brown stains are positive signals. Tissues were counter stained with hematoxylin
(purple). (Original magnifications: !200.)
to reduced AF volume postterm. Further ontogenic Together with the demonstrated AQP9 function as
study of AQP9 expression in ovine fetal membranes and a water and neutral solute (including urea) channel,
correlation of the AQP9 level with the AF volume and and our prior report of the high level of AQP9
composition changes throughout gestation will provide expression in human fetal membranes, this study pro-
insight to this hypothesis. vides further evidence supporting our hypothesis
In summary, we have cloned the full-length cDNA that AQP9 is a major water and solutes channel
sequence of ovine AQP9 and demonstrated its mRNA mediating intramembranous AF fluid absorption. Clon-
and protein expression in ovine amnion and allantois. ing and identification of AQP9 expression in ovine fetal
848 Wang et al
membranes is a critical step toward establishing an 12. Wang KS, Komar AR, Ma T, Filiz F, McLeroy J, Hoda K, et al.
animal model, allowing for future in vivo studies of the Gastric acid secretion in aquaporin-4 knockout mice. Am J Physiol
Gastrointest Liver Physiol 2000;279:G448-53.
role of AQP9 in intramembranous AF transport. 13. Wang KS, Ma T, Filiz F, Verkman AS, Bastidas JA. Colon water
Although beyond the score of the current article, further transport in transgenic mice lacking aquaporin-4 water channels.
studies are needed to (1) address the physiologic Am J Physiol Gastrointest Liver Physiol 2000;279:G463-70.
function of AQP9 in fetal membrane fluid transfer by 14. Yang B, Verbavatz JM, Song Y, Vetrivel L, Manley G, Kao WM,
determining whether amniotic fluid reduction and am- et al. Skeletal muscle function and water permeability in aqua-
porin-4 deficient mice. Am J Physiol Cell Physiol 2000;278:
niotic fluid volume loading will alter fetal membrane C1108-15.
AQP9 expression in the sheep, (2) explore whether small 15. Gilbert WM, Brace RA. The missing link in amniotic fluid
molecule and endocrine factors (eg, vasopressin) can volume regulation: intramembranous flow. Obstet Gynecol 1989;
alter AQP9 expression in fetal membrane and change 74:748-54.
amniotic fluid dynamic, and (3) investigate the possibil- 16. Wang S, Kallichanda N, Song W, Ramierez B, Ross MG.
Expression of aquaporin 8 in human placenta and chorioamniotic
ity of altering AQP9 expression as a therapeutic ap- membranes: evidence of molecular mechanism for intramembra-
proach to oligohydramnios and polyhydramnios. nous amniotic fluid resorption. Am J Obstet Gynecol 2001;
185:1226-31.
References 17. Mann SE, Ricke EA, Yang BA, Verkman AS, Taylor RN.
Expression and localization of aquaporin 1 and 3 in human fetal
1. Knepper MA, Wade JB, Terris J, Ecelbarger CA, Marples D, membranes. Am J Obstet Gynecol 2002;187:902-7.
Mandon B, et al. Renal aquaporins. Kidney Int 1996;49:1712-7. 18. Wang S, Chen J, Beall M, Zhou W, Ross MG. Expression of
2. Agre P, Kozono D. Aquaporin water channels: molecular mech- aquaporin 9 in human chorioamniotic membranes and placenta.
anisms for human diseases. FEBS Lett 2003;555:72-8. Am J Obstet Gynecol 2004;191:2160-7.
3. Zeidel ML, Nielsen S, Smith BL, Ambudkar SV, Maunsbach AB, 19. Lingwood B, Wintour EM. Amniotic fluid volume and in vivo
Agre P. Ultrastructure, pharmacologic inhibition, and transport permeability of ovine fetal membranes. Obstet Gynecol 1984;
selectivity of aquaporin channel-forming integral protein in 64:368-72.
proteoliposomes. Biochemistry 1994;33:1606-15. 20. Tsukaguchi H, Shayakul C, Berger UV, Mackenzie B, Devidas S,
4. Ishibashi K, Sasaki S, Fushimi K, Uchida S, Kuwahara M, Saito Guggino WB, et al. Molecular characterization of a broad
H, et al. Molecular cloning and expression of a member of the selectivity neutral solute channel. J Biol Chem 1998;273:24737-43.
aquaporin family with permeability to glycerol and urea in 21. Lingwood BE, Wintour EM. Permeability of ovine amnion and
addition to water expressed at the basolateral membrane of kidney amniochorion on urea and water. Obstet Gynecol 1983;61:227-32.
collecting duct cells [see comments]. Proc Natl Acad Sci U S A 22. Shandley L, Alcorn D, Wintour EM. Ovine amniotic and allantoic
1994;91:6269-73. epithelia across gestation. Anat Rec 1997;248:542-53.
5. Echevarria M, Windhager EE, Tate SS, Frindt G. Cloning and 23. Schroder HJ, Dehne K, Andreas T, Rago S, Rybakowski C.
expression of AQP3, a water channel from the medullary collecting Diffusive transfer of water and glucose across the chorionic plate
duct of rat kidney. Proc Natl Acad Sci U S A 1994;91:10997-1001. of the isolated human term placenta. Placenta 1999;20:59-63.
6. Deen PM, Verdijk MA, Knoers NV, Wieringa B, Monnens LA, 24. Wintour EM, Earnest L, Alcorn D, Butkus A, Shandley L,
van Os CH, et al. Requirement of human renal water channel Jeyaseelan K. Ovine AQP1: cDNA cloning, ontogeny, and control
aquaporin-2 for vasopressin-dependent concentration of urine. of renal gene expression. Pediatr Nephrol 1998;12:545-53.
Science 1994;264:92-5. 25. Butkus A, Earnest L, Jeyaseelan K, Moritz K, Johnston H, Tenis
7. Berry V, Francis P, Kaushal S, Moore A, Bhattacharya S. N, et al. Ovine aquaporin-2: cDNA cloning, ontogeny and control
Missense mutations in MIP underlie autosomal dominant ‘poly- of renal gene expression. Pediatr Nephrol 1999;13:379-90.
morphic’ and lamellar cataracts linked to 12q. Nat Genet 26. Johnston H, Koukoulas I, Jeyaseelan K, Armugam A, Earnest L,
2000;25:15-7. Baird R, et al. Ontogeny of aquaporins 1 and 3 in ovine placenta
8. Tsubota K, Hirai S, King LS, Agre P, Ishida N. Defective cellular and fetal membranes. Placenta 2000;21:88-99.
trafficking of lacrimal gland aquaporin-5 in Sjogren’s syndrome. 27. Liu H, Hooper SB, Armugam A, Dawson N, Ferraro T,
Lancet 2001;357:688-9. Jeyaseelan K, et al. Aquaporin gene expression and regulation in
9. Steinfeld S, Cogan E, King LS, Agre P, Kiss R, Delporte C. the ovine fetal lung. J Physiol 2003;551:503-14.
Abnormal distribution of aquaporin-5 water channel protein in 28. Wang S, Su Y, Fang R, Ramierez B, Ross MG. Cloning and
salivary glands from Sjogren’s syndrome patients. Lab Invest expression of aquaporin 8 water channel in ovine and human
2001;81:143-8. chorioamniotic membranes: molecular mechanism of the intra-
10. Vallon V, Verkman AS, Schnermann J. Luminal hypotonicity in membranous pathway for amniotic fluid resorption. J Soc Gynecol
proximal tubules of aquaporin-1-knockout mice. Am J Physiol Investig 2000;7:183A.
Renal Physiol 2000;278:F1030-3. 29. Brace RA, Gilbert WM, Thornburg KL. Vascularization of the
11. Ma T, Song Y, Yang B, Gillespie A, Carlson EJ, Epstein CJ, et al. ovine amnion and chorion: a morphometric characterization of the
Nephrogenic diabetes insipidus in mice lacking aquaporin-3 water surface area of the intramembranous pathway. Am J Obstet
channels. Proc Natl Acad Sci U S A 2000;97:4386-91. Gynecol 1992;167:1747-55.
www.ajog.org
Limited differentiation to neurons and astroglia from

neural stem cells in the cortex and striatum after
ischemia/hypoxia in the neonatal rat brain
Tomoaki Ikeda, MD,a,* Masanori Iwai, MD,b Takeshi Hayashi, MD,b Isao Nagano, MD,b
Mikio Shogi, MD,b Tsuyomu Ikenoue, MD,a Koji Abe, MDb
Department of Obstetrics and Gynecology, Miyazaki Medical College, University of Miyazaki, Miyazaki, Japana;
Department of Neurology, Graduate School of Medicine and Dentistry, Okayama University, Okayama, Japanb
KEY WORDS Objective: We examined whether progenitor neural stem cells can differentiate successfully into
Hypoxic-ischemic mature neurons and astrocytes in a rat model of neonatal hypoxic-ischemic encephalopathy.
encephalopathy Study design: Seven-day-old Wistar rats were subjected to hypoxic-ischemic stress. At days 5 to 7
Neonatal rat after hypoxic-ischemic stress, 5-bromodeoxyuridine (an early marker of cell proliferation) was
Neural stem cell injected, and the brains were retrieved at 14, 28, and 42 days after hypoxic-ischemic stress.
Differentiation Immunohistochemical and immunofluorescent studies were carried out for 5-bromodeoxyuridine,
neuronal nuclear antigen (a marker protein of matured neuron), and glial fibrillary acidic protein
(a protein marker of mature astrocytes).
Results: Only 1% of neuronal nuclear antigen–positive and 4.6% of glial fibrillary acidic protein–
positive cells could be detected among the 5-bromodeoxyuridine-immunopositive cells in the peri-
infarcted area of the cortex and the striatum, respectively, at 14 days after hypoxic-ischemic stress.
There were no such double-staining cells at 28 and 42 days after hypoxic-ischemic stress.
Conclusion: The intrinsic ability for neurologic self-repair was limited at the maturation step after
hypoxic-ischemic stress in the neonatal rat brain.
Although there have been enthusiastic efforts to set up

new therapeutic modalities for the treatment of perinatal
hypoxic/ischemic (H/I) brain damage (such as brain
Supported in part by Grants-in-Aid for Scientific Research (B) hypothermia1 or special medication2), there is no stan-
15390273 and (Hoga) 15659338 and the National Project on Protein dard therapy that has been established for this disorder.
Structural and Functional Analyses from the Ministry of Education, Many infants with this problem have permanent neuro-
Science, Culture and Sports of Japan and by grants from the Ministry logic deficits (such as cerebral palsy, seizure, sensory
of Health and Welfare of Japan.
deficit, and mental retardation). Such chronic states of
* Reprint requests: Tomoaki Ikeda, MD, Department of Obstetrics
and Gynecology, Miyazaki Medical College, University of Miyazaki,
defect do not seem to be amenable to conventional
5200 Kihara, Kiyotake-cho, Miyazaki, 889-1692 Japan. therapy. Recent progress in regenerative medicine and
E-mail: tikeda44@hotmail.com in our understanding of neurogenesis may lead to the
doi:10.1016/j.ajog.2005.01.029
850 Ikeda et al
Figure 1 Time schedule of 5-bromodeoxyuridine (BrdU) administration and brain retrieval. H/I, Ischemic/hypoxic insult;
PND, postnatal day; Brain rtv, brain retrieval.
recovery of neuronal functions that were lost by perinatal Briefly, the pups were anesthetized with ether, and the
H/I-induced brain injury, with stem-cell therapy. left common carotid artery was sectioned between
Neurogenesis comprises at least 3 processes: cell pro- double ligatures of 4-0 surgical silk. The pups were
liferation, migration, and differentiation.3,4 Recently, allowed to recover for 2 hours and were then exposed to
some research institutes, which include our laboratory, an additional 2 hours of hypoxia (8% oxygen, 92%
reported evidence of cell proliferation and migration nitrogen) in a plastic container at 33(C. Sham-operated
during neurogenesis, with the use of a rat model of control animals were treated identically, except for the
neonatal H/I encephalopathy that leads to unilateral section of the left carotid artery and the subsequent
brain damage. There were significant increases in pro- hypoxia.
liferation of the neural stem cells in the subventricular
region, which is the most important germinal matrix of Bromodeoxyuridine labeling and
the rat, not only on the injured side but also on the experimental design
noninjured side.5 Immunohistochemistry for doublecor-
A cell proliferation marker, bromodeoxyuridine (Sigma
tin, which is a marker of migrating neuronal precursors,6,7
Chemical Company, St. Louis, Mo), was dissolved in
showed significant migration of doublecortin-positive
0.9% saline solution. Bromodeoxyuridine (50 mg/kg)
newly divided cells gathering around the infarction in
solution was injected intraperitoneally twice daily for
the injured side, but not in the noninjured side (un-
3 days from 5 to 7 days after the H/I treatment (ie, on
published data).
postnatal days 12-14; Figure 1). We chose days 5 to 7
These results encouraged us to test whether these
after H/I treatment because this time frame is within the
progenitor neuronal cells could differentiate successfully
maximal period of H/I-induced cell proliferation in our
into mature neurons. In the present study, we used
previous study and to avoid labeling proliferating glia.5
a thymidine analog, 5-bromodeoxyuridine, as an early
The animals were killed 14 days after H/I treatment
marker of cell proliferation in the normal adult rodent
(postnatal day 21), 28 days after H/I treatment (post-
brain and antibodies against neuronal nuclear antigen
natal day 35), and 42 days after H/I treatment (postnatal
(NeuN),8,9 which is a marker protein of matured
day 49). Each group contained 6 pups. For each group
neuron, and glial fibrillary acidic protein (GFAP), which
of rats, a sham control group also received bromodeox-
is a marker protein of mature astrocytes.
yuridine without the H/I insult (6 in each group).
Material and methods Tissue preparation
Animal model The pups were killed under anesthesia with pentobarbital
(50 mg/kg, intraperitoneally) and were then perfused
The Animal Research Committee of Okayama Univer- transcardially with phosphate-buffered saline solution
sity approved this study. Pregnant Wistar rats were followed by 4% paraformaldehyde in phosphate-buffered
purchased from Japan Charles River (Shizuoka, Japan). saline solution (pH 7.4). The brains were removed,
On postnatal day 7, each pup was subjected to a mod- postfixed overnight, and rapidly frozen in 2-methylbutane
ified Levine procedure to produce H/I brain injury.10 that was chilled by liquid nitrogen after cryoprotection.
Ikeda et al 851
The brains were cut serially into 10-mm–thick coronal Cell counting
sections on a cryostat at ÿ30(C, and the sections for
caudate and dorsal hippocampus levels were mounted on For single immunohistochemical staining, the numbers
glass slides and stored at ÿ80(C until used for immuno- of bromodeoxyuridine-labeled cells in each side of the
histochemistry. cortex and striatum were counted in 5 coronal sections
(10 mm thick, spaced 50 mm apart) for each animal.
Immunohistochemistry Each 25 square areas of 660 ! 660 mm were selected
randomly from the cortex and striatum just adjacent to
For immunohistochemical detection of bromodeoxyur- the infarcted area (that is, peri-infarcted areas). For cells
idine labeling, the sections were incubated in 1N HCl at that were immunofluorescently positive for both bro-
65(C for 1 hour to denature DNA and then rinsed in 0.1 modeoxyuridine and NeuN or for bromodeoxyuridine
mol/L boric acid (pH 8.5) at 25(C for 10 minutes. and GFAP, the areas around the region of infarction
Sections were incubated with mouse monoclonal anti- were photographed at 460.6 ! 460.6 mm for 5 of the
bromodeoxyuridine immunoglobulin G (IgG; 1:200; coronal sections.
Oncogene Research Products, Boston, Mass) overnight
at 4(C and were then incubated with biotinylated horse Statistical analysis
anti-mouse IgG (1:200; Vector Laboratories, Burlin-
game, Calif) for 1 hour at room temperature. Cell numbers are given as means G SDs. We compared
Immunoreactivities were developed in a horseradish the sham-operated controls with the ipsilateral and
peroxidase-streptavidin-biotin complex solution (Vec- contralateral sides of the H/I-treated rats. Differences
tastain ABC Kit; Vector Laboratories) for 30 minutes were tested with analysis of variance, followed by
and were incubated for 5 minutes in a peroxidase Tukey’s test for post-hoc analysis to evaluate intergroup
reactive solution (0.02% diaminobenzidine, 0.02% differences. A probability value of !.01 was assumed to
H2O2). Peroxidase staining was examined with an be statistically significant.
Olympus microscope (BX-51; Olympus, Tokyo, Japan).
Double immunofluorescence Results

For double immunofluorescence detection of bromo- All the rats survived from the time of H/I insult until the
deoxyuridine and NeuN, DNA-denatured sections time of death. All brains of H/I-treated rats were
were first incubated with sheep polyclonal IgG anti- damaged macroscopically on the left side.
bromodeoxyuridine (1:500; Biodesign, Saco, Mass) and
mouse monoclonal IgG anti-NeuN (1:500; Chemicon, Histologic changes
Temecula, Calif) overnight at 4(C and were then
In the cortex, the H/I-treated rats showed a specific type
incubated with FITC-labeled donkey anti-sheep IgG
of damage: Comb-like columnar degeneration appeared
(1:500; Molecular Probes, Eugene, Ore) and rhodamine-
perpendicular to the cortical surface, with alternating
labeled goat anti-mouse IgG (1:500; Chemicon) for 1
regions of preserved neurons and gliosis. Bromodeox-
hour at room temperature. We optimized the titration of
yuridine-positive cells predominated in the areas of
2 different antibodies for bromodeoxyuridine to obtain
gliosis (Figure 2, A, B, and C). In regions with massive
the equivalent sensitivity and specificity.
tissue infarction, bromodeoxyuridine-positive cells sur-
For double immunofluorescence detection of bromo-
rounded the infarcted area. In the striatum, the later-
deoxyuridine and GFAP, DNA-denatured sections were
odorsal portion was predominantly affected, and
first incubated with mouse monoclonal IgG anti-
bromodeoxyuridine-positive cells were identified mainly
bromodeoxyuridine (1:200; Oncogene Research Products)
in the damaged areas where neuronal cells were sparse
and goat polyclonal IgG anti-GFAP (1:200; Santa Cruz
(Figure 2, D, E, and F ).
Biotechnology Inc, Santa Cruz, Calif) overnight at 4(C
and were then incubated with FITC-labeled horse anti- Numbers of bromodeoxyuridine-positive cells
mouse IgG (1:500; Vector Laboratories) and Alexa
Fluor 546-labeled donkey anti-goat IgG (1:500; Molec- The number of bromodeoxyuridine-positive cells in the
ular Probes) for 1 hour at room temperature. The ipsilateral cortex of rats at 14 days after H/I treatment
specificity of primary antibodies that we used was tested was significantly higher than that in the contralateral
by omission of primary antibodies. cortex of rats at 14 days after H/I treatment or in the
Sections were scanned with a confocal microscope cortex of the control rats (121.2 G 24.8/mm2 vs 31.5 G
that was equipped with argon and HeNe1 lasers (LSM- 6.9/mm2 and 26.6 G 10.1/mm2, respectively [P ! .01];
510; Zeiss, Jena, Germany). Sets of fluorescent images Figure 3). The numbers of bromodeoxyuridine-positive
were acquired sequentially for the red and green cells in the ipsilateral cortex decreased to 43.4 G 20.9/
channels to prevent crossover of color signals. mm2 and 37.2 G 20.0/mm2 at 28 days after H/I
852 Ikeda et al
Figure 2 Immunohistochemistry for bromodeoxyuridine (BrdU) in the cerebral cortex (CTX; A to C) and striatum (ST; D to F) on
the left (infarcted) side in coronal sections of H/I-treated rats at 14 days (A, D), 28 days (B, E), and 42 days (C, F) after H/I insult.
Scale bar = 200 mm.
treatment and 42 days after H/I treatment, respectively. respectively [P ! .01 at 24 days after H/I treatment];
On these days, the ipsilateral cortex still showed 83.3 G 47.1/mm2 vs 3.7 G 3.0/mm2 and 6.2 G 7.1/mm2,
significantly more bromodeoxyuridine-positive cells respectively [P ! 0.01 at 42 days after H/I treatment]).
compared with the contralateral cortex and the control
cortex (43.4 G 20.9/mm2 vs 10.1 G 4.4/mm2 and 8.7 G Double immunostaining for bromodeoxyuridine
2.8/mm2, respectively [P ! .01 at 24 days after H/I and NeuN or GFAP
treatment]; 37.2 G 20.0/mm2 vs 6.2 G 3.2/mm2 and
4.1 G 2.5/mm2, respectively [P ! .01 at 42 days after Cells staining for both bromodeoxyuridine and NeuN
H/I treatment]). were observed sparsely in the ipsilateral cortex at 14
The number of bromodeoxyuridine-positive cells in the days after H/I treatment, but not at 28 days after H/I
ipsilateral striatum of rats at 14 days after H/I treatment treatment or 42 days after H/I treatment. These cells
was significantly higher than that in the contralateral resembled mature neurons and had unipolar or bipolar
striatum of rats at 14 days after H/I treatment or in the cell shapes (Figure 4, A-F). Such double-positive cells
striatum of the control rats (202.5 G 48.2/mm2 vs 34.8 G were never found in the contralateral cortex or other
8.7/mm2 and 30.5 G 12.6/mm2, respectively; P ! .01). areas, including the striatum, at any days after H/I
The numbers of bromodeoxyuridine-positive cells in the treatment. The concentration of such double-positive
ipsilateral striatum decreased to 103.8 G 71.6/mm2 at 28 cells was 1.2 G 2.2/mm2; thus on average, the bromo-
days after H/I treatment and 83.3 G 47.1/mm2 at 42 days deoxyuridine-positive cells that were seen in the ipsilat-
after H/I treatment. On these days after H/I treatment, eral cortex of rats at 14 days after H/I treatment
the ipsilateral striatum still showed significantly more contained 1% of cells that were also positive for NeuN.
bromodeoxyuridine-positive cells compared with the Cells that were immunopositive for both bromodeox-
contralateral striatum and the control striatum (103.8 yuridine and GFAP were observed sparsely in the
G 71.6/mm2 vs 7.6 G 5.1/mm2 and 9.0 G 6.0/mm2, ipsilateral striatum at 14 days after H/I treatment, but
Ikeda et al 853
Figure 3 Changes in the numbers of bromodeoxyuridine (BrdU) single-labeled cells in the cortex and striatum. The asterisks
indicate statistical significance (P ! .01). Note the significant increases in the number of bromodeoxyuridine-labeled cells in the left
(infarcted) side but not in the right (uninfarcted) side at all time points compared with the sham-operated controls. The number of
bromodeoxyuridine-positive cells decreased with time after the H/I treatment.
not at 28 days after H/I treatment or 42 days after H/I 5 to 7 days after the H/I insult), when cells would have
treatment. These resembled mature astroglia (Figure 4, been undergoing most proliferation in the subventricu-
C). Such double-positive cells were never found in the lar zone. This cohort of cells must have migrated to the
contralateral striatum or other areas, including the infarcted area in the cortex at 7 days after I/H insult.
cortex, on any day after H/I treatment. The concentra- Our unpublished study showed that, at 7 days after H/I
tion of such double-stained cells was 9.4 G 7.7/mm2, or treatment, 108 of 456 bromodeoxyuridine-positive cells
4.6% of the bromodeoxyuridine-positive cells. around infarcted area were immunopositive for double-
cortin, which indicated that 23.7% of newly generated
neural stem cells had become neural progenitor cells
Comment (Figure 5). At 14 days after H/I treatment, however,
only a single cell differentiated into a mature neuron of
It was unexpected and rather disappointing that only 108 migrating neural progenitor cells. These few newly
1% of mature neurons (NeuN immunopositive) could formed neurons could be found no more than 28 days
be detected among the bromodeoxyuridine-immuno- after the H/I insult. This finding is consistent with the
positive cells in the infarcted and glial cell–replaced observation by Morshead and van der Kooy11 that most
areas in the cortex of rats at 14 days after H/I treatment. newly generated neurons in the adult appear to under-
The bromodeoxyuridine-positive cells represented those go programmed cell death rather than survive to
cells that would have divided 7 to 9 days previously (ie, make a neuronal network. Furthermore, the nuclear
854 Ikeda et al
Figure 4 A-F, Double immunofluorescence for bromodeoxyuridine (BrdU; green) and NeuN (red) of the left (infarcted) cortex 14
days after H/I treatment. bromodeoxyuridine-positive cells in A and NeuN positive cells in B were overlaid in C. In this photograph, 4
double immunopositive cells (arrows) were identified. However, the concentration of such double positive cells was 1.2 G 2.2/mm2
(mean G SD) on average. D-F, Three-dimensional reconstruction shows co-localization of bromodeoxyuridine and NeuN. Higher
magnification of boxed area in C. bromodeoxyuridine positive cells in D and NeuN positive cells in E were overlaid in F. G-I, Double
immunofluorescence for bromodeoxyuridine (green) and GFAP (red) of the left striatum at 14 days after H/I insult. bromodeoxyuridine
positive cells in G and GFAP positive cells in H were overlaid in I. Scale bars are 50 mm in A-C and G-I and 10 mm in D-F.
morphologic condition of the cells that were doubly neurons to make a neural network or to contribute even
immunopositive for bromodeoxyuridine and NeuN sug- partially to a functional recovery. This significantly poor
gested that they were some form of interneurons, not ability to differentiate contrasts with the successful
pyramidal neurons. Thus, it appears that these newly differentiation into neurons and astroglia of externally
recruited neurons are not eligible to connect with other injected human neural stem cells in experiments with
Ikeda et al 855
Figure 5 Schematic explanation of limited differentiation from neural progenitor cells that are doubly immunopositive for 5-
bromodeoxyuridine (BrdU) and doublecortin (DCX) to mature neurons positive for bromodeoxyuridine and NeuN. Our
unpublished study showed that at 7 days after H/I treatment, 108 of 456 bromodeoxyuridine-positive cells that were around the
infarcted area were immunopositive for doublecortin. At 14 days after H/I treatment, however, only a single cell differentiated into
mature neuron of 108 migrating neural progenitor cells. PND, Postnatal day.
normally developing animal brains.12 This poor ability to infants16) may be required to facilitate cell differentia-
differentiate may be explained by the poor environment tion. Bromodeoxyuridine-labeled cells from the subven-
of the unilateral infarcted cortical area in this H/I brain- tricular zone that co-expressed doublecortin, and later
damage model with neonatal rats. For example, there NeuN, appear to have migrated into the ischemic cortex,
may be poor neural input and output or a poor humoral but not into the ischemic striatum. This finding is
milieu (such as low levels of neurotrophic factors). The contrary to the results of Arvidsson et al.17 Differences
poor self-repairing ability in this model seems to be in brain maturity in the models may explain this
associated with the large size of cell degeneration. Daval regional discrepancy, and further studies clearly are
et al13 reported complete self-repair in the hippocampus needed.
region CA1 of neonatal rats. In their study, transient As for astrogliosis, it was also unexpected that
hypoxic stress resulted in a rather small decrease in GFAP-positive cells, which were abundant through the
neuronal cells in this region. Therefore, our present H/I cortex and the striatum on the ipsilateral side, did not
insult seems to be too severe for successful neuronal self- show bromodeoxyuridine staining, except in the stria-
repair. tum at 14 days after H/I insult. Zaidi et al18 injected
The severely limited ability to differentiate from bromodeoxyuridine at either postnatal day 21 or 22 and
neuroblasts to mature neurons that are shown in the killed animals at postnatal day 35 to allow proliferating
present study implies that attempts to produce func- glial cells to express markers of differentiated cells. In
tional recovery by stem cell transplantation14 will be that study, some of the GFAP-positive cells were
unsuccessful for treating neonatal H/I encephalopathy, bromodeoxyuridine-positive not only in the ipsilateral
although exogenous stem-cell transplantation in this but also in the contralateral side. In the present study,
model has not been attempted. Additional manipula- bromodeoxyuridine was injected at an earlier time, so
tions (such as the administration of neurotrophic glial cell precursors that formed immediately after H/I
factors15 or an enriched environment surrounding the insult might not have survived.
856 Ikeda et al
In conclusion, we believe that successful stem-cell 7. Gleeson JG, Lin PT, Flanagan LA, Walsh CA. Doublecortin is
therapy for treating neonatal H/I encephalopathy is a microtubule-associated protein and is expressed widely by
migrating neurons. Neuron 1999;23:257-71.
limited by poor differentiation from neural precursors to 8. Mullen RJ, Buck CR, Smith AM. NeuN, a neuronal specific
mature neurons, at least in this animal model. New nuclear protein in vertebrates. Development 1992;116:201-11.
strategies to overcome this limitation will be needed. 9. Magavi SS, Leavitt BR, Macklis JD. Induction of neurogenesis in
the neocortex of adult mice. Nature 2000;405:951-5.
10. Levine S. Anoxic-ischemic encephalopathy in rats. Am J Pathol
1960;36:1-17.
References 11. Morshead CM, van der Kooy D. Postmitotic death is the fate of
constitutively proliferating cells in the subependymal layer of the
1. Gunn AJ. Cerebral hypothermia for prevention of brain injury adult mouse brain. J Neurosci 1992;12:249-56.
following perinatal asphyxia. Curr Opin Pediatr 2000;12:111-5. 12. Ourednik V, Ourednik J, Flax JD, Zawada WM, Hutt C, Yang C,
2. Ikeda T, Xia YX, Kaneko M, Sameshima H, Ikenoue T. Effect of et al. Segregation of human neural stem cells in the developing
the free radical scavenger, 3-methyl-1-phenyl-2-pyrazolin-5-one primate forebrain. Science 2001;293:1820-4.
(MCI-186), on hypoxia-ischemia-induced brain injury in neonatal 13. Daval JL, Pourie G, Grojean S, Lievre V, Strazielle C, Blaise S,
rats. Neurosci Lett 2002;329:33-6. et al. Neonatal hypoxia triggers transient apoptosis followed by
3. Iwai M, Sato K, Omori N, Nagano I, Manabe Y, Shoji M, et al. neurogenesis in the rat CA1 hippocampus. Pediatr Res 2004;55:
Three steps of neural stem cells development in gerbil dentate 561-7.
gyrus after transient ischemia. J Cereb Blood Flow Metab 14. Szentirmai O, Carter BS. Genetic and cellular therapies for
2002;22:411-9. cerebral infarction. Neurosurgery 2004;55:283-6.
4. Iwai M, Sato K, Kamada H, Omori N, Nagano I, Shoji M, et al. 15. Nakatomi H, Kuriu T, Okabe S, Yamamoto S, Hatano O,
Temporal profile of stem cell division, migration, and differentia- Kawahara N, et al. Regeneration of hippocampal pyramidal
tion from subventricular zone to olfactory bulb after transient neurons after ischemic brain injury by recruitment of endogenous
forebrain ischemia in gerbils. J Cereb Blood Flow Metab neural progenitors. Cell 2002;110:429-41.
2003;23:331-41. 16. Johansson BB. Environmental influence on outcome after exper-
5. Plane JM, Liu R, Wang TW, Silverstein FS, Parent JM. Neonatal iment brain infarction. Acta Neurochir Suppl (Wien) 1996;66:63-7.
hypoxic-ischemic injury increases forebrain subventricular zone 17. Arvidsson A, Collin T, Kirik D, Kokaia Z, Lindvall O. Neuronal
neurogenesis in the mouse. Neurobiol Dis 2004;16:585-95. replacement from endogenous precursors in the adult brain after
6. Francis F, Koulakoff A, Boucher D, Chafey P, Schaar B, Vinet stroke. Nat Med 2002;8:963-70.
MC, et al. Doublecortin is a developmentally regulated, micro- 18. Zaidi AU, Bessert DA, Ong JE, Xu H, Barks JD, Silverstein FS,
tubule-associated protein expressed in migrating and differentiat- et al. New oligodendrocytes are generated after neonatal hypoxic-
ing neurons. Neuron 1999;23:247-56. ischemic brain injury in rodents. Glia 2004;46:380-90.
www.ajog.org
Intrauterine therapy of goitrous hypothyroidism in

a boy with a new compound heterozygous mutation
(Y453D and C800R) in the thyroid peroxidase gene.
A long-term follow-up
Kirsten Börgel, MD,a,* Joachim Pohlenz, MD,b Wolfgang Holzgreve, MD,c
Jurgen H. Bramswig, MDa
Department of Pediatrics, University Children’s Hospital Münster, Münster, Germanya; Children’s Hospital of
the Johannes Gutenberg-University of Mainz, Mainz, Germanyb; Department of Obstetrics and Gynecology,
University of Basel, Basel, Switzerland c
Received for publication July 28, 2004; revised January 9, 2005; accepted January 25, 2005
KEY WORDS We report the results of intrauterine L-thyroxine therapy, and the long-term follow-up in a fetus
Fetal goiter who presented at 32 weeks’ gestation with goitrous hypothyroidism, hyperextension of the neck,
Congenital and polyhydramnios. Spontaneous delivery was possible and hypothyroidism improved.
hypothyroidism Molecular analysis revealed a new compound heterozygous mutation (Y453D/C800R) in the
Intrauterine therapy TPO gene.
TPO gene Ó 2005 Mosby, Inc. All rights reserved.
Congenital hypothyroidism occurs in 1:3000-4000 Case report

newborns. Fifteen to twenty percent have defects in
thyroid hormone synthesis, including mutations in the A 34-year-old woman, gravida 3, para 2, presented with
thyroid peroxidase (TPO) gene. We had the unique polyhydramnios and premature labor during the 32nd
opportunity to follow a patient with fetal goitrous week of gestation. Her thyroid function and thyroid
hypothyroidism caused by TPO mutations into adult- antibodies were normal, history revealed no exposure to
hood. He received L-thyroxine intraamniotically and goitrogens.
intraumbilically, as well as long-term substitution Fetal ultrasound (with an Acuson 128, Acuson,
therapy. Mountain View, Calif) demonstrated a bilobed symmet-
ric anterior neck mass measuring 4.5 ! 8.0 cm. It caused
hyperextension of the head, esophageal obstruction, and
polyhydramnios, making spontaneous delivery impossi-
ble. Fetal blood sampling (W.H.) at 32C6 weeks revealed
severe hypothyroidism with a serum total thyroxine (TT4)
* Reprint requests: Dr Kirsten Börgel, MD, Universitätskinderkli-
nik Münster, Albert-Schweitzer-Strasse 33, D-48149 Münster, Ger-
of 2.4 mg/dL (reference values 3.2-16 mg/dL, measured by
many. enzyme immunoassay ES 600, Böhringer, Mannheim,
E-mail: ullriki@uni-muenster.de Germany) and a thyroid-stimulating hormone (TSH) of
doi:10.1016/j.ajog.2005.01.060
858 Börgel et al
O200mIU/mL (reference values 3-14 mIU/mL, measured have been reported.1-3 Fetal goiter decreased rapidly in
by enzyme immunoassay ES 600, Böhringer), so that 250 size in all patients shortly after L-thyroxine therapy was
mg L-thyroxine were injected intraamniotically. given in 1 or more doses varying from 10 mg/kg to 500
Hypothyroidism had improved at 36C5 weeks with a mg.1-3
normal TT4 of 4.3 mg/dL, and a TSH of 80 mIU/mL. To our knowledge, our patient is the first who
At this time 200 mg L-thyroxine were injected into the received intraumbilical injections of L-thyroxine. We
umbilical vein. The procedure was repeated at 37C6 observed that hypothyroidism improved, and thyroid
weeks with another 100 mg L-thyroxine. Thyroid func- volume decreased, alleviating esophageal obstruction
tion had improved further. TT4 was 6.3 mg/dL, TSH and polyhydramnios.
36.6 mIU/mL. The goiter was reduced to 4.2 ! 5.1 cm so We had the unique opportunity to compare the long-
that a male infant with a small goiter was born by term outcome in our patient with the outcome in his
uncomplicated vaginal delivery at 39C6 weeks. Cord older brother affected by the same mutation in the TPO
blood showed a TT4 of 5.2 mg/dL (11.4-22.4 mg/dL), and gene, but starting L-thyroxine therapy postnatally at 3
an elevated TSH above 200 mIU/mL. Therapy was weeks of age. The prenatally treated patient had higher
started with 25 mg L-thyroxine and adjusted according IQ values, even though sensorineural hearing loss could
to weight and TSH levels during the following 25 years. not be prevented.
IQ tests showed median values of 112. Because of Prenatal thyroid hormone treatment can be consid-
sensorineural and high tone hearing loss, hearing aids ered in carefully selected patients with goitrous hypo-
were prescribed. thyroidism, especially when spontaneous delivery is
The first child was born at 36 weeks with Apgar scores impossible. Intrauterine therapy should only be done
of 4/8/9. The initial TSH screening was negative. When by experienced physicians. It is unclear whether in-
hypotonia, macroglossia, and constipation developed, trauterine L-thyroxine therapy improves central nervous
hypothyroidism was diagnosed with a TSH of 272 mIU/ system development, intellectual outcome, and auditory
mL and a slightly enlarged thyroid gland. L-thyroxine function. This case report demonstates definite short-
treatment was started at 3 weeks. Developmental tests term and possible long-term beneficial effects of this
showed an IQ of 80. Hearing function was normal at therapy.
18 years. In goitrous familiar hypothyroidism, molecular anal-
Informed consent was obtained from all individuals ysis should be performed to identify the molecular defect
for molecular analysis. Genomic DNA was isolated so that parents and physicians are aware of the genetic
from peripheral blood leukocytes using the QIAamp risk for further pregnancies.
Blood Kit (Qiagen, Hilden, Germany). The human TPO
gene was amplified and sequenced directly using an
automated sequencing system (A 377, Applied Biosys- References
tems, Weiterstadt, Germany). The 2 brothers and the
1. Abuhamad AZ, Fisher DA, Warsof SL, Slotnick RN, Pyle PG,
mother showed a heterozygous TPO gene mutation in
Wu SY, et al. Antenatal diagnosis and treatment of fetal goitrous
exon 9 (Y453D). Both brothers and the father were hypothyroidism: case report and review of the literature. Ultra-
heterozygous carriers of another mutation in exon 14 sound Obstet Gynecol 1995;6:368-71.
(C800R). Thus, the children were compound heterozy- 2. Caron P, Moya CM, Malet D, Gutnisky VJ, Chabardes B, Rivolta
gous, the parents carriers of 1 affected allele. CM, et al. Compound heterozygous mutations in the thyroglobulin
gene (1143delC and 6725G-OA [R2223H]) resulting in fetal goitrous
hypothyroidism. J Clin Endocrinol Metab 2003;88:3546-53.
3. Gruner C, Kollert A, Wildt L, Dorr HG, Beinder E, Lang N.
Comment Intrauterine treatment of fetal goitrous hypothyroidism controlled
by determination of thyroid-stimulating hormone in fetal serum. A
Only a few cases of prenatal intraamniotic thyroid case report and review of the literature. Fetal Diagn Ther
hormone treatment for fetal goitrous hypothyroidism 2001;16:47-51.
www.ajog.org
CLASSIC PAGES
Syndrome of hemolysis, elevated liver enzymes, and

low platelet count: A severe consequence of
hypertension in pregnancy
Louis Weinstein
American Journal of Obstetrics and Gynecology 1982;142:159-67.
doi:10.1016/j.ajog.2005.02.113
860 Classic Pages
Commentary by Lawrence D. Longo, MD thousands of patients with this life-threatening disorder.

Figure 1 depicts crenated erythrocytes (Burr cells or
Hemolysis, in conjunction with abnormal liver function spinocytes), while Figure 2 illustrates the irregularly
tests and thrombocytopenia, has been recognized for shaped schistocytes (split cells, also called helmet cells)
many years as a complication of severe preeclampsia- typical of microangiopathic hemolytic and other ane-
eclampsia. Yet it remained until 1982 for Weinstein, on mias.1 Experimental evidence suggests this latter cell
the basis of 29 patients with these complications, to coin conformation is a consequence of interaction between
the term HELLP Syndrome (‘‘H’’ for Hemolysis, ‘‘EL’’ red cells and fibrin strands under arterial flow condi-
for elevated liver enzymes, and ‘‘LP’’ for low platelet tions.2 Weinstein described the HELLP Syndrome while
count). Such patients often constitute a diagnostic at the University of Arizona, Tucson. Presently, he is
dilemma for the physician, who may view the hemolysis, Chair of the Department of Obstetrics and Gynecology
thrombocytopenia, hepatic disorder with possible hem- at the Thomas Jefferson University, Philadelphia.
orrhage, and/or uremic syndrome as separate and/or
unrelated entities (many patients do not, in fact, have
hypertension). Patients with this syndrome are subject to
increased risk of adverse outcomes, including death and/ References
or death of her fetus/newborn infant, such as the one
mother with severe preeclampsia and microangiopathic 1. Brain MC, Dacie JV, Hourihane DO. Microangiopathic haemolytic
anemia: the possible role of vascular lesions in pathogenesis. Brit J
hemolytic anemia and 3 infants in the present account. Haemat 1962;8:358-74.
Since this original description, and as of this writing, 2. Bull BS, Kuhn IN. The production of schistocytes by fibrin strands
there have appeared over a thousand reports of many (a scanning electron microscope study). Blood 1970;35:104-11.
doi:10.1016/j.ajog.2005.06.057
It has been a great ride: The history of HELLP syndrome

Louis Weinstein, MD*
Department of Obstetrics & Gynecology, Thomas Jefferson University, Philadelphia, PA
‘‘It is circumstances and proper timing that give an in Division I (the lying-in hospital) of Vienna’s Allge-
action its character and make it either good or bad.’’ meines Krankenhaus [general hospital], where the med-
Agesilaus (444–400 BC) ical students practiced obstetrics. It was in the hospital’s
This quote, written in approximately 400 BC, is as Division II, where obstetrics was practiced by the
true today as it was when it was stated. Timing is midwives, that Semmelweiss observed a much lower
everything. There are many examples of the value of mortality rate among the patients who were delivered, as
timing and the importance of observation that are related to puerperal infection. The importance of ob-
applicable to medicine. In 1846, Ignaz Philipp Semmel- servation, timing, and being in the right place presented
weiss (1818-1865) was appointed first assistant (lecturer) itself when Semmelweiss’s colleague, Jakob Kolletschka,
cut his hand while performing an autopsy on a post-
partum patient who had died of childbed fever.
* Reprint requests: Louis Weinstein, MD, Department of Obstet- Semmelweiss observed that Jakob’s friend’s death was
rics and Gynecology, Thomas Jefferson University, 834 Chestnut St, identical to that of the women who died from puerperal
Philadelphia, PA 19107.
fever and that the cause was likely to be from the
E-mail: louis.weinstein@jefferson.edu
transmission of the putrid organic material from exam-
0002-9378/$ - see front matter Ó 2005 Mosby, Inc. All rights reserved. iner to patient. By the initiation of the simple task that
doi:10.1016/j.ajog.2005.06.058 was instituted by Semmelweiss of scrubbing one’s hands
Classic Pages 861
in a chlorinated lime solution between patients, a eller. Neither of them was sure of the clinical diagnosis;
marked decrease in the mortality rate was noted in the but, because of the concern for acute fatty liver, we all
postpartum patients on the medical student ward.1 A agreed to proceed with delivery and to support the
similar story exists about the discovery of penicillin by patient metabolically. We administered intravenous
Sir Alexander Fleming (1881-1955). It was his observa- fluids, and labor induction ensued. Six hours later she
tion of the inhibition of the growth of bacteria around was delivered of an anencephalic infant. Over the next
the site of a mold that was growing in an open glass dish 18 hours, the patient’s platelet count continued to
that had been left in his laboratory that led to the decrease; her hematocrit level decreased because of
discovery of the antibiotic penicillin, which revolution- hemolysis with no active bleeding, and she remained
ized the treatment of bacterial infections.2 profoundly hypoglycemic but lucid. A solution of 10%
Both of these examples demonstrate the power of dextrose was administered intravenously with no change
being in the right place at the right time and making in blood sugar levels; the plan was to consider plasma-
some simple observations. This principle presented itself pheresis. Approximately 24 after delivery, she became
during my career and led to my discovery of the variant comatose, experienced cardiorespiratory arrest, and was
of preeclampsia now called hemolysis, elevated liver unable to be resuscitated.
enzymes, and low platelet count (HELLP) syndrome.3 Postmortem examination revealed a massively swol-
In 1979 after completing my obligatory military ser- len liver with multiple petechial hemorrhages, a large
vice, I was recruited to the University of Arizona to be quantity of ascites, a pancreas with internal hemorrhage,
their first maternal-fetal medicine fellow. At that time, and no obvious cerebral findings. Microscopic exami-
fellows functioned in an independent manner and took nation revealed disruption of both liver and pancreas
call independently as a faculty member. In early 1980, I cells with no obvious necrosis. A small amount of fatty
received a telephone call from an Indian Health Service deposits were present in the liver, but these were not
physician in Tuba City, Ariz, about transferring a consistent with the diagnosis of acute fatty degenera-
patient to the University hospital in Tucson. The history tion. Although the patient probably had preeclampsia,
of the patient was that she was at 29 weeks of gestation, we had no adequate explanation for her death.
had minimal elevation of blood pressure, 1 to 2+ This had a terrible personal impact on me. I had
proteinuria, blood sugar of 25 mg/dL (patient was lucid), never experienced a maternal death, and I felt that it was
platelet count of 52,000, hematocrit level of 30%, and my fault that the patient died. I then decided to devote
abnormal liver function test results. It was being the much of my time to determine the cause of her demise
attending physician on call that started my process of and to try to educate myself as to how to treat the next
being in the right (or wrong) place at the right time. patient with similar findings.
Obviously, it was apparent to me on the telephone My initial thought was that the patient had some
that the laboratory workup of the patient was in error variant of preeclampsia. I started by doing an extensive
and that I would correct this when she arrived. The literature search and looking for any pregnancy that
patient, who was brought by air transport, arrived late was complicated by hemolysis, abnormal liver functions,
in the afternoon. Her history and physical examination thrombocytopenia, and/or hypoglycemia that was asso-
were unremarkable, except that she stated that she had ciated with maternal death. I was amazed by the number
felt sick for approximately 1 week and that she had of articles, many of which were not in the obstetric
experienced right upper quadrant pain for several days. literature and many of which described patients similar
Her blood pressure was 130/84 mm Hg, and she had to the one I had treated. The key article that described a
2+ proteinuria. She was lucid and able to answer all similar entity was the report of 3 cases in the New
questions but was quite stoic, as were most of the England Journal of Medicine in 1954 by Pritchard et al4
Native American patients. As I was not able to from Texas. Their 3 patients all experienced eclampsia
determine the presentation of the fetus, an ultrasound with 1 survivor. McKay5 in 1972, Kitzmiller et al6 in
examination was performed that revealed the fetus to 1974, and Killam et al7 in 1975, all reported patients
be anencephalic. Laboratory testing was performed and similar to the patient that I had treated, each being
revealed a blood sugar of 30 mg/dL, platelet count of linked to preeclampsia. There were many other isolated
35,000, hematocrit level of 25% with schistocytes and reports in the medical literature of the presence of
Burr cells on peripheral smear, liver function test hemolysis and thrombocytopenia with various diagnoses
results that were 10 times normal, and an elevated of hemolytic uremic syndrome and thrombotic throm-
bilirubin level, with the majority being indirect. She bocytopenic purpura. The surgical literature described
had no evidence of any active bleeding or uterine the entity of spontaneous rupture of the liver that
activity. occurred in pregnant patients who all appeared to have
The clinical picture was very confusing to me, so I findings that were compatible with preeclampsia. The
consulted 2 of the best physicians I have ever known, my paper that impacted me the most was by Goodlin8 in
mentors, C. Donald Christian and William Droegemu- 1976 (‘‘Severe pre-eclampsia: Another great imitator’’).
862 Classic Pages
It was becoming more obvious to me, even early in my Over my academic career, I have been called numer-
career, that preeclampsia was truly the great imitator. ous times to discuss the treatment of patients with
Now that I am a senior clinician, I am absolutely HELLP syndrome and have reviewed many medical
convinced of this. records to determine whether negligent medical care
After I had finished educating myself by my literature occurred. Several observations that I have made are
review, I sent a notice to the physicians who were clearly repetitive in nature.
practicing obstetrics in Arizona that I was looking for The first observation is that the disease is progressive,
pregnant patients with unexplained thrombocytopenia, and often the patients do not look very sick. The second
hemolysis, and elevated liver enzymes and that I would observation is that the major marker for death is the
care for them if they were referred. Over the next 30 presence of hypoglycemia and that the blood sugar should
months, I had the opportunity to care for 29 patients be checked frequently during the labor process and that
and to learn much about this variant of preeclampsia. every attempt should be made to keep the blood sugar at
As I put the information that I gathered together, I O60 mg/dL. Often the patient will need an infusion of
appreciated that I was seeing a form of preeclampsia 10% dextrose or a push of 50% dextrose to maintain the
that often did not have hypertension of O140/90 mm blood sugar level. I believe that the hypoglycemia is
Hg, proteinuria, or edema. I was able to determine that related to decreased glycogen stores in the liver and
the most of the women who were affected had general- increased levels of circulating insulin from disruption of
ized malaise during the week before hospital admission the acini of the pancreas from cellular edema. The third
that was out of proportion with her being in the late observation is that I cannot tell from either the history or
second or early third trimester. Many of the patients physical examination the levels of the patient’s platelet
were experiencing nausea with or without vomiting and count or liver enzymes. Therefore, I often order these tests
right upper quadrant pain. Looking at the natural with minimal clinical indications other than the possibil-
progression of this entity, it appeared that the throm- ity of preeclampsia. The fourth observation is that, of all
bocytopenia occurred first, elevated liver enzymes sec- the records that I have reviewed of patients who died from
ond, and hemolysis third. Also, the disease was HELLP syndrome, the most common family of drugs that
progressive, and delivery was the only means of ending all patients have received in the days before death is not
the process. However, approximately 25% of the pa- an antihypertensive or magnesium sulfate but an antacid.
tients had their worst manifestations during the post- Most, if not all, patients with HELLP syndrome complain
partum period. What became apparent to me was that I of ‘‘heartburn’’ and therefore have been advised to take
was seeing a variant of preeclampsia. an antacid. If a patient complains of heartburn or right
As I started to put together the data for the 29 upper quadrant pain during the third trimester, clinical
patients that would result in the 1982 American Journal and laboratory evaluations are warranted. The key to
of Obstetrics and Gynecology article,3 I wanted to help the treatment of any patient with suspected HELLP
educate my fellow clinicians to assist them in recogniz- syndrome is simply recognition. Once the clinician has
ing these patients to prevent what happened to the first identified that the entity is present and that it is a variant
patient who I had seen with this entity. It was my belief of severe preeclampsia, delivery can be expedited, with the
that I needed to have a distinctive name for this prolongation of pregnancy offered only to administer
‘‘syndrome’’ so that most physicians would not forget antenatal steroids for the benefits of the developing fetus.
it once they heard it. The problem with these patients, I I am somewhat pessimistic that any direct single cure
realized, was not how to care for them when diagnosed will be found for preeclampsia. I believe that it is a
but how to recognize that they were sick and needed disease of the placenta and microvascular system that
care. It was then that the light dawned on me that what has an immunologic basis. It is a form of rejection of
these women needed was what the entity should be the paternal half of the fetal genetic makeup and
called; therefore, the term HELLP syndrome was born. signifies that it is best for the baby to leave the hostile
I was pleased that this term was both descriptive and intrauterine environment and began life in a brave new
reminded the clinician of what to give to the patient. I world.
then collected data for 57 patients, which resulted in a Currently, much research is being performed to
second article in 1985.9 My colleague at the University evaluate the immunologic basis for the disease, but little
of Tennessee, Baha Sibai, went on to publish several progress in the clinical management has ensued. I believe
larger series, because his patient volume was much that the cure for preeclampsia will remain elusive but
greater than mine.10,11 What has been most interesting that we shall improve in prolonging the pregnancy and
and gratifying since the original 1982 article is the safely inducing labor when necessary. I have done many
plethora of papers that have been published about things during my academic career and am proud of my
patients with HELLP syndrome. Little has changed in accomplishments and articles. The question we all need
the description: The entity is still a variant of severe to ask during and near the end of our careers is ‘‘Did we
preeclampsia, and delivery remains the prime therapy. truly make a difference?’’ I believe that my experience
Classic Pages 863
with this patient who died allowed me to identify this 4. Pritchard JA, Weisman R, Ratnoff OD, Vosburgh GJ. Intravas-
entity and to make other health care providers aware of cular hemolysis, thrombocytopenia and other hematologic abnor-
malities associated with severe toxemia of pregnancy. N Engl J
the serious nature of this disease so that proper medical Med 1954;250:89-98.
treatment can be offered and morbidity and death can be 5. McKay DG. Hematologic evidence of disseminated intravascular
avoided. I believe that the time and effort that I put into coagulation in eclampsia. Obstet Gynecol Surv 1972;27:399-417.
learning, writing, and teaching about HELLP syndrome 6. Kitzmiller JL, Lang JE, Yelenosky PF, Lucas WE. Hematologic
have resulted in better care and possibly the saving of assays in pre eclampsia. Am J Obstet Gynecol 1974;118:362-7.
7. Killam AP, Dillard SH, Patton RC, Pederson PR. Pregnancy-
life for other patients. I am truly grateful to my teachers induced hypertension complicated by acute liver disease and
and mentors who planted within me the seed for disseminated intravascular coagulation. Am J Obstet Gynecol
scientific curiosity and the desire to make a difference. 1975;123:823-8.
I can truly say that it has been a great ride. 8. Goodlin RC. Severe pre-eclampsia: another great imitator. Am J
9. Weinstein L. Preeclampsia/eclampsia with hemolysis elevated
liver enzymes, and thrombocytopenia. Obstet Gynecol 1985;66:
References 657-60.
10. Sibai BM, Taslimi MM, El-Nazer A, Amon E, Mabie BC,
1. Semmelweis IP. Die Aetiologie, der Begriff und die Prophylaxis des Ryan GM. Maternal-perinatal outcome associated with the
Kindbettfiebers [in German]. Vienna: C.A. Hartleben; 1861. syndrome of hemolysis, elevated liver enzymes, and low platelets
2. Fleming A. On the antibacterial action of cultures of a Penicillium, in severe preeclampsia-eclampsia. Am J Obstet Gynecol
with special reference to their use in the isolation of B influenzae. 1986;155:501-9.
Br J Exp Pathol 1929;10:226-36. 11. Sibai BM, Ramadan MK, Usta I, Salama M, Mercer BM,
3. Weinstein L. Syndrome of hemolysis, elevated liver enzymes, and Friedman SA. Maternal morbidity and mortality in 442 pregnan-
low platelet count: a severe consequence of hypertension in cies with hemolysis, elevated liver enzymes, and low platelets
pregnancy. Am J Obstet Gynecol 1982;142:159-67. (HELLP syndrome). Am J Obstet Gynecol 1993;169:1000-6.
www.ajog.org
CASE REPORTS
Recombinant human activated protein C treatment

of septic shock syndrome in a patient at 18th
week of gestation: A case report
László Medve, MD,a,* István Kis Csitári, MD,b Zsolt Molnár, MD, PhD,c
Ádám László, MD, PhDd
Departments of Anesthesiologya and Obstetrics and Gynecology,b Szent Lázár County Hospital, Salgótarján;
Medical University of Pećs,c Pećs; Bajcsy-Zsilinszky Hospital, Budapest,d Hungary
Received for publication January 11, 2005; revised March 16, 2005; accepted April 19, 2005
KEY WORDS Recombinant human-activated protein C (rhAPC) has been suggested to treat sepsis. We present
Septic shock syndrome the case of a 19-year-old pregnant patient at the 18th week of gestation with septic shock
Pregnancy syndrome that originated from urinary tract infection and was successfully treated with rhAPC.
Activated protein C
therapy Ó 2005 Mosby, Inc. All rights reserved.
Sepsis during pregnancy carries a substantial risk for cology because of back pain and fever. In 4 hours the
both mother and fetus. Recombinant human-activated symptoms of severe sepsis developed: she became un-
protein C (rhAPC) therapy has been suggested to treat conscious, tachypnoeic, tachycardic, anuric, and her
severe septic shock syndrome in nonpregnant patients.1,2 blood pressure was hardly measurable. Some of the
RhAPC treatment in pregnancy has been recently baseline parameters are summarized in the Table. Ex-
reviewed by von Dadelszen et al.3 Pregnancy was 1 of the amination of the urine sample confirmed Escherichia coli
exclusion criteria of rhAPC treatment in the PROWESS greater than 105. Septic shock syndrome originating
trial.2 However, obvious contraindications of rhAPC from urinary tract infection was diagnosed with at least
treatment during pregnancy are not known. We report 2 major (respiratory and circulatory) organ system
on a successful rhAPC treatment of severe septic shock failures with rapid progression. The patient was trans-
syndrome in a pregnant patient. ferred to the intensive care unit, where ventilation,
infusions, norepinephrine and dobutamine, hydrocorti-
Case report sone, furosemide, meropenem, and transfusion were
given. The APACHE II score was 27. Despite the
A 19-year-old patient at the 18th week of gestation was therapeutic efforts, the patient’s condition further dete-
registered to the Department of Obstetrics and Gyne- riorated, thus rhAPC treatment was decided at the
recommended dose of 24 mg/kgÿ1 hÿ1 and continued
* Reprint requests: László Medve, MD, Department of Anesthe-
siology, Szent Lázár County Hospital, Füleki út 64, Salgótarján,
for 96 hours.1 After the administration of rhAPC, the
H-3100, Hungary. patient’s general condition rapidly improved (see T24
E-mail: medve561@axelero.hu values in the Table). There was no sign of any adverse
doi:10.1016/j.ajog.2005.04.046
Medve et al 865
Table Baseline characteristics on admission (T0) and after Comment

24 hours (T24)
Severe septic shock syndrome was diagnosed in a 19-year-
Values old patient at the 18th week of pregnancy that originated
Parameters T0 T24 from a urinary tract infection. According to the ‘‘Surviv-
Mean arterial pressure (mm Hg) 40 92 ing Sepsis Campaign’’ guidelines1 aggressive treatment
Central venous pressure (mm Hg) 18 12 was started. RhAPC treatment has been suggested to
Central venous oxygen saturation (%) 53 78 treat severe sepsis in nonpregnant patients.1,2 Because of
Heart rate (beats/min) 120 108 the deteriorating condition of the patient, we started
PaO2 (mm Hg) 48 92 rhAPC administration. She rapidly improved and was
PaCO2 (mm Hg) 21 34 soon discharged from the hospital. At the 39th week of
actHCO3ÿ (mmol/L) 16.4 25.1 pregnancy a healthy female infant weighing 2900 g was
pHart 7.33 7.48
delivered by caesarean section because of fetal distress.
Creatinine (mmol/L) 94 88
At present little is known about rhAPC treatment
Prothrombin (%) 54 79
Platelet count 154 177 during pregnancy.3 In our case no adverse effects of
Procalcitonin (ng/mL) 19 9.5 rhAPC therapy on mother or fetus were detected.
C reactive protein (mmol/L) 240 225 According to our knowledge, this is the first report on
Hemoglobin (g/L) 71 107 rhAPC treatment for severe septic shock syndrome
APACHE II score 27 12 during pregnancy, with a favorable outcome for both
mother and fetus, suggesting the effectiveness and safety
of rhAPC therapy during pregnancy.
effects of rhAPC treatment. The fetus had normal heart References

rate throughout the assessment. The fetus and the
1. Dellinger RP, Carlet J, Masur H, Gerlach H, Calandra T, Cohen J,
quantity of amniotic fluid were found to be normal by et al. Surviving Sepsis Campaign guidelines for management of
ultrasound before, during, and after rhAPC treatment. severe sepsis and septic shock. Crit Care Med 2004;32:858-72.
On day 7, the patient was discharged from the hospital 2. Bernard GR, Vincent JL, Laterre PF. Efficacy and safety of
and 21 weeks later, at the 39th week of gestation, she recombinant human activated protein C for severe sepsis. N Engl
gave birth to a healthy female infant weighing 2900 g J Med 2001;344:699-709.
3. von Dadelszen P, Magee LA, Lee SK, Stewart SD, Simone S, Koren
(cesarean section was performed because of fetal distress G, et al. Activated protein C in normal human pregnancy and
probably caused by the umbilical cord wrapped around pregnancies complicated by severe preeclampsia: a therapeutic
the neck). opportunity? Crit Care Med 2002;30:1883-92.
www.ajog.org
Clinicopathologic features of six cases of primary

cervical lymphoma
John K. Chan, MD,a,* Vera Loizzi, MD,b Alessandra Magistris, MD,b
Mark I. Hunter, MD,b Joanne Rutgers, MD,c Philip J. DiSaia, MD,b
Michael L. Berman, MDb
Division of Gynecologic Oncology, Department of Obstetrics and Gynecology, Stanford Cancer Center, Stanford
University School of Medicine,a Stanford, CA; Division of Gynecologic Oncology, Department of Obstetrics and
Gynecology, Chao Family Comprehensive Cancer Center, University of California, Irvine Medical Center,b Orange,
CA; Department of Pathology, Long Beach Memorial Medical Center,c Long Beach, CA
Received for publication December 17, 2004; revised March 13, 2005
KEY WORDS Objective: Primary lymphoma of the uterine cervix is rare, with less than 60 cases reported. We
Cervix present a series of 6 patients with cervical lymphoma and review the literature.
Lymphoma Study design: Between 1988 and 2003, we identified 6 women with primary lymphoma of the
uterine cervix treated at our institutions. Data for analysis were obtained from hospital charts,
office records, and tumor registry files. We also reviewed 20 published reports on cervical
lymphoma, providing information on 58 additional patients.
Results: The median age at diagnosis was 52 years (range 40-76). Three patients had an abnormal
Papanicolaou test within 6 months of the diagnosis. Mean tumor size was 8.3 cm (range 3-14 cm).
On the basis of the Ann Arbor system of staging where ‘‘E’’ denotes extranodal tumor origin, 2
patients had stage IE, 1 had stage IIIE, and 3 had stage IVE disease. The median follow-up for
these 6 women was 33 months (range 12-120). Adding the 6 patients in our series to the 58
patients obtained from published reports, 43 had stage IE, 14 had stage IIE, 2 had stage IIIE, and
5 had stage IVE disease. There was no consistent pattern of treatment identified from our
literature review.
Conclusion: Primary lymphoma of the uterine cervix is a rare malignancy. Most patients present
with stage IE disease. Women with localized disease typically respond to various combinations of
surgery, chemotherapy, and radiotherapy. Combination chemotherapy with tailored radiother-
apy appears to be the preferred treatment option in women with advanced disease.
Primary malignant lymphoma of the genital tract is a

rare disease accounting for only 1% of extranodal
* Reprint requests: John K. Chan, MD, Stanford University lymphomas.1 Of these genital tract lymphomas, only
Medical Center, Department of Obstetrics and Gynecology, Division
of Gynecology Oncology, 300 Pasteur Dr HH333, Stanford, CA
0.6% arise from the uterine cervix.2-4 Freeman et al5
94305-5317. reported a series of 1467 patients with extranodal
E-mail: johnchan@stanford.edu lymphomas and found only 3 cases of primary cervical
doi:10.1016/j.ajog.2005.04.044
Chan et al 867
Table I International working formulation for classification of Non-Hodgkin’s lymphoma

Low grade
A Malignant lymphoma
Small lymphocytic
Consistent with chronic lymphocytic leukemia
Plasmocytoid
B Malignant lymphoma follicular, predominantly small cleaved cell
Diffuse areas
Sclerosis
C Malignant lymphoma follicular, mixed small cleaved cell and large cell
Diffuse areas
Sclerosis
Intermediate grade
D Malignant lymphoma, follicular
Predominantly large cell
Diffuse areas
Sclerosis
E Malignant lymphoma, diffuse small cleaved cell
F Malignant lymphoma, diffuse mixed-, small- and large-cell sclerosis
Epithelioid component
G Malignant lymphoma, diffuse
Large cell
Cleaved cell
Noncleaved cell
Sclerosis
High grade
H Malignant lymphoma, large cell, immunoblastic
Plasmacytoid
Clear cell
Polymorphus
Epithelioid component
I Malignant lymphoma lymphoblastic
Convoluted cell
Nonconvoluted cell
J Malignant lymphoma, small noncleaved cell
Burkitt’s
Follicular area
lymphoma. Over the last 25 years, less than 60 cases of Methods

cervical lymphoma have been reported in the literature.
The clinical presentation of primary cervical lym- Between December 1988 and January 2003, 6 women with
phoma often is similar to that of squamous cell carci- primary malignant lymphoma of the cervix were diag-
noma of the cervix. Most patients experience abnormal nosed and treated at University of California, Irvine
bleeding and have a large, bulky cervix on pelvic Medical Center and Long Beach Memorial Medical
examination. Because cervical lymphomas typically Center. After Institutional Review Board approvals
arise within the cervical stroma rather than the mucosa, from both institutions, these patients were identified
cytology is not a sensitive screening tool. Histologically, from tumor registry databases. All patients underwent
cervical lymphoma appears similar to lymphomas aris- staging with bilateral trephine marrow biopsy and com-
ing in other sites. It is important to make the correct puter tomography of the thorax, abdomen, and pelvis.
diagnosis of this uncommon disease because the treat- On the basis of the Ann Arbor staging classification for
ment for cervical lymphomas differs from more common extranodal lymphomas,6 stage IE is defined as the in-
cervical cancers. We report a series of 6 patients and volvement of a single extra-lymphatic organ or site, stage
provide a review of the literature on primary lymphoma IIE as a localized involvement of extra-lymphatic organ
of the uterine cervix. or site and of 1 or more lymph node regions on the same
868 Chan et al
Table II Clinicopathologic features and outcome of 6 cases of primary cervical lymphoma

Patients no. 1 2 3 4 5 6
Age (y) 62 40 41 49 76 52
Race White White White White Asian Asian
Menopausal Yes No No No Yes Yes
status
HRT Yes No No No No Yes
Presenting Vaginal Vaginal Severe Vaginal Vaginal bleeding Vaginal
complaints discharge bleeding weakness bleeding bleeding,
weight loss
and fatigue
Pelvic pain Abdominal pain Abdominal pain, Abdominal
weight loss, pain
fever, night
sweats
Cervical Normal Not available ASCUS Normal ASCUS CIS
cytology
Clinical 6 cm 12 cm pelvic 12 cm barrel- 6 cm cervical 3 cm cervical mass 14 cm pelvic
findings cervical mass mass shaped cervix mass mass
Ann Arbor IEA IVEA IIIEB IVEA IEA IVEB
stage
IWF histology J (high-grade, G (intermediate, G (intermediate I (high, G (intermediate, D (intermediate,
small cell, large cell, to high, large lymphoblastic) large cell, diffuse) predominantly
non-Burkitt) diffuse) cell, diffuse) large cell)
HRT, Hormone replacement therapy; ASCUS, atypical cells of undetermined significance; CIS, carcinoma in situ.
side of the diaphragm, stage IIIE as a localized involve- women had a previous history of gynecologic tract
ment of extra-lymphatic organ or site and of lymph node infection; one with genital herpes and the other with
regions on both sides of the diaphragm, and stage IVE as a syphilis. Both women were treated without sequelae.
diffuse or disseminated involvement of 1 or more extra- The most common presenting symptom was abnormal
lymphatic organs or tissues with or without associated vaginal bleeding, occurring in 4 of 6 patients. Two
lymph node involvement. Furthermore, lymphomas are women had 1 or more B symptoms typically associated
classified histologically according to the International with lymphomas, which include weight loss, weakness,
Working Formulation (IWF) for non-Hodgkin’s lym- night sweats, and fever. Three women also had history
phomas (Table I).7 Specimens were reviewed by the of an abnormal Papanicolaou (Pap) test within 6
pathologists at the institutions where the patients were months of diagnosis, 2 with atypical squamous cells of
treated. To confirm the diagnosis of lymphoma, all undetermined significance and 1 with cytologic features
pathologic specimens were subjected to immunohisto- of carcinoma in situ. All 6 patients had either a cervical
chemical and molecular studies to identify the cellular or cervicopelvic mass ranging from 3 to 14 cm. The
lineage, in particular, to differentiate B-cell and T-cell diagnoses of lymphoma were made based on cervical
lymphomas. The clinical presentation, stage of disease, punch biopsies in 3 of 6 patients. Of 2 women with
medical/surgical treatment, recurrence, and survival data cervical biopsy specimens not showing a malignancy,
were collected from hospital charts, office records, and one had a computer tomography–guided biopsy of the
tumor registry files. We performed a review of the liter- cervicopelvic mass and the other underwent an explor-
ature with 58 additional patients from 20 published atory laparotomy, bilateral salpingo-oophorectomy to
reports using a Medline search. make the diagnoses of primary cervical lymphoma. The
remaining patient had a cervical biopsy showing a
Results poorly differentiated carcinoma and underwent a radical
trachelectomy and pelvic lymphadenectomy. The cor-
We identified 6 women with cervical lymphoma among rect diagnosis of lymphoma was made on the excised
1330 treated for cervical cancers at the 2 institutions cervix. All patients underwent a metastatic workup that
from 1988 and 2003. The patients’ characteristics are included chest x-ray film, computer tomography or
summarized in Table II. The median age at diagnosis gallium scan of the abdomen and pelvis, and bone
was 52 years (range: 40-76). Of the 6 women, the median morrow biopsy. The median time from initial presenta-
parity was 2. One patient used tobacco in the past. Two tion to diagnosis was 5 months (range: 1 to 8 months).
Chan et al 869
Table III Clinical outcome of patients in published literature based on Ann Arbor stage and histology
Ann Arbor Therapy
and
No. histology
Author patients group Surgery radiation Chemotherapy Recurrence Survival
Steinfield et al16 1 IEA G H BSO Pelvis None None NED 9 mo
Tunca et al17 1 IEA G None Pelvis Vincristine None NED 2.5 y
Volpe et al18 1 IIEA E None None CHOP (6) 6 mo DOD 3 y
Harris et al9 17 4 IEA B H BSO Pelvis None None NED 2 y
C H BSO None None None NED 14 y
C H BSO Pelvis None None NED 13 y
C H RSO None None None NED 4 y
9 IEA D H BSO None Yes None NED ! 1 y
D H BSO Pelvis None None NED 3.3 y
G None Pelvis None None NED 9 y
G None PelvisCPA None None NED 6.3 y
G H None None None NED 9 y
vaginectomy
G H BSO Pelvis None None NED 7.1 y
G H Pelvis None None NED 2.2 y
G H BSO Pelvis, None 13 mo NED 6 y
vagina,
kidney
G H BSO None CVP NA DOD ! 1 y
1 IEA H H PelvisCPA None None NED 11 y
3 IIEA G H None None None NED 3 y
H H Pelvis None 2 mo DOD 10 mo
H H BSO None 1 cycle NA DOD 10 mo
Komaki et al8 3 3 IIEA F None PelvisCPA None None NED 3 y
G None Pelvis, abdomen None 4y NED 11 y
G None PelvisCPA None None NED 7 y
Taki et al19 1 IEA G H BSO Pelvis None 5 mo DOD 7 mo
Gharpure et al20 1 IEA G None Pelvis None 6 wks DOD 8 mo
Bar et al21 1 IEA G H BSO None CHOP (6) 10 mo DOD 10 mo
Mann et al22 1 IEA J H BSO Pelvis None 6 mo DOD 10 mo
Andrews et al23 1 IVE J None None CVM 6 mo DOD 6 mo
Muntz et al11 5 IEA B H BSO PelvisCPA None None NED 4.5 y
G RH BSO PLD Pelvis None None NED 10 y
G None PelvisCPA None None NED 5 y
G Staging lap Pelvis None None NED 5 y
G H BSO PLD Pelvis None None NED 4.5 y
Perren et al10 2 1 IIEA F Staging lap PelvisCPA None None DOO 18 y
1 IIEB F None None CHOP (6) None NED 6 y
Awwad et al24 1 IEA G None Pelvis None 6 mos DOD 7 mo
Stroh et al14 7 4 IEA G None Pelvis, inguinals CHOP-bleo None NED 14.4 y
G None Pelvis, abdomen CHOP-bleo None NED 13.7 y
G None Pelvis CHOP None NED 5 y
G None Pelvis ASAP NA DOD 1 y
1 IIEA G None Pelvis CHOP-bleo None NED 11.8 y
1 IIIEB G None Pelvis, abdomen, CHOP-bleo None NED 9.7 y
inguinals
1 IVA G None None CHOP-bleo Never free DOD 4 mo
Papadopoulos et al25 1 IEB H H Pelvis CHOP (8) None NED 4 y
Biswal et al26 1 1 IE G None Pelvis CHOP (3) None NED 5 mo
Nasu et al27 1 IEA G None None THP-COP None NED 1.5 y
Lee et al15 2 2 IEA G H BSO PLD Pelvis None None NED (NA)
G H BSO PLD Pelvis None None NED (NA)
870 Chan et al
Table III (Continued)

Ann Arbor Therapy
and
No. histology
Author patients group Surgery radiation Chemotherapy Recurrence Survival
Vang et al4 9 6 IE G None Pelvis Yes None NED 7 mo
G H BSO None None None NED 4.5 y
E Conization None None None NED (NA)
G None None Yes Yes DOD 1 y
G Conization None Yes None NED 5 y
G Conization None None None DOC 9 y
3 IIE G None Pelvis Yes None NED 10 y
G None Pelvis Yes None NED 5 y
B H BSO Pelvis Yes None NED 6 y
Yokoyama et al28 1 IIEA G None Pelvis CEOP (3) None NED 1 y
Chan et al, 2004 6 2IEA G H BSO PLD Pelvis None None NED 1 y
J TRACH PLD None None 3 mo DOO 6 y
1IIIEB G None None CHOP (8) 29 mo DOD 3 y
2IVEA G None Pelvis CHOP (8) None NED 10 y
I H BSO PLD None CHOP (9) None NED 2.5 y
1IVEB D None None CHOP None NED 1 y
ASAP, Doxorubicin, solumedrol, araC, cisplatin; CEOP, cyclophosphamide, epirubicin, vincristine, prednisolone; CHOP-bleo, CHOP and bleomycin; CVP,
cyclophosphamide, vincristine, prednisolone; DOD, dead of disease; DOO, died of other causes; H, hysterectomy; NED, no evidence of disease; NA, not
assessed; PA, para-aortic nodes; PLD, pelvic lymphadenectomy; RH, radical hysterectomy; RSO, residual salpingo-oophorectomy; TRACH, trachelectomy;
BSO, bilateral salpingo-oophorectomy; THP-CHOP, pirarubicin and CHOP.
Of the 6 patients, 2 had stage IE, 1 had stage IIIE, The patient with stage IIIEB disease underwent 8
and 3 had stage IVE disease. Two patients, with stage cycles of cyclophosphamide, doxorubicin, vincristine,
IIIE and IVE disease, presented with B symptoms. and prednisolone (CHOP). She had a disease-free inter-
According to the IWF classification, 4 women had val of 29 months and recurred in the abdominal retro-
intermediate grade B-cell lymphomas, 3 with group G peritoneal lymph nodes. She was treated with CHOP
(diffuse, large cell), and 1 with group D (large cell). and rituxan for her recurrence but died 11 months later.
Moreover, 2 patients had high-grade B-cell lymphomas, All 3 patients with stage IVE disease received combina-
1 with group I (lymphoblastic) and 1 with group J (small tion chemotherapy. One was diagnosed with an inter-
cell, non-Burkitt).7 Of the 2 patients with stage IE mediate grade (group G) disease and received 8 courses
disease, one had a cervical biopsy specimen that revealed of CHOP with pelvic radiotherapy for a 12-cm pelvic
a poorly differentiated carcinoma. Because she had a tumor. She is disease free 10 years after her diagnosis.
previous subtotal hysterectomy for leiomyoma, she The other one with intermediate grade (group D) disease
underwent a trachelectomy with bilateral pelvic lym- received 8 cycles of CHOP with rituxan and is alive
phadenectomy. The final diagnosis was consistent with a without disease 1 year after her treatment. The remain-
high-grade, small cell, non-Burkitt B-cell lymphoma. ing patient had high-grade (group I) disease for which
Adjuvant chemotherapy was delayed secondary to post- she received 5 cycles of CHOP with rituxan. Because of
operative complications, including a distal colonic ob- persistent, bulky, cervical disease, she underwent a total
struction requiring a transverse loop colostomy and an abdominal hysterectomy. Postoperatively, she was trea-
acute cholecystitis for which she underwent a cholecys- ted with an additional 3 cycles of chemotherapy and is
tectomy. Three months after the primary surgery for her alive without disease 3 years after her diagnosis.
lymphoma, she experienced an intra-abdominal recur- Immunohistochemical data were available in all 6
rence to the liver and spleen. She received 6 courses of patients. All cases stained positive for CD-20. BCL-2
bleomycin, doxorubicin, cyclophosphamide, and vin- and monotypic immunoglobulin lambda light chain was
cristine (BACOD) and remained free of disease for more found to be positive in 4 speicmens and CD-19 in 3
than 6 years. She subsequently died of causes unrelated specimens. In addition, 1 tumor sample was positive for
to her lymphoma. The other patient with stage IE CD-10, CD-23, and CD43. We obtained information on
lymphoma underwent a total abdominal hysterectomy, 58 additional patients from 20 published reports using a
bilateral salpingo-oophorectomy, and pelvic lymphade- Medline search. Including the 6 patients in our series, 43
nectomy. She received adjuvant pelvic radiotherapy and patients had stage IE, 14 had stage IIE, 2 had stage IIIE,
remains free of disease 14 months after her diagnosis. and 5 had stage IVE disease. There was no consistent
Chan et al 871
pattern of treatment identified from the published apy. On the other hand, others have suggested combi-
reports (Table III). nation CHOP chemotherapy with radiation, particularly
in patients with bulky stage I to II disease.14 Because the
survival rates in patients with earlier stage cervical
Comment lymphomas are excellent with either surgery, chemo-
therapy, or radiotherapy, young women who desire to
Lymphoma of the uterine cervix is an uncommon retain their fertility will benefit from combination che-
disease that provides a diagnostic challenge for the motherapy.10
clinician and the pathologist. Because the majority of On the basis of previous studies of patients with
patients present with abnormal vaginal bleeding, a extranodal non-Hodgkin’s lymphoma, combination
gynecologist usually performs the initial evaluation chemotherapy has been used in treating patients with
and treatment.8-11 Because cervical lymphomas typically disease that is more advanced than stage IE, with CHOP
originate from the cervical stroma, the superficial squa- as the most common regimen used.15 All 4 women with
mous epithelium is often preserved, leading to a false- stage IIIE to IV disease in our series received the CHOP
negative cytologic smear.11-13 Of 5 patients who had regimen, with or without surgery or radiotherapy. Three
cytologic smears in our series, 2 were normal, 2 had are alive without any evidence of disease at 1, 2.5, and
atypical cells of undetermined significance, and the 10 years after their diagnosis. The patient who died from
remaining 1 showed carcinoma in situ. Similarly, Harris her disease received CHOP after a disease-free interval
et al9 found that only 5 of 10 patients in their study had of 29 months and died 11 months after her recurrence
a positive cytologic smear showing malignant cells. despite further treatment. Lastly, chemotherapy alone
Although the Pap smear test can show false-negative may not be sufficient in the treatment of stage II to IV
results in patients with cervical lymphoma, 3 of 6 disease with bulky tumors in the pelvis. Some authors
women in our study had an abnormal Pap smear test. have advocated surgery or radiotherapy in addition to
The diagnosis of cervical lymphoma is typically made combination chemotherapy to decrease the risk of
with a histologic analysis of a deep cervical biopsy. central recurrence.10,14
Harris et al9 reported that up to 67% of these tumors Malignant lymphoma of the uterine cervix is a rare
presented with a subepithelial mass without obvious disease. The diagnosis is typically made with a histologic
ulceration or epithelial abnormality. The authors rec- analysis of a deep cervical biopsy. Patients with localized
ommended a deep incisional or excisional biopsy to stage IE disease with nonbulky lesions often will re-
establish a definitive diagnosis when the initial biopsy is spond to primary surgery, chemotherapy, or radiother-
nondiagnostic. In such instance, the initial biopsy usu- apy alone; however, combined therapy seems to be
ally contains atypical epithelial cells coexisting with favored in many centers. Combination chemotherapy
lymphoid infiltrates. The authors recognize that the with tailored radiotherapy is the preferred treatment
Ann Arbor staging system has been replaced by option in women with more advanced disease.
the Word Health Organization classification. Because
the majority of case reports provided in this extensive
literature review used the Ann Arbor staging system, we
reported our series as such to provide a standard for References
comparison of the therapeutic outcome. 1. Dang HT, Terk MR, Colletti PM, Schlaerth JB, Curtin JP. Primary
Primary cervical lymphomas have a rapid growth lymphoma of the cervix: MRI findings with gadolinium. Magn
pattern. Six patients had a normal pelvic examination Reson Imaging 1991;9:941-4. Available from http://www.ncbi.nlm.
within the past year but presented with a 6-cm or greater nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=
cervical lesion. Muntz et al11 reviewed a series of Citation&list_uids=1766319. Accessed on March 1, 2005.
2. Abbas MA, Birdwell R, Katz DS, Chang H, Ostrow K. Primary
patients with stage IE cervical lymphomas and found lymphoma of the cervix in a heart transplant patient. AJR Am J
that half of these women had tumors larger than 4 cm at Roentgenol 1996;167:1136-8. Available from http://www.ncbi.nlm.
presentation. nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=
Because cervical lymphoma is an uncommon malig- Citation&list_uids=8911165. Accessed on March 1, 2005.
nancy, the standard treatment has not been established. 3. Kuo HC, Chou CY, Chang CH, Liu MT, Tzeng CC, Huang KE.
Primary malignant lymphoma of the uterine cervix shows favor-
Some authors have recommended primary surgical able response to neoadjuvant chemotherapy. Gynecol Oncol
treatment for stage IE disease, whereas others have 1994;52:408-10. Available from http://www.ncbi.nlm.nih.gov/
suggested combination chemotherapy with or without entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Citation&list_
radiotherapy.11,14 Muntz et al11 advocated primary uids=7512523. Accessed on March 1, 2005.
surgery, followed by individualized radiation therapy 4. Vang R, Medeiros LJ, Ha CS, Deavers M. Non-Hodgkin’s
lymphomas involving the uterus: a clinicopathologic analysis of
for those with stage IE disease localized in the pelvis. 26 cases. Mod Pathol 2000;13:19-28. Available from http://www.
For women with stage IE diffuse bulky disease, they ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed
recommended high-dose radiotherapy with brachyther- &dopt=Citation&list_uids=10658906. Accessed on March 1, 2005.
872 Chan et al
5. Freeman C, Berg JW, Cutler SJ. Occurrence and prognosis of 17. Tunca JC, Reddi PR, Shah SH, Slack ST. Malignant non-Hodg-
extranodal lymphomas. Cancer 1972;29:252-60. Available from kin’s-type lymphoma of the cervix uteri occurring during pregnancy.
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db= Gynecol Oncol 1979;7:385-93. Available from http://www.ncbi.nlm.
PubMed&dopt=Citation&list_uids=5007387. Accessed on March nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=
1, 2005. Citation&list_uids=447122. Accessed on March 1, 2005.
6. Carbone PP, Kaplan HS, Musshoff K, Smithers DW, Tubiana M. 18. Volpe R, Tirelli U, Tumolo S, Talamini R, Grigoletto E, Carbone
Report of the Committee on Hodgkin’s Disease Staging Classifica- A. Stage IIE diffuse small cleaved cell lymphoma of the cervix.
tion. Cancer Res 1971;31:1860-1. Available from http://www. Pathologica 1983;75:887-92. Available from http://www.ncbi.nlm.
ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt= nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=
Citation&list_uids=5121694. Accessed on March 1, 2005. Citation&list_uids=6674906. Accessed on March 1, 2005.
7. National Cancer Institute sponsored study of classifications of 19. Taki I, Aozasa K, Kurokawa K. Malignant lymphoma of the uterine
non-Hodgkin’s lymphomas: summary and description of a work- cervix: cytologic diagnosis of a case with immunocytochemical
ing formulation for clinical usage. The Non-Hodgkin’s Lymphoma corroboration. Acta Cytol 1985;29:607-11. Available from http://
Pathologic Classification Project. Cancer 1982;49:2112-35. Avail- www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=
able from http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd= PubMed&dopt=Citation&list_uids=3895797. Accessed on March
Retrieve&db=PubMed&dopt=Citation&list_uids=6896167. Ac- 1, 2005.
cessed on March 1, 2005. 20. Gharpure KJ, Mahesh D, Bhargava MK. Malignant lymphoma of
8. Komaki R, Cox JD, Hansen RM, Gunn WG, Greenberg M. the uterine cervix–a report of two cases with review of literature.
Malignant lymphoma of the uterine cervix. Cancer 1984;54: Indian J Cancer 1985;22:296-302. Available from http://www.ncbi.
1699-1704. Available from http://www.ncbi.nlm.nih.gov/entrez/ nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=
query.fcgi?cmd=Retrieve&db=PubMed&dopt=Citation&list_uids= Citation&list_uids=3916333. Accessed on March 1, 2005.
6383595. Accessed on March 1, 2005. 21. Bar BM, Reijnders FJ, Keuning JJ, Bal H, van Beek M. Primary
9. Harris NL, Scully RE. Malignant lymphoma and granulocytic malignant lymphoma of the uterine cervix associated with cold-
sarcoma of the uterus and vagina: a clinicopathologic analysis reacting autoantibody-mediated hemolytic anemia. Acta Haematol
of 27 cases. Cancer 1984;53:2530-45. Available from http:// 1986;75:232-5. Available from http://www.ncbi.nlm.nih.gov/
www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db= entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Citation&list_
PubMed&dopt=Citation&list_uids=6585265. Accessed on March uids=3096057. Accessed on March 1, 2005.
1, 2005. 22. Mann R, Roberts WS, Gunasakeran S, Tralins A. Primary
10. Perren T, Farrant M, McCarthy K, Harper P, Wiltshaw E. lymphoma of the uterine cervix. Gynecol Oncol 1987;26:127-34.
Lymphomas of the cervix and upper vagina: a report of five cases Available from http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?
and a review of the literature. Gynecol Oncol 1992;44:87-95. Avail- cmd=Retrieve&db=PubMed&dopt=Citation&list_uids=3792930.
able from http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd= Accessed on March 1, 2005.
Retrieve&db=PubMed&dopt=Citation&list_uids=1730432. Ac- 23. Andrews SJ, Hernandez E, Woods J, Cook B. Burkitt’s-like
cessed on March 1, 2005. lymphoma presenting as a gynecologic tumor. Gynecol Oncol
11. Muntz HG, Ferry JA, Flynn D, Fuller AF Jr, Tarraza HM. Stage 1988;30:131-6. Available from http://www.ncbi.nlm.nih.gov/
IE primary malignant lymphomas of the uterine cervix. Cancer entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Citation&
1991;68:2023-32. Available from http://www.ncbi.nlm.nih.gov/ list_uids=3366389. Accessed on March 1, 2005.
entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Citation 24. Awwad JT, Khalil AM, Shamseddine AI, Mufarrij AA. Primary
&list_uids=1913551. Accessed on March 1, 2005. malignant lymphoma of the uterine cervix: is radiotherapy the best
12. Matsuyama T, Tsukamoto N, Kaku T, Matsukuma K, Hirakawa therapeutic choice for stage IE? Gynecol Oncol 1994;52:91-3.
T. Primary malignant lymphoma of the uterine corpus and cervix: Available from http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?
report of a case with immunocytochemical analysis. Acta Cytol cmd=Retrieve&db=PubMed&dopt=Citation&list_uids=8307508.
1989;33:228-32. Available from http://www.ncbi.nlm.nih.gov/ Accessed on March 1, 2005.
entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Citation 25. Papadopoulos AJ, Pambakian H, Devaja O, Raju KS. High grade
&list_uids=2522702. Accessed on March 1, 2005. non-Hodgkins stage IEB primary malignant lymphoma of the
13. Grace A, O’Connell N, Byrne P, Prendiville W, O’Donnell R, cervix and upper vagina: a case report. Eur J Gynaecol Oncol
Royston D, et al. Malignant lymphoma of the cervix: an un- 1996;17:484-6. Available from http://www.ncbi.nlm.nih.gov/
usual presentation and a rare disease. Eur J Gynaecol Oncol entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Citation&
1999;20:26-8. Available from http://www.ncbi.nlm.nih.gov/entrez/ list_uids=8971523. Accessed on March 1, 2005.
query.fcgi?cmd=Retrieve&db=PubMed&dopt=Citation&list_uids= 26. Biswal BM, Sharma A, Sharma MC, Prasad RR, Roy K, Rath
10422676. Accessed on March 1, 2005. GK. Malignant lymphoma of the uterine cervix: a case report and
14. Stroh EL, Besa PC, Cox JD, Fuller LM, Cabanillas FF. Treat- review of literature. J Indian Med Assoc 1997;95:434, 40 Available
ment of patients with lymphomas of the uterus or cervix with from http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&
combination chemotherapy and radiation therapy. Cancer 1995; db=PubMed&dopt=Citation&list_uids=9425847=Retrieve&db=
75:2392-9. Available from http://www.ncbi.nlm.nih.gov/entrez/ PubMed&dopt=Citation&list_uids=9425847. Accessed on
query.fcgi?cmd=Retrieve&db=PubMed&dopt=Citation&list_uids= March 1, 2005.
7712450. Accessed on March 1, 2005. 27. Nasu K, Yoshimatsu J, Urata K, Miyakawa I. A case of primary
15. Lee KM, Seah ES, Sethi VK. Primary non-Hodgkin’s lymphoma non-Hodgkin’s lymphoma of the uterine cervix treated by combi-
of the uterine cervix: case report of long-term survival of two nation chemotherapy (THP-COP). J Obstet Gynaecol Res
patients treated with surgery and radiotherapy. Australas Radiol 1998;24:157-60. Available from http://www.ncbi.nlm.nih.gov/
1998;42:126-7. Available from http://www.ncbi.nlm.nih.gov/ entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Citation&
entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Citation list_uids=9631606. Accessed on March 1, 2005.
&list_uids=9599826. Accessed on March 1, 2005. 28. Yokoyama Y, Sato S, Xiao YH, Tanaka K, Maruyama H, Saito
16. Steinfeld AD. Histiocytic lymphoma of the cervix. Gynecol Oncol Y. Primary non-Hodgkin’s lymphoma of the uterine cervix. Arch
1979;8:97-103. Available from http://www.ncbi.nlm.nih.gov/ Gynecol Obstet 2001;265:108-11. Available from http://www.
entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Citation&list_ ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&
uids=378776. Accessed on March 1, 2005. dopt=Citation&list_uids=11409473. Accessed on March 1, 2005.
www.ajog.org
Port site ischemic necrosis: An unforeseen complication

of laparoscopic surgery
R. Oliver, MBBS, MRCOG, A. Coker, MBBS, MRCOG, Cert Lap Sur (RCOG), Cert BSCCP*
Harold Wood Hospital, Romford, United Kingdom
Received for publication January 23, 2005; accepted March 4, 2005
KEY WORDS A case of necrosis of the 10-mm umbilical port site after laparoscopic closure of enterocele in a
Laparoscopy 66-year-old woman is reported. Ischemic necrosis of the superficial tissue with cellulitis of the
Port site umbilical wound was seen during debridement. This highlights a hitherto unforeseen complica-
Necrosis tion of laparoscopic surgery.
Laparoscopic techniques have become the forefront complications. The 10-mm port incision was closed
of gynecologic operating in the past decade. In parallel with 2 stitches of no. 1 vicryl.
with the increase in the number of procedures per- She was discharged home as asymptomatic, on the
formed, we are also seeing an increase in different third postoperative day. On the fifth postoperative
techniques and methods. In conjunction with the above day, the patient noticed pain and discoloration of the
mentioned rise, we are observing an expansion in the wound, which worsened over the next 48 hours. She was
types of complications that could arise. readmitted on the seventh postoperative day and the
A hitherto unreported complication of ischemic ne- scar appeared to be tender, with areas of skin discolor-
crosis of the tissue at the site of suturing of the port ation and erythema. Intravenous antibiotic therapy was
incision is reported. commenced. She was seen by the surgeons on the tenth
postoperative day and examination showed the wound
to be necrotic with cellulitis and abscess formation.
Case report Debridement of the wound was performed, no com-
munication was seen intra-abdominally and the skin and
A 66-year-old woman was referred in June 2003 for superficial tissues, which were found to be necrotic, were
laparoscopic closure of enterocele. A 10-mm umbilical excised. The postoperative period was uneventful.
port was introduced for the laparoscope. An open Histology showed skin and subcutaneous tissue ne-
laparoscopic entry (Hasson) was performed for entry crosis of 1.8 ! 1 cm area, with abscess and granulation
into the abdominal cavity, and adhesions were found in tissue formation.
the upper abdomen. There were no intra-operative
* Reprint requests: Mr A. Coker, MBBS, MRCOG, Cert. Lap. Comment

Sur (RCOG), Cert, BSCCP, Consultant Gynaecologist, Harold Wood
Hospital, Gubbins Lane, Romford-RM3 OBE, United Kingdom. An infective cause for the initiation of the wound
E-mail: yemicoker@aol.com necrosis was not considered during the management,
doi:10.1016/j.ajog.2005.03.039
874 Oliver and Coker
as the symptoms started with pain and discoloration of area of skin and subcutaneous tissue caused by suturing
the skin. The clinical course suggested that infection was has not been raised. To the best of our knowledge, this
superimposed on to the necrosed tissue by the tenth rare complication has not been previously reported.
postoperative day. Also, the patient was not diabetic
and did not give a previous history of poor wound
healing. Hence, an ischemic cause for the wound necro- Conclusion
sis was considered.
The causative factors for postoperative ischemic The rapidly expanding field of laparoscopic surgery has
necrosis would be difficult to ascertain, but it could be resulted in a greater frequency and variety of complica-
due to an inherent defect of the anterior abdominal wall tions.2 Modifications of the technique of closure of port
anatomy and blood supply or it could be due to site incisions may be required to prevent obstruction of
problems arising during the suturing. In the light of blood supply to the skin and subcutaneous tissues,
the above considerations, it might be more pertinent to which could be the primary causative factor in the
use rapidly absorbable sutures in view of the fact that above case of necrosis. Reviews of the current closure
anatomic approximation for only a minimal number of techniques may be necessary to avoid further similar
days is necessary to facilitate complete healing. Also, complications.
techniques of closure could be modified with devices to
enable approximation under better vision to ensure only
the wound edges and not a wide area of tissue is taken in
the suture. References
Complications occurring at port sites are not uncom-
1. Shindholimath VV, Seenu V, Parshad R, Chaudhry R, Kumar A.
mon and have been described mainly as infections,
Factors influencing wound infection following laparoscopic chole-
bleeding, haematoma formation, and also long-term cystectomy. Trop Gastroenterol 2003;24:90-2.
complications such as hernia formation.1 So far the 2. Munro MG. Laparoscopic access: complications, technologies, and
question of obstructing the blood supply to a confined techniques. Curr Opin Obstet Gynecol 2002;14:365-74.
www.ajog.org
Spontaneous epidural hematoma of the spine

in pregnancy
Ashley S. Case, MD,* Patrick S. Ramsey, MD
Department of Obstetrics and Gynecology, Center for Research in Women’s Health, University of Alabama
at Birmingham, Birmingham, AL
Received for publication December 13, 2004; revised February 16, 2005; accepted February 22, 2005
KEY WORDS Spontaneous spinal epidural hematomas are uncommon. Progressive neurologic deficits that are
Spontaneous epidural associated with epidural hematomas can develop rapidly and require prompt treatment. We
hematoma present a case of spontaneous epidural hematoma of the thoracic spine that complicated a term
Pregnancy pregnancy.
Spontaneous spinal epidural hematomas (SSEHs) are strength, absent lower extremity reflexes, no proprio-
exceedingly rare, with an incidence of approximately ception, and complete sensory loss bilaterally to the level
1 per 1,000,000 individuals.1 Clinical manifestations of T7. Magnetic resonance imaging (MRI) revealed a
include the acute onset of back or neck pain with 4-cm intraspinal mass with an intensity that was consis-
subsequent rapid and progressive loss of sensory and tent with acute hemorrhage that extended from T6
motor function. Prompt surgical decompression remains to T9 spinal levels anterior to the cord with extrinsic com-
the optimal treatment.1 We present a case of SSEH of pression noted (Figure).
the thoracic spine that complicated a term pregnancy. Emergent cesarean delivery was performed and fol-
lowed by a T7-T9 laminectomy with evacuation of an
Case report epidural hematoma. Follow-up MRI on postoperative
day 1 revealed decompression and partial re-expansion of
A healthy 30-year-old woman (para 1001) was seen at 37 the spinal cord. By postoperative day 6, she had regained
weeks 6 days of gestation with a 10-hour history of back slight lower extremity sensation with some movement in
pain and acute progressive lower extremity paresthesias, her distal lower extremities. She underwent intensive
weakness, and sensory loss. On examination she was rehabilitation and regained ambulatory status by the
afebrile with stable vital signs aside from elevated blood fourth month after her delivery. By 6 months, she had
pressures and new onset proteinuria that was consistent regained all of her strength and sensation. At the 1-year
with preeclampsia. Neurologic examination was re- follow-up evaluation, she continued to have excellent
markable for lower extremity paraplegia with 0/5 motor neurologic recovery with no residual morbidities.
* Reprint requests: Ashley S. Case, MD, Department of Obstetrics Comment

and Gynecology, University of Alabama at Birmingham, 618 South
19th St, Old Hillman Building- Room 340, Birmingham, AL 35233. We have presented a case of SSEH that compli-
E-mail: Ashley.case@obgyn.uab.edu cated pregnancy. A review of the English literature
doi:10.1016/j.ajog.2005.03.015
876 Case and Ramsey
Figure Sagittal T1-weighted MRI shows a large epidural hematoma (arrows) with direct compression of the spinal cord.
(MEDLINE 1966-2000; keywords: spontaneous spinal Complete and permanent neurologic deficits can occur;
epidural hematoma and pregnancy) revealed only 4 death has been reported rarely.1 The extent of the
previous reported cases of SSEH in pregnancy.2-5 symptoms is typically dependent on the spinal level
SSEH, although rare, can cause rapidly evolving neu- and degree of compression. Less commonly, patients
rologic deficits within minutes to hours. The onset is may present in a subacute manner.
typically acute, with an onset of sudden back or neck SSEHs are thought to result from low pressure within
pain followed quickly by signs of cord compression. the epidural space relative to venous pressure, which
Case and Ramsey 877
predisposes to rupture of a pre-existing pathologic decisions in pregnancy depend on the gestational age.
venous system.1,5 Factors such as hypertension, hemo- Decompression can be performed adequately and suc-
philia, vasculitis, anticoagulants, and Valsalva maneu- cessfully intrapartum, with the patient in the lateral
ver may also predispose to bleeding and hematoma position.2,3 For women at term, cesarean delivery may
formation.1 During pregnancy, a combination of the be advantageous, followed by surgical spinal decom-
aforementioned factors and the changing venous dy- pression, especially because the early delivery of the
namics of pregnancy may accentuate the potential for fetus decreases epidural venous engorgement. Func-
rupture.2-5 Preeclampsia may have contributed signifi- tional recovery after surgery is related directly to the
cantly to the development of SSEH in this patient, interval between symptom onset and surgical decom-
which is a finding that has never been reported. pression and the rate of development and size of the
MRI is the diagnostic imaging modality of choice. hematoma.1 Although uncommon, SSEH is an acute
The exact location and size of the lesion must be neurologic emergency that mandates prompt recogni-
ascertained as accurately as possible to prepare for tion and treatment.
surgical treatment. Computerized tomography can be
falsely negative, and myelography is less specific and
more invasive. The presence of spinal cord edema
References
or ischemia is especially worrisome and substantially
reduces the potential for functional recovery. Prompt 1. Groen RJ, van Alphen HA. Operative treatment of spontaneous
surgical decompression by laminectomy with hematoma spinal epidural hematomas: a study of the factors determining
evacuation is the standard of care in most circum- postoperative outcome. Neurosurgery 1996;39:494-509.
2. Bidzinski J. Spontaneous spinal epidural hematoma during preg-
stances. Nonoperative treatment has been reported with
nancy. J Neurosurg 1966;24:1017.
good outcomes in patients with mild symptoms. Mea- 3. Carroll SG, Malhotra R, Eustace D, Sharr M, Morcos S. Sponta-
sures such as immobilization, the administration of neous spinal extradural hematoma during pregnancy. J Matern
steroids, and the treatment of coagulopathy may result Fetal Med 1997;6:218-9.
in rapid improvement. Before laminectomy, measures 4. Yonekawa Y, Mehdorn HM, Nishikaw M. Spontaneous
spinal epidural hematoma during pregnancy. Surg Neurol
such as hyperventilation and minimization of crystalloid
1975;3:327-8.
and glucose levels improve spinal cord perfusion. How- 5. Steinmetz MP, Kalfas IH, Willis B, Chalari A, Harlan R. Successful
ever, surgery is recommended if progressive neurologic surgical management of a case of spontaneous epidural hematoma
symptoms develop or recovery is not rapid.3 Treatment of the spine during pregnancy. Spine J 2003;3:539-42.
www.ajog.org
Incisional hernia on the 5-mm trocar port site and

subsequent wall endometriosis on the same site:
A case report
Rodolfo Sirito, MD,a,* Andrea Puppo, MD,b Maria Grazia Centurioni, MD,b
Claudio Gustavino, MDb
Department of Obstetrics and Gynaecology, Ospedale Evangelico Internazionalea; Department of

Gynaecology Oncology, Istituto Nazionale per la Ricerca sul Cancro, Genova, Italyb
Received for publication November 2, 2004; accepted February 22, 2005
KEY WORDS A 26-year-old nulliparous woman underwent a laparoscopy to remove a 10-cm endometrial cyst
Incisional hernia on the left ovary (type II Nezhat). The cyst was extracted from the 10-mm umbilical incision; the
Wall endometriosis other 2 trocars were inserted through 5-mm incisions. One year later, in correspondence to the
Trocar previous 5-mm incision site, a hernia occurred that contained omentum and was reduced easily
with a local anesthetic. After 2 years of good health, an aching nodule occurred on the same
trocar site; vaginal ultrasound examination showed another left ovarian cyst. A second
laparoscopy was performed; the cyst was very adherent and was removed in fragments. The wall
nodule was removed, and the histologic examination classified it as endometriosis.
A 26-year-old nulliparous patient, with recurrent One year later, after a physical strain, the patient
pelvic pain and dysmenorrhea, was given a gynecologic noticed a tumefaction in correspondence to the previ-
examination. The examination revealed a 10-cm cyst on ous 5-mm incision site; the clinical examination
the left ovary, probably endometriosis, at ultrasound suggested the existence of a hernia, which contained
examination. A laparoscopy was performed to remove omentum, which was reduced easily with a local
the cyst, which was very easily cleavable (type II Nezhat). anesthetic.
The cyst was extracted from the 10-mm umbilical After 2 years of good health, an aching nodule arose
incision; the other 2 trocars were put through 5-mm on the same trocar site; vaginal ultrasound examination
incisions. Histologic evaluation confirmed the diagnosis. showed another left ovarian cyst. A second laparoscopy
Subsequently, the patient was treated with gonadotropin- was performed; the cyst was very adherent and was
releasing hormone analogues for 6 months. removed in fragments. The wall nodule was removed,
and the histologic examination classified it as endome-
triosis (Figure).
* Reprint requests: Rodolfo Sirito, MD, V.le C. Canepa 22, 16154,
The patient has been treated with gonadotropin-
Genoa, Italy. releasing hormone analogues and oral contraceptives
E-mail: rodolfo.sirito@fastwebnet.it and is experiencing complete remission.
doi:10.1016/j.ajog.2005.02.121
Sirito et al 879
Figure Wall endometriosis nodule sectioned after removal.
Comment breach in a major fascial defect, not well repaired

during abdominal suture.
This clinical case raises 2 main questions: What caused As regard to the second question, we recall that, for a
the incision hernia on the 5-mm trocar site? How could long time, essentially 2 theories were formulated to
endometriosis wall implantation happen without direct explain the presence of ectopic endometrial tissue; this
contact during laparoscopic surgery? case could provide another cause of discussion in
Although the incidence of incision hernias on the addition to the controversial and partly unknown
trocar incision site was largely documented in literature pathogenesis of this disease.
(incidences, 21/100,000),1 with regard to the first ques- Coming back to our clinical case, a question remains
tion it is interesting to consider that, at the moment on about a possible explanation about the cause of the
the basis of available data, the recommendation is to endometriosis nodule on the 5-mm incision site, which
repair iatrogenic fascial damages that are O10 mm. In again highlights the fact that no endometrial lesions
fact, cases of incisional hernia on trocar incision sites were extracted from this site.
that are !5 mm are described on occasion.2 At last, making clear that the comprehension of
To comprehend the reason that an incisional hernia pathogenetic mechanisms of this disease is determined
could occur on a 5-mm incision site, we formulate the by immunologic evaluation, we could exhort surgeons
following hypothesis: The existence of an intrinsic to evaluate carefully the entity of parietal damages
parietal defect is made worse by the trocar’s insertion. because of the trocar’s passing and to repair the iatro-
The creation of an iatrogenic defect, followed by trocar genic defect if required. Moreover, a careful disease’s
movements during surgery, could transform a 5-mm staging should be done to identify the additional
880 Sirito et al
implantations in addition to main ovarian lesion so that References

we could give considerable diagnostic and prognostic
help for the clinical follow-up evaluation of these 1. Lajer H, Widecrantz S, Heisterberg L. Hernias in trocar incisions
following abdominal laparoscopy: a review. Acta Obstet Gynecol
patients. We are waiting for a more efficient compre-
Scand 1997;76:389-93.
hension of the factors that are involved in the disease’s 2. Montz FJ, Holschneider CH, Munro MG. Incision hernia following
development and recognize the possible use of specific laparoscopy: a survey of the American Association of Gynaecologic
drugs. Laparoscopists. Obstet Gynecol 1994;84:881-4.
www.ajog.org
Prenatal diagnosis of amniotic sheets by magnetic

resonance imaging
Kiyoshi Kato, MD,a Tanri Shiozawa, MD,a,* Takashi Ashida, MD,a
Nobuya Unno, MD,b Ikuo Konishi, MDa
Department of Obstetrics and Gynecology, Shinshu University School of Medicinea; Department of

Obstetrics, Center for Perinatal Medicine, Nagano Children’s Hospital, Nagano, Japanb
KEY WORDS We report prenatal diagnosis of an amniotic sheet by magnetic resonance imaging (MRI) at 28
Amniotic sheets weeks’ gestation. The amniotic sheet divided the uterine cavity into 2 compartments, with the
Uterine synechia fetus and placenta located in the upper portion of the amniotic cavity. Although prenatal
Pregnancy diagnosis of amniotic sheets can be performed with ultrasonography, the wider field of view and
Magnetic resonance excellent tissue contrast provided by MRI may allow better spatial visualization of the amniotic
imaging (MRI) sheet and, therefore, improve the diagnostic accuracy.
Ultrasound Ó 2005 Mosby, Inc. All rights reserved.
The amniotic sheets, shelf-like structures in the uterus Case report

originated from ‘‘wrapping’’ of the amniochorionic
membrane over a uterine synechia, have been mainly A 27-year-old woman, gravid 1, para 1, was referred to
diagnosed using ultrasonography during pregnancy.1,2 our hospital at 28 weeks of gestation for a work-up and
Because amniotic sheets are benign, the accurate prena- the management of abnormal placentation. Her previ-
tal diagnosis, including differential diagnosis with other ous pregnancy had been uneventful, but she received a
disorders such as amniotic band syndrome, is important postpartum curettage because of an adherent placenta.
to avoid unnecessary medical interventions.1 Magnetic The amount of menstrual flow after the first pregnancy
resonance imaging (MRI) features the wider field of was normal. The present pregnancy was confirmed at
view and excellent tissue contrast, and the use of MRI in a previous clinic at 7 weeks’ gestation (Figure 1A). At
combination with ultrasound makes the prenatal diag- 13 weeks’ gestation, ultrasonography showed protruded
nosis more accurate. In the present study, we report the anterior and posterior walls with a normal fetal heart
prenatal diagnosis of an amniotic sheet by MRI. beat and normal fetal growth (Figure 1B). At 22 weeks’
gestation, ultrasonography revealed the presence of
another amniotic cavity with polyhydramnios beneath
the ‘‘original’’ amniotic cavity, which carried a normal-
looking fetus (Figure 1C). She was then referred to our
* Reprint requests: Tanri Shiozawa, MD, Department of Obstet-
rics and Gynecology, Shinshu University School of Medicine, 3-1-1,
hospital at 28 weeks’ gestation. On vaginal examination,
Asahi Matsumoto 390-8621, Japan. the cervix was dilated 3 cm and 70% effaced. Because
E-mail: tanri@hsp.md.shinshu-u.ac.jp she complained of frequent uterine contractions, an
doi:10.1016/j.ajog.2005.02.111
882 Kato et al
the septum-like membrane (Figure 2). There were no

direct connections between the fetus and septum-like
structure. Although the edge of septum was not evident,
we concluded that the septum was an amniotic sheet
secondary to uterine synechia with a part of placenta on
it. Based on these findings, the patient was diagnosed as
having a third-trimester pregnancy complicated by an
amniotic sheet.
An ultrasound scan showed that the fetus was in the
left acromiodorsoanterior position with remarkable
polyhydramnios. The fetal growth was normal with no
apparent gross anomalies. The pulsed and color Dop-
pler imaging demonstrated a fine blood flow between the
posterior wall and placenta. On the second day in
hospital, an administration of MgSO4 was commenced
in addition to ritodrine because of the increased fre-
quency of uterine contractions. At 34 weeks’ gestation,
uterine contraction was further augmented with a com-
plaint of severe abdominal fullness, and the forebag
eventually protruded into the vaginal cavity. An ultra-
sound examination at 34 weeks’ gestation indicated that
the estimated fetal body weight was 2721 g with no
detectable anomalies, and there were no apparent mor-
phologic changes in the septum-like structure. Because
of the possibility of abruption or umbilical cord pro-
lapse, a cesarean delivery was performed at 35 weeks’
gestation and a live 2744 g male with Apgar scores of 8/9
at 1/5 minutes was delivered. The amount of amniotic
fluid was approximately 1000 mL. After the delivery, a
membranous septum was found in the right side of the
uterine cavity, and the placenta occupied the lower part
of the upper lumen. The backside of the septum was
whitish and thickened with a fibrous texture. There was
a route between the upper and lower cavities on the left
side of the placenta. The placenta was spontaneously
removed from the anterior wall with the septum; how-
ever, the placenta attachment on the posterior wall was
removed manually without notable bleeding. Pathologic
examination revealed that the septum was made of a
thin layer of myometrium containing invading tropho-
blasts. The patient’s puerperal course was uneventful.
Figure 1 Ultrasonography of the uterus. (A) Five weeks’
gestation. No abnormal findings were noted. (B) Thirteen
weeks. Parts protruding into the uterine cavity on the anterior
and posterior walls are noted. (C) Twenty-two weeks. An Comment
abnormal placenta (long arrow) is observed along with a
membraneous structure (short arrow). In general, ultrasonography is suitable for the observa-
tion of localized areas of the uterus or fetus. MRI is
more fitted to understand whole intrauterine structures
intravenous administration of ritodrin hydrochloride by the 3-dimensional observation, and to observe the
was started. MRI (1.5-T scanner system, Gyroscan posterior side of uterine cavity, especially in the third
Intera T15, Philips) showed that the uterine cavity had trimester, when the ultrasound scanning is hampered by
2 amniotic cavities, ie, upper and lower, which were the fetus. Therefore, we consider that the use of MRI in
interconnected. Between the 2 amniotic cavities a septum- combination with ultrasonography can be a potent
like structure was noted. The fetus and placenta were strategy in the diagnosis of amniotic sheets.
located in the upper amniotic cavity, and the placenta The MRI scan of the present case was peculiar, ie, the
attached from the anterior through posterior wall across uterine cavity was divided into 2 components separated
Kato et al 883
Figure 2 Magnetic resonance images at 28 weeks’ gestation showing 2 cavities in the uterus. Short arrowheads indicate the
placenta and arrows indicate the amniotic sheet. (A) Sagittal section. The placenta locates from the fundus, through the uterine
anterior wall, to the posterior wall across the amniotic sheet. The thick septum-like structure is made of the amniotic sheet (lower
side) with a part of placenta on it. *Fetal head. (B) Coronal section. The fetus is in the left acromiodorsoanterior position with
remarkable polyhydramnios. A part of placenta is attached on the amniotic sheet, and a fetal hand is ‘‘pushing’’ the amniotic sheet.
by the amniotic sheets with a part of placenta. Korbin fore, we prepared an autologous blood transfusion before
et al reported that the placenta implants around the the operation. With regard to the mode of delivery, we
synechia were observed in approximately one fourth of selected an elective cesarean delivery because of the risk of
the cases.3 The most important abnormality for differ- placental abruption and umbilical cord prolapse during
ential diagnosis is amniotic band syndrome. Amniotic labor.
band syndrome was excluded by the lack of direct In conclusion, we reported a case with an amniotic
connections between the septum and fetus, and the lack sheet in the third trimester detected by MRI. It is very
of apparent fetal anomalies.4 We also suspected congen- important that the amniotic sheets not be confused with
ital uterine anomalies, such as a septate uterus. However, the bands of amniotic band syndrome; while amniotic
a septate uterus was not likely because the septum in a band syndrome is associated with multiple fetal anom-
septate uterus is usually in the fundus and oriented in a alies, amniotic sheets appear to be a totally benign
sagittal plane, whereas the septum in the present case was process unassociated with fetal deformation. Misinter-
located midportion of the uterus with an axial plane.4 pretation of the finding of amniotic sheets could lead to
Therefore, to our knowledge, this is the first report of unwarranted abortion of normal fetuses. Therefore, the
MRI detecting an amniotic sheet in the third trimester of use of MRI, in addition to ultrasonography, can im-
pregnancy. The pathogenetic process of the present case prove the diagnostic accuracy in earlier gestational
is totally unknown. However, based on Figures 1 and 2, weeks and, thus, contribute for proper managements
we speculate that this rare situation was caused by the of amniotic sheets secondary to uterine synechia.
implantation of an ovum near the adhesion site, and the
zygote grew encompassing the adhesion site without
disruption of the string-like septum.
References
Uterine synechia (Asherman’s syndrome) during preg- 1. Randel SB, Filly RA, Callen PW, Anderson RL, Golbus MS.
nancy has been reported to be associated with an adherent Amniotic sheets. Radiology 1988;166:633-6.
placenta. Jewelewicz et al reported that the ratio of 2. Finberg HJ. Uterine synechiae in pregnancy: expanded criteria for
placenta accreta in patients with uterine synechia was recognition and clinical significance in 28 cases. J Ultrasound Med
1991;10:547-55.
9%.5 In the present case, we also predicted the presence 3. Korbin CD, Benson CB, Doubilet PM. Placental implantation on
of an adherent placenta because we observed blood flow the amniotic sheet: effect of pregnancy outcome. Radiology 1998;
between the posterior uterine wall and placenta. There- 206:773-5.
884 Kato et al
4. Brown DL, Felker RE, Emerson DS. Intrauterine shelves in 5. Jewelewicz R, Khalaf S, Neuwirth RS, Vande Wiele RL. Obstetric
pregnancy: sonographic observations. Am J Roentgenol 1989;153: complications after treatment of intrauterine synechiae (Asherman’s
821-4. syndrome). Obstet Gynecol 1976;47:701-5.
www.ajog.org
Posterior reversible leukoencephalopathy

in a case of postpartum eclampsia
Maryam Parisaei, MRCOG,a Iris Derwig, MBBS,a Jeannie Yoon, MRCOG,a
Katrina J. Erskine, MRCOG,a Paul R. Jarman, PhDa,b
Department of Obstetrics and Gynaecology, Homerton University Hospital a; National Hospital

for Neurology and Neurosurgery, Queen Square,b London, United Kingdom
KEY WORDS This case report describes an atypical presentation of eclampsia. A 26-year-old lady presented
Late onset eclampsia 5 days’ postpartum with a series of grand mal seizures after an uneventful pregnancy and delivery.
Posterior reversible An MRI scan of the brain showed areas of low signal involving cerebral white matter and right
leukoencephalopathy cerebellum. Within 2 weeks, all symptoms and radiologic abnormalities had resolved.
Postpartum headache Ó 2005 Mosby, Inc. All rights reserved.
Case report blood pressure was noted to be 159/86 mm Hg com-

pared with her booking blood pressure of 110/70 mm
A 26-year-old Bengali lady living in the UK booked Hg. Urinalysis revealed a trace of protein. A full blood
routinely at 12 weeks of pregnancy. She had no significant count showed no major abnormalities, with a platelet
medical or family history, and was not taking any regular count of 282. Her electrolytes were normal. Aspartase
medication. A previous pregnancy in 1996 was unevent- transaminase and uric acid levels were marginally ele-
ful, and resulted in the spontaneous vaginal delivery of vated at 48 and 0.39, respectively. Detailed neurologic
a baby girl. During this pregnancy her antenatal period examination showed no abnormalities. A CT scan of the
was also uneventful. At 38 weeks’ gestation, labor was head was performed immediately after admission and
induced 36 hours after prelabor spontaneous rupture was normal. Lumbar puncture showed an elevated
of membranes. After a normal vaginal delivery, she opening pressure at 40 cm. Protein was elevated at
was discharged home the following day. She had been 0.94 g/L with other CSF constituents normal.
normotensive throughout the pregnancy, and before She was started on magnesium sulphate, given low-
discharge her blood pressure was 110/75 mm Hg. molecular-weight heparin and high-dose acyclovir. The
Five days later she presented to the accident and differential diagnosis at this stage was late onset ec-
emergency department feeling generally unwell, com- lampsia, venous sinus thrombosis, and encephalitis. She
plaining of a sudden onset of severe headache and had multiple grand mal seizures over the next 3 days and
flashing lights. On admission, she had a generalized was started on phenytoin. An MRI scan of the brain 6
grand mal seizure, which terminated spontaneously. Her days after presentation showed confluent areas of high
signal on FLAIR sequences, and low signal on T1-
weighted images involving the cerebral white matter and
Reprints not available from the authors. superior part of the right cerebellum (Figure). Within a
doi:10.1016/j.ajog.2005.02.077
886 Parisaei et al
Postpartum eclampsia is categorized as early (!48

hours) or late (O48 hours) eclampsia. A recent study
revealed that up to 33% of eclampsia occurred postpar-
tum, with 79% of these arising 48 hours after delivery.
Interestingly, only 22% of those with late onset eclamp-
sia had been previously diagnosed with preeclampsia.
The majority had at least 1 prodromal symptom, with
headache and epigastric pain being the most common.1
In cases of late onset postpartum eclampsia edema,
proteinuria and hypertension are often not present until
just before the onset of seizures, making early diagnosis
difficult.2
Posterior reversible encephalopathy is an infrequent
diagnosis in day-to-day obstetric care but is known to be
part of a spectrum of conditions associated with ec-
lampsia. It is characterized by headache, altered mental
status, cortical blindness, seizures, and a characteristic
MRI picture showing high signal lesions in the posterior
cerebral white matter.3 Although the name suggests that
it is usually reversible, early diagnosis and treatment is
essential because irreversible neurologic deficits or death
may occur. Treatment involves lowering mean arterial
blood pressure to below 125 mm Hg and anticonvulsant
therapy. The syndrome is not unique to pregnancy. A
Figure Axial T1-weighted MRI image showing patchy high similar clinical and radiologic picture may occur in a
signal change in the cerebellum. number of conditions, including malignant hypertension
and cyclosporine toxicity, and is believed to be caused
by a loss of integrity of the blood brain barrier and
few days her fits had stopped. A subsequent repeat MRI vascular leaks.4
of the brain showed a complete resolution of the It is crucial to consider late onset postpartum ec-
previous abnormalities. A repeat lumbar puncture was lampsia in any woman who presents after delivery with
also normal with negative oligoclonal bands. All other severe headache, epigastric pain, or visual disturbance,
microbiological, autoimmune, and thrombophilia stud- even in those without a previous diagnosis of pree-
ies showed no abnormalities. Two weeks after her first clampsia. Eclampsia will continue to present us with
grand mal seizure she had made a full recovery with no diagnostic challenges, and efforts need to be directed to
neurologic sequelae, and was discharged from hospital. educate patients and health care workers to report
A diagnosis of reversible posterior leukoencephalopathy symptoms early and to investigate them appropriately.
syndrome was made. She has since had no recurrence of
her seizures.
References
Comment
1. Chames MC, Livingstone JC, Ivester TS, Barton JR, Sibai BM.
Reversible posterior leukoencephalopathy syndrome is a Late postpartum eclampsia: a preventable disease? Am J Obstet
Gynecol 2002;186:1174-7.
condition seldom diagnosed in women with eclampsia.
2. Veltkamp R, Kupsch A, Polasek J, Yousry TA, Pfister HW. Late
In this case, eclampsia occurred 5 days after delivery onset postpartum eclampsia without pre-eclamptic prodromi: clin-
with no prior warning. Eclampsia continues to be a ical and neurological presentation in two patients. J Neurol
poorly understood condition associated with significant Neurosurg Psychiatry 2000;69:824-7.
mortality and morbidity. The neurologic signs vary from 3. Servillo G, Striano P, Striano S, Tortora F, Boccella P, De Robertis
seizures to altered states of consciousness, including E, et al. Posterior reversible encephalopathy syndrome in critically
ill obstetric patients. Intensive Care Med 2003;29:2323-6.
coma. It is thought that eclampsia results from the 4. Hinchley J, Chaves C, Appignani B, Breen J, Pao L, Wang A, et al.
loss of autoregulation of cerebral blood flow, causing A reversible posterior leukoencephalopathy syndrome. N Engl J
cerebral edema and hemorrhage. Med 1996;334:494-500.
www.ajog.org
Twin-to-twin transfusion syndrome at

11 weeks of gestation
Marieke Sueters, MD,a Johanna M. Middeldorp, MD,a Dick Oepkes, PhD,a
Enrico Lopriore, MD,b Frank P. H. A. Vandenbussche, PhDa,*
Departments of Obstetricsa and Pediatrics,b Leiden University Medical Center, Leiden, The Netherlands
KEY WORDS Second-trimester twin-to-twin transfusion is well known, but first-trimester cases rarely have
Monochorionic twins been described. We present the case of a monochorionic twin at 11 C 0 weeks of gestation with
Twin-to-twin single increased nuchal translucency and normal karyotypes. At 12 C 5 weeks of gestation,
transfusion double intrauterine death was diagnosed, followed by delivery of a strikingly red and
syndrome white fetus.
First trimester Ó 2005 Mosby, Inc. All rights reserved.
Ultrasound
A 33-year-old woman, gravida 3, visited our unit at At 12C5 weeks of gestation, double intrauterine
11 C 0 weeks of gestation for ultrasound examination. death was diagnosed. At that time, the crown-rump
A monochorionic-diamniotic twin pregnancy was seen. length of twin A and twin B both measured 6.56 cm.
Both fetuses showed a crown-rump length appropriate Twin B’s nuchal translucency was still increased, and the
for gestational age (twin A, 4.71 cm; twin B, 4.70 cm) amniotic fluid was normal for both. We sampled amni-
and a normal amount of amniotic fluid. Nuchal trans- otic fluid from each sac, which revealed normal female
lucency thickness in fetus A was normal (0.16 cm); karyotypes.
nuchal translucency of fetus B was markedly increased Two days later, a very pale fetus and a dark red fetus,
(0.68 cm, far above the 95th percentile). No other fetal both with a body weight of 20 g, were delivered (Figure).
anomalies were seen. There were no anatomic or Dop- Because of a retained placenta, dilatation and curettage
pler signs that were suggestive of impending cardiac was performed. Therefore, dye injection into chorionic
failure. The differential diagnosis included a chromo- vessels could not be performed. Autopsy was refused.
somal defect in 1 or both twins, cardiac abnormality of
twin B (not yet visible sonographically), and extreme
early onset twin-to-twin transfusion syndrome (TTTS). Comment
Chronic TTTS develops as a result of a small and
chronically unbalanced blood flow between monochori-
* Reprints not available from authors. Address correspondence to onic twins through vascular anastomoses on the pla-
F. P. H. A. Vandenbussche, MD, PhD, Department of Obstetrics, K6-
37, Leiden University Medical Center, PO Box 9600, 2300 RC Leiden,
centa. The diagnosis is based on ultrasonography
The Netherlands. showing a combination of polyuric polyhydramnios in
E-mail: frank.vandenbussche@lumc.nl the recipient’s sac and oliguric oligohydramnios in the
doi:10.1016/j.ajog.2005.02.095
888 Sueters et al
the first trimester. Nuchal translucency thickness tends

to normalize after 14 weeks of gestation. Fetal urine
production then commences. We speculate that only
then does volume overload in the recipient result in the
excessive urine production that causes polyhydramnios,
which is the main indicator of second-trimester TTTS.
Because fetal diuresis contributes little to the amniotic
fluid in the first trimester, it seems plausible that the
classic oligo/polyhydramnios sequence cannot yet occur.
However, as possibly happened in this case, severe first-
trimester TTTS may worsen so rapidly that single or
double fetal death occurs before the second trimester is
reached. Benirschke and Masliah2 described the 2 ear-
liest cases of first-trimester TTTS. Those were strongly
suggestive of such long-lasting circulatory imbalance,
as was revealed by the high cardiac-to-body ratio in
recipient twins. Another possibility is that fetus B had a
lethal malformation. It is well known that severe cardiac
and other malformations can cause subcutaneous edema
in the nuchal region. Fetal anomalies were not seen on
ultrasound, but autopsy was not performed. Thus, both
early chronic TTTS and acute TTTS after single fetal
dead should be considered here. Whether severe first-
trimester TTTS or a lethal malformation of fetus B was
the primary cause of intrauterine death will remain
unknown. Either way, inter-twin vascular connections,
which are present invariably in all monochorionic
placentas, enabled the second twin to exsanguinate
Figure Picture taken after the delivery of the very pale and
into the dead fetus, which caused the co-twin death
dark red fetus after double intrauterine death at 12 C 5 weeks
of gestation.
because of acute TTTS. This would imply, in our case,
the red twin to be the fetus with increased nuchal
translucency who died first, and the pale twin to be
donor’s sac. Quintero et al developed a staging system the co-twin who died consequently through exsangui-
to determine the severity of the circulatory imbalance nation.
between the twins.1 Thus defined, TTTS complicates In conclusion, TTTS can be seen in various ways at
approximately 15% of monochorionic-diamniotic twin different gestational ages. Besides the well-known risks
pregnancies in the second and early third trimester. of severe second-trimester TTTS, we believe that TTTS
Acute TTTS can occur after the intrauterine death of can cause fetal death or neurologic damage,4 even in the
one twin and involves a sudden loss of a substantial first trimester of pregnancy. The only presenting symp-
amount of blood. The survivor then exsanguinates tom may be a single increased nuchal translucency.
through vascular anastomoses, presumably an arterio-
arterial one, into the low-pressure circulation of the
dying or dead twin, followed by hypovolemia, ischemic References
brain injury, or death.
In our patient, one of the earliest TTTS cases to be 1. Quintero RA, Morales WJ, Allen MH, Bornick PW, Johnson PK,
reported thus far,2 the first and only symptom was an Kruger M. Staging of twin-twin transfusion syndrome. J Perinatol
1999;19:550-5.
increased nuchal translucency in 1 twin. Although this 2. Benischke K, Masliah E. The placenta in multiple pregnancy:
finding is related strongly to chromosomal defects, both outstanding issues. Reprod Fertil Dev 2001;13:615-22.
twins had normal karyotypes. Increased nuchal trans- 3. Sebire NJ, Souka A, Skentou H, Geerts L, Nicolaides KH. Early
lucency has been reported previously to be associated prediction of severe twin-to-twin transfusion syndrome. Hum
with an increased risk (odds ratio, 3.5) of the subsequent Reprod 2000;15:2008-10.
4. Weiss JL, Cleary-Goldman J, Tanji K, Budorick N, D’Alton ME.
development of TTTS.3 Sebire et al3 suggest that the Multicystic encephalomalacia after first-trimester intrauterine fetal
recipients’ hypervolemia causes heart failure with a death in monochorionic twins. Am J Obstet Gynecol 2004;190:
subsequent accumulation of fluid behind the neck in 563-5.
www.ajog.org
Spontaneous closure of the hymen during pregnancy

M. A. Onan, MD,a,* A. B. Turp, MD,a C. Taskiran, MD,a C. Ozogul, MD,b
O. Himmetoglu, MDa
Departments of Obstetrics and Gynecologya and Histology and Embryology,b Gazi University Faculty of Medicine,
Besevler, Ankara, Turkey
Received for publication December 2, 2004; revised February 21, 2005; accepted March 7, 2005
KEY WORDS Imperforate hymen is a rare disorder that is usually discovered at the onset of menstruation. In
Spontaneous closure the literature, secondary closure of the hymen has been reported in 2 cases, both of which
of hymen occurred subsequent to surgical procedures that involved the hymen. We report an interesting
Pregnancy case of the spontaneous formation of an imperforate hymen during pregnancy in the absence of
previous surgical procedures. Electron microscopic findings indicate hymenal tissue reorganiza-
tion.
The hymen marks the distal most extent of the week fetus that was concordant to the gestational age.
vagina. Imperforate hymen is a rare genital disorder, An amniotic index of 250 mm indicated hydramnios.
which is diagnosed generally during adolescence conse- Sonography revealed funneling, and a dense fluid col-
quent to hematocolpos. The case presented here, how- lection in the vagina was seen. A surgical hymenotomy
ever, describes spontaneous closure of hymen during was performed. A stellate incision was made with local
pregnancy. anesthesia. On incision, purulent material was dis-
charged from vagina. A sample was taken for microbi-
ologic assay. In the pelvic examination, the vagina and
Case report
cervix were found to be structurally normal with a
A 23-year-old pregnant woman was seen at our clinic cervical dilation of 4 cm and 80% effacement with
with uterine contractions. On initial examination, regu- bulging amnion. Despite attempted tocolysis, a 2350-g
lar intensive contractions, with a duration of 30 seconds, male fetus was delivered with an Apgar score of 2/4 at
were observed. Vaginal examination was not possible; 1 and 5 minutes. The fetus had no spontaneous respi-
however, a fully intact and bulging hymen with no ration and a heart rate of 60 beats/minute and was
papillomatosis was noted (Figure 1). She immediately cyanotic. Intubation was performed promptly by the
was evaluated by ultrasonography, which showed a 30- pediatricians. Physical examination of the newborn
infant revealed no esophageal atresia. The baby was
transferred to the intensive care unit. After a few hours,
the baby died of respiratory distress.
* Reprint requests: M. A. Onan, MD, Gazi University Faculty of
Medicine, Department of Obstetrics and Gynecology, Besevler,
In her detailed history, we learned that menstruation
Ankara, Turkey. began normally without any problems. She had no
E-mail: maonan@gazi.edu.tr complaint related with menstrual activity. She also
doi:10.1016/j.ajog.2005.03.025
890 Onan et al
Figure 1 Intact hymen before the surgical intervention.
experienced no coital abnormalities after her marriage.

She had coital activity until she learned that she was
pregnant.
A hymenal biopsy specimen was taken during deliv-
ery to evaluate the ultrastructure of hymen by electron
microscopy (Carl Zeiss EM 900; Zeiss, Jena, Germany).
The ultrastructure of the hymen showed metabolically
active connective tissue cells. Mitotic figures were espe-
cially evident in the basal epithelial cells. The connective
tissue cells contained abundant rough endoplasmic
reticulum (RER) and ribosomes that indicated cell
activity. Increased vascularization and widespread dis-
tribution of collagen fibers under epithelial tissue were
strong evidence of accelerated hymenal tissue reorgani- Figure 2 A, Normal appearance of basal cells of epithelial
zation (Figure 2, A and B). tissue, mitochondrial crystalysis observed (thin arrows), nor-
Although she had not had intercourse, the surgically mal appearance of basal membrane (thick arrows), and con-
created hymenal orifice was still patent at postpartum nective tissue rich in collagen (star). (Electron microscopy,
week 8. The couple began coital activity and have not uranyl acetate-lead citrate; original magnification, !3000.) B,
Active rough endoplasmic reticulum (RER) cisterns of basal
reported any coital or menstrual problem until now.
cells (thick arrows); ribosomes in cytoplasm (thin arrows).
(Electron microscopy, uranyl acetate-lead citrate; original
magnification, !7000.)
Comment
The hymen marks the distal most extent of the vagina lower genital tract abnormalities. None of these prob-
and the most proximal boundary of the vulvar vestibule. lems were associated with the present case. Secondary
It may be imperforate, round, annular, septate, cribri- closure of hymen has been reported in 2 cases.1,2
form, or parous. On the vaginal surface, there is a Berkowitz et al1 documented a 5-year-old abused child
nonkeratinized stratified squamous epithelium, which is with distortion of hymen, laceration of perineal body,
more or less glycogenated in response to estrogen and loss of normal anal tone. Seven months after
exposure, as seen in women of reproductive age, new- surgical correction, the hymen was replaced by a thick
born infants, pregnant women, and postmenopausal and opaque scar with no orifice. Chao-Hsi and Ching-
women who receive estrogen therapy. On the vulvar Chung2 reported a 32-year-old pregnant woman who
surface, the vestibular epithelium appears similar to the experienced closure of the hymen during week 37 of
vaginal epithelium in women of reproductive age. gestation of her second pregnancy. The baby was
Imperforated hymen is a rare genital disorder. Gen- delivered by cesarean delivery, but a thick fibrotic
erally, it is diagnosed during adolescence consequent to membrane at the vaginal orifice blocked the passage of
hematocolpos. These patients may demonstrate other lochia. Therefore, a surgical incision was made to enable
Onan et al 891
drainage. She had been subjected to a hymenotomy at microscopy. The connective tissue cells demonstrated
age 13 years because of imperforate hymen and hema- markedly increased rough endoplasmic reticulum and
tocolpometra. At age 18 years, after cessation of normal free ribosomes. Between these cells, vascularization and
menstruation because of significant weight loss, a thin connective tissue collagen fibers were obvious. These
membrane was observed and excised surgically. No ultrastructural appearances indicate reorganized hy-
problem occurred during her first pregnancy. Although menal tissue.
spontaneous closure of a hymen-like structure occurred
in these 2 cases, there were mitigating circumstances that
suggest the formations were related to extensive scar References
formation. The current case, by contrast, involves
spontaneous closure of the hymen in the absence of 1. Berkowitz CD, Evlik SL, Logan M. A simulated ‘‘acquired’’
previous surgical procedures. In our opinion, after the imperforate hymen following genital trauma of sexual abuse. Clin
Pediatr (Phila) 1987;26:307-9.
cessation of coital activity, a healing process that was 2. Chao-Hsi L, Ching-Chung L. Hymen re-formation after hymenot-
possibly related to the pregnancy may explain the omy associated with pregnancy. Aust N Z J Obstet Gynecol
closure. Active growth was demonstrated by electron 2002;42:559-60.
www.ajog.org
Small bowel obstruction due to adhesive disease observed

after uterine fibroid embolization
Jay Goldberg, MD, MS,a,* Kristyne Boyle, MD,a Monica Choi, CRNP, PhD,a
Narhari Panchal, MDb
Department of Obstetrics and Gynecology, Jefferson Medical College,a Philadelphia, PA; Department of Surgery,
SUNY Buffalo,b Buffalo, NY
Received for publication December 29, 2004; accepted February 21, 2005
KEY WORDS After uterine fibroid embolization (UFE), the development of intra-abdominal adhesions,
Uterine fibroid especially those involving the bowel, is a very rare complication. Seven months after UFE, a
embolization patient had a complete small bowel obstruction develop that was caused by an adhesive band
Adhesion between the posterior fibroid and cul-de-sac. She underwent an exploratory laparotomy, lysis of
Bowel obstruction adhesion, and myomectomy. No bowel resection was needed. Inflammation after UFE may cause
the development of intraperitoneal adhesions. We report an unanticipated case of a complete
small bowel obstruction caused by an adhesion observed after UFE.
Uterine fibroid embolization (UFE) is an increasingly examination. She denied a history of prior surgeries,
used, effective, and minimally invasive treatment for pelvic pain, endometriosis, or pelvic infections. A pelvic
symptomatic leiomyoma. UFE is a treatment alternative ultrasound verified a uterus measuring 13.9 ! 10.8 !
to hysterectomy or myomectomy, 2 surgeries with high 13.5 cm with multiple myomas, the largest, 9 ! 8 ! 6.5
postsurgical rates of adhesive disease, often affecting the cm, arising posteriorly. She elected to undergo a uterine
bowel. The development of adhesions, especially those fibroid embolization performed via a 4-French sheath in
involving the bowel, after UFE is a very rare complica- the right common femoral artery. Coaxial catheteriza-
tion.1,2 We present an unusual case of a delayed complete tion with a 3-French catheter was used to deliver 500 to
small bowel obstruction caused by adhesions observed 700 mm Embospheres (Biosphere Medical, Inc, Rock-
after UFE. land, Mass) selectively into both right and left uterine
arteries under fluoroscopic guidance until near-stasis of
Case report flow was achieved. Both embolizations were performed
with the catheter tip in the transverse portion of the
A 32-year-old nulligravid woman with menorrhagia and uterine artery. No reflux of particles out of either
anemia had a fibroid uterus diagnosed on clinical uterine artery was observed. The patient had no acute
postoperative complications and returned to work in
approximately one week. Her prior symptoms soon
* Reprint requests: Jay Goldberg, MD, MS, Department of
Obstetrics and Gynecology, Jefferson Medical College, 834 Chestnut
thereafter improved.
St, Suite 400, Philadelphia, PA 19107. Seven months after the UFE, she experienced an
E-mail: jaygoldbergmd@yahoo.com acute onset of crampy abdominal pain, accompanied by
doi:10.1016/j.ajog.2005.02.133
Goldberg et al 893
nausea and vomiting. In an emergency department, her dures, which could have also been an cause of her
abdominal examination revealed diffuse tenderness and adhesive disease.
rebound. A computed tomographic scan had findings The proposed mechanism of action of adhesion
consistent with a distal complete small bowel obstruc- formation in our case is similar to that encountered in
tion. An exploratory laparotomy found a dense adhesive surgical patients. The large posterior fibroid, in direct
band running between a 2-cm subserosal posterior contact with the cul-de-sac, underwent devasculariza-
fibroid and the cul-de-sac, trapping a dilated loop of tion and ischemia after UFE. The resulting inflamma-
small bowel. The location of this fibroid corresponded tory reaction set off a cascade of intra-abdominal
with the preembolization 9 ! 8 ! 6.5-cm fibroid. No adhesion formation between the peritoneal surfaces.
other abdominal disease, including other adhesions or Inflammation of the adjacent peritoneal surfaces initi-
endometriosis, was visualized. The adhesion was lysed, ated adhesion formation with the formation of a fibrin
releasing the obstructed loop of bowel. This posterior matrix in the presence of suppressed fibrinolysis. Local
fibroid was then removed. A careful inspection of the ischemia allowed persistence of the fibrin matrix. Vas-
previously obstructed area of small bowel revealed no cular granulation tissue, containing fibroblasts, macro-
discoloration or nonviable areas, allowing the avoidance phages, and giant cells, then gradually replaced the
of a resection. Postoperatively, the patient did well. matrix. As the adhesion band slowly matured, it was
After slowly advancing her diet, she was discharged on covered by mesothelium and connective tissue fibers.
the fourth postoperative day. Although it is possible that the adhesion could have
preceded the UFE, given her lack of previous intra-
abdominal surgeries, or any evidence of prior pelvic
Comment infections or endometriosis, our patient’s adhesion and
subsequent complete bowel obstruction is most plausi-
A MEDLINE search using the MeSH terms uterine bly a result of the UFE. Although adhesion formation
artery embolization, uterine fibroid embolization, adhe- is a common occurrence after myomectomy and hys-
sion, and bowel obstruction, identified only 2 articles that terectomy, it may also occur when fibroid disease is
potentially attributed adhesion formation as a result of treated with UFE, which may subsequently lead to a
UFE.1,2 One of these articles also reported a partial small bowel obstruction, in this case 7 months post-
small bowel obstruction caused by adhesive disease.2 embolization. In patients presenting with abdominal
Intrauterine adhesions causing infertility were re- pain who have previously undergone UFE, small
ported by Honda et al1 in 4 patients after UFE. In their bowel obstruction should be included in the differential
discussion, a hypothesized etiology of adhesion forma- diagnosis. In our case, prompt recognition of and
tion was infection and inflammation resulting from intervention for the obstruction prevented the need for
sloughed necrotic fibroids obstructing the cervix and a small bowel resection.
interfering with the passage of intrauterine discharge.1
They did not report any intra-abdominal adhesions in
their series. References
Payne and Haney2 reported a patient who had a
partial small bowel obstruction develop 2 weeks after 1. Honda I, Sato T, Adachi H, Kobayashi Y, Shimada K, Watanabe
UFE. On laparotomy, the patient underwent extensive H, et al. Embolization for leiomyoma: complications and effects on
fertility. [Japanese] Nippon Igaku Hoshasen Gakkai Zasshi
adhesiolysis, including separating small bowel adherent
2003;63:294-302.
to the uterus. Also found was a large ovarian adeno- 2. Payne JF, Haney AF. Serious complications of uterine artery
carcinoma. No mention was made of whether the embolization for conservative treatment of fibroids. Fertil Steril
patient had previously undergone any surgical proce- 2003;79:128-31.
www.ajog.org
LETTERS TO THE EDITORS
Selective fetocide reverses preeclampsia

in discordant twins
To the Editors: We read with interest the article by anism is consistent with the historic observations of res-
Heyborne and Porreco,1 who reported on 2 cases of olution of preeclampsia after spontaneous death of
resolution of preeclampsia after selective fetocide in 1 twin. We believe that the interruption of placental
discordant twin pregnancies. We recently reported on a blood flow may act as a mechanism of placental exclu-
similar case,2 in which a 32-year-old nulliparous woman sion, thus leading to decrease or arrest of the release
with a dichorionic discordant twin pregnancy had pree- of the substances that are involved in the maintenance
clampsia at 28 weeks of gestation. Twin A was developing of the disease. These observations open prospects
normally; twin B showed severe intrauterine growth re- for further research on the mechanisms of this complex
striction. Expectant management was favored initially, disease and to offer an acceptable alternative therapy in
but the maternal condition worsened, with hypertension selected cases.
necessitating high doses of nifedipine and labetalol and
proteinuria increasing up to 6.5 g per day. Selective feto- Francois Audibert, MD
cide of the compromised twin was presented as an alter- Universite de Montreal
native to delivery, and the patient accepted this option. Hopital Sainte-Justine
After selective termination at 32 weeks of gestation, we Montreal, Quebec, Canada
observed a dramatic improvement in the clinical and bio- E-mail: francois.audibert@montreal.ca
logic condition of the woman, who eventually was deliv-
ered of a healthy infant who weighed 2560 g at 38 weeks Laurent J. Saloman, MD
of gestation and of a stillborn second twin who weighed René Frydman, MD
330 g. The postpartum course was uneventful. Hospital Antoine Beclere
These concordant reports provide further clinical Université Paris XI
evidence that preeclampsia is caused by placental factors. Clamart, France
Heyborne and Porreco1 hypothesize that placental invo-
lution might be necessary for the resolution of the dis-
ease. However, placental involution is known to occur References
after several weeks, whereas preeclampsia resolved within
1. Heyborne KD, Porreco RP. Selective fetocide reverses preeclampsia
several days in our patient and in the 2 cases described by in discordant twins. Am J Obstet Gynecol 2004;191:477-80.
the authors. We would suggest rather that the interrup- 2. Audibert F, Salomon LJ, Castaigne-Meary V, Alves K, Frydman R.
tion of placental blood flow in the pathologic placenta Selective termination of a twin pregnancy as a treatment of severe
may have been enough to reverse the disease. This mech- pre-eclampsia. BJOG 2003;110:68-9.
doi:10.1016/j.ajog.2005.02.091
Reply
To the Editors: We thank Audibert et al for their failing to reference their case report that was published
comments regarding our manuscript and apologize for while our manuscript was in preparation. We agree
Letters to the Editors 895
that their similar experience strengthens the link between required. We are not certain of the precise meaning
a pathologic placenta and the disease process that we of placental exclusion. Certainly the fetal circulation
recognize as preeclampsia and further supports selective within the placenta ceases immediately with selective
termination as a reasonable management strategy in a termination, but there is no a priori mechanism to im-
few carefully selected cases. mediately ‘‘exclude’’ the placenta from the maternal
Inasmuch as the underlying disease processes that circulation.
results in the clinical syndrome of preeclampsia remain Another potential mechanism for disease resolution is
largely hypothetical, the mechanisms by which selective the removal of the fetal metabolic ‘‘load’’ from the
termination reverses these processes are doubly so. We placenta, which presumably improves the placenta’s
offered placental involution as a potential mechanism ischemic condition. In any case, it seems likely that what-
because, in our cases, the time course of disease resolu- ever mechanism is involved, it does not entirely correct the
tion, which was possibly aided by the administration of underlying disease process immediately but simply reduces
corticosteroids, accorded reasonably well with the it below some threshold beneath which the signs and
course of placental involution that was known to us symptoms of preeclampsia resolve sufficiently to allow
from the somewhat analogous situation of delayed- the safe continuation of the pregnancy.
interval delivery.
Audibert et al suggest a more rapid mechanism, Kent D. Heyborne, MD*
although their paper indicates a 2-week interval for their Richard P. Porreco, MD
patient’s 24-hour urinary protein excretion to diminish Swedish Medical Center
to approximately 2 g, which is an elevated level that Perinatal Center
persisted for the remainder of the pregnancy; and the Englewood, CO 80110
continued administration of antihypertensive agents was *E-mail: kheyborne@msn.com
doi:10.1016/j.ajog.2005.02.092
Is circulating extracellular VEGF increased in

preeclamptic women?
To the Editors: I read with interest the article of before serum is collected reduces the time-dependent,
McKeeman et al.1 Although the authors should be non-uniform release of VEGF4; this approach should
congratulated by their observations on sFlt-1 levels in be chosen when serum VEGF is measured.
normotensive and preeclamptic women, I would like Furthermore, it would be interesting to know whether
to discuss some issues concerning the measurement of the force that was used to centrifuge blood samples
vascular endothelial growth factor (VEGF) concen- and the length of centrifugation were standardized dur-
trations. ing the study period because they may affect platelet
The authors found that there is a trend for serum activation by mechanical stress and, consequently, may
VEGF levels to be higher in preeclamptic, compared have an impact on serum VEGF levels.3
with normotensive, women at 12, 20, and 30 weeks of Considering that serum VEGF concentrations are
gestation and that this increase becomes statistically sig- correlated to platelets counts,4 it would also be interest-
nificant at 37 weeks of gestation and before delivery. ing to evaluate whether the differences in VEGF levels
It is well known that VEGF is stored in a granules of between preeclamptic and normotensive women are
circulating platelets and is released in serum during blood confirmed after correcting for platelet count.
clotting in a time-dependent manner.2,3 Although the au- In light of the fact that serum VEGF concentrations
thors reported that serum was ‘‘separated by centrifuga- reflect blood platelets counts rather than VEGF synthe-
tion after clotting,’’ they did not provide evidence of sis by peripheral tissues,2 it has been suggested that
standardization of the interval between venipuncture plasma and not serum should be used for the determina-
and separation of serum from blood cells; this limit of tion of circulating extracellular VEGF. In plasma, plate-
the study may have contributed to reduce the difference let degranulation is minimized by adding anticoagulants
in VEGF levels between preeclamptic and normotensive to blood samples; CTAD (citrate, theophylline, adeno-
women. Importantly, it has been proved that allowing sine, dipyridamole) tubes should be used for sample
whole blood samples to clot for between 2 and 6 hours collection when plasma VEGF is measured.
896 Letters to the Editors
I believe that a meticulous processing of serum blood is increased throughout gestation in patients who have
samples and the use of plasma may improve the value of preeclampsia develop. Am J Obstet Gynecol 2004;191:
1240-6.
VEGF measurement throughout the whole pregnancy. 2. Webb NJ, Bottomley MJ, Watson CJ, Brenchley PE. Vascular
endothelial growth factor (VEGF) is released from platelets
Simone Ferrero, MD* during blood clotting: implications for measurement of circu-
Department of Obstetrics and Gynaecology lating VEGF levels in clinical disease. Clin Sci (Lond) 1998;94:
San Martino Hospital 395-404.
3. Hormbrey E, Gillespie P, Turner K, Han C, Roberts A,
University of Genoa McGrouther D, et al. A critical review of vascular endothelial
16132 Genoa, Italy growth factor (VEGF) analysis in peripheral blood: is the
*E-mail: simone.ferrero@fastwebnet.it current literature meaningful? Clin Exp Metastasis 2002;19:
651-63.
References 4. Werther K, Christensen IJ, Nielsen HJ. Determination of vascular
endothelial growth factor (VEGF) in circulating blood: significance
1. McKeeman GC, Ardill JE, Caldwell CM, Hunter AJ, McClure N. of VEGF in various leucocytes and platelets. Scand J Clin Lab
Soluble vascular endothelial growth factor receptor-1 (sFlt-1) Invest 2002;62:343-50.
doi:10.1016/j.ajog.2005.02.087
Cytokines, preeclampsia, and uterine denervation?

To the Editors: McKeeman et al1 have found increased associated with different cytokine responses that include
maternal circulating levels of cytokines and their recep- vascular endothelial growth factor.3 In this detailed lon-
tors (vascular endothelial growth factor and vascular gitudinal dataset, does the maternal cytokine response
endothelial growth factor receptor-1) at different gesta- vary with the gestational onset of preeclampsia? Does
tional ages in preeclampsia. Several explanations may it vary with the other possible clinical ‘‘phenotypes’’?
account for these observations that include a prior or Are other cytokines (of neural origin) involved?
continuing injury to the placental bed. Variable uterine
denervation at the deciduomyometrial interface may M. J. Quinn, MD
prevent appropriate endovascular trophoblast invasion Consultant in Obstetrics and Gynaecology
and generate a maternal cytokine response.2 Possible Hope Hospital
sources of regional uterine denervation include curet- Salford, UK M6 8HD
tage, back injury, or, most commonly, a low residue E-mail: quinnobgyn@aol.com
diet that results in sustained constipation. The clinical
‘‘phenotype’’ of the preeclamptic syndrome may vary
with the extent of the injury to the placental bed References
(inversely proportional to gestational age of onset), the 1. McKeeman GC, Ardill JES, Caldwell CM, Hunter AJ, McClure N.
placental site (denervated placental site, recurrent hyper- Soluble vascular endothelial growth factor receptor-1 (sFlt-1) is
tensive syndrome), the concomitant loss of uterine affer- increased throughout gestation in patients who have preeclampsia
ent nerve fibers (intrauterine growth restriction, no develop. Am J Obstet Gynecol 2004;191:1240-7.
2. Quinn MJ. Preeclampsia and partial uterine denervation. Med
maternal hypertension), and the increases in maternal
Hypoth 2005;64:449-54.
systemic vascular compliance (HELLP [hemolysis, ele- 3. Hobson MI, Green CJ, Terenghi G. VEGF enhances intraneural
vated liver enzymes, and low platelet count] syndrome). angiogenesis and improves nerve regeneration after axotomy.
Denervation-reinnervation in other tissues has been J Anat 2000;197:591-605.
doi:10.1016/j.ajog.2005.02.089
Reply
To the Editors: We are pleased with the interest our Quinn for their comments. We hope that this re-
study has received, and we thank Drs Ferrero and sponse will clarify a number of issues regarding the
Letters to the Editors 897
importance of circulating VEGF and sFlt-1 in pree- be avoided because of the binding interactions that can
clampsia. occur between VEGF, sFlt-1, and heparin.5 We agree
First, we would like to address the queries from that it is important to reduce the release of VEGF
Dr Ferrero regarding the treatment of blood samples. from platelets during processing (by implementing strin-
A standard method was implemented for the collection gent sample handling procedures) so that any changes in
of all serum samples throughout the study. After veni- VEGF concentrations are representative of events that
puncture, blood samples were left at room temperature occur in vivo.
for 4 to 6 hours until clotted and serum was separated Second, we find the views of Dr Quinn on the reasons
by centrifugation at 1200g (4(C) for 20 minutes. These for preeclampsia interesting. Our study was an observa-
standardized methods were enforced for each sample. tional study that demonstrated that, at the time of onset,
We did not consider it necessary to correct the VEGF circulating VEGF and sFlt-1 concentrations are in-
data for platelet counts, although this is something creased in a preeclamptic group. Although these cyto-
that could be investigated in future studies. kines may not be the root cause of the syndrome, they
Before this work, a small preliminary study was per- could act as influential mediators. There is not enough
formed to assess the optimum specimen for VEGF mea- data to correlate with gestational age of onset and not
surement throughout pregnancy. VEGF concentrations sufficient patients to correlate for a phenotypic factor.
were analyzed in serum, EDTA plasma and lithium hep- With regard to other cytokines of neural origin, this is
arin plasma collected from both pregnant (n = 4) and beyond the scope of this paper but certainly raises ques-
nonpregnant (n = 4) subjects. There was little variation tions for future research.
in VEGF concentrations that were measured in serum
and plasma from either the pregnant or nonpregnant vol- Gareth C. McKeeman*
unteers. Therefore, on the basis of these data and the fact Joy E. S. Ardill
that the original VEGF radioimmunoassay method was Carolyn M. Caldwell
optimized in serum,1 we opted for the use of serum for Neil McClure
the measurement of VEGF throughout pregnancy. Department of Medicine
Although plasma recently has been suggested as the Queen’s University Belfast
optimum medium for VEGF measurement, the use of se- Belfast, BT12 6BW Northern Ireland
rum is important in cases in which platelets are activated *E-mail: g.mckeeman@qub.ac.uk
and in which platelet-derived VEGF is important.
Because of this, serum has been suggested as being References
the more useful specimen for the assessment of VEGF
1. Anthony FW, Evans PW, Wheeler T, Wood PJ. Variation in de-
in cancer patients.2 In the case of preeclampsia, some tection of VEGF in maternal serum by immunoassay and the possible
women can go on to experience HELLP (hemolysis, ele- influence of binding proteins. Ann Clin Biochem 1997;34:276-80.
vated liver enzymes, and low platelet count) syndrome, 2. Lee J-K, Hong Y-J, Han C-J, Hwang D-Y, Hong S-I. Clinical
which involves platelet activation and the additional re- usefulness of serum and plasma vascular endothelial growth factor
in cancer patients: Which is the optimal specimen? Int J Oncol
lease of VEGF. Therefore, it could be important to assess
2000;17:149-52.
total VEGF concentrations under such circumstances 3. Brockelsby JC, Anthony FW, Johnson IR, Baker PN. The effects
because excess circulating VEGF has been linked to the of vascular endothelial growth factor on endothelial cells: a potential
endothelial cell damage that is seen in these patients.3 role in preeclampsia. Am J Obstet Gynecol 2000;182:176-83.
Furthermore, recent investigations have also confirmed 4. Bussen S, Dietl J. VEGF concentrations in pregnancy [letter]. Am J
the involvement of elevated VEGF serum concentrations Obstet Gynecol 2004;191:1834.
5. Soker S, Goldstaub D, Svahn CM, Vlodavsky I, Levi B-Z,
in the onset of preeclampsia.4 Neufold G. Variations in the size and sulfation of heparin modulate
Although we do not eliminate the use of plasma for the effect of heparin on the binding of VEGF165 to its receptors.
VEGF analysis, the use of heparinized containers should Biochem Biophys Res Commun 1994;203:1339-47.
doi:10.1016/j.ajog.2005.02.090
Accidental fetal lacerations during cesarean delivery:

Experience in an Italian level III university hospital
To the Editors: We read with interest the article by ing cesarean delivery. Several methods are used during
Dessole et al1 concerning accidental fetal laceration dur- cesarean delivery to minimize the risk of fetal accidental
898 Letters to the Editors
laceration occurrence.2 Nonetheless, those procedures Fumikazu Kotsuji, MD, PhD

do not ensure sufficient reliability. We recommend trans- *Department of Obstetrics and Gynecology
abdominal or transvaginal ultrasonographic examina- University of Fukui
tion by the surgeon immediately before the initiation of Matsuoka-cho
the operative procedure. In only a few minutes, ultraso- Yoshida-gun, Fukui 910-1193, Japan
nography can provide important information: the uterine E-mail: kojigyne@fmsrsa.fukui-med.ac.jp
wall thickness, the fetal presentation or orientation, and
the amniotic fluid volume. Those findings can help the
surgeon characterize the inherent risk of fetal laceration References
injury at the cesarean delivery. Careful consideration of
the possibility of accidental fetal laceration at the instant 1. Dessole S, Cosmi E, Balata A, Uras L, Caserta D, Capobianco G,
et al. Accidental fetal lacerations during cesarean delivery: experi-
of uterine incision is the most important safety device.
ence in an Italian level III university hospital. Am J Obstet Gynecol
2004;191:1673-7.
Koji Nishijima, MD* 2. Weiner JJ, Westwood J. Fetal lacerations at cesarean section.
Ken-ichi Shukunami, MD J Obstet Gynecol 2002;22:23-4.
doi:10.1016/j.ajog.2005.02.124
Reply
To the Editors: We appreciate the interest and the nated. Thus, before the performance of cesarean deliv-
comments of Nishijima et al about our article concerning ery, the patient always should be informed about this
accidental fetal lacerations during cesarean delivery.1 risk of accidental fetal lacerations.
Nishijima et al recommend transabdominal or trans-
vaginal ultrasonographic examination by the surgeon Salvatore Dessole, MD*
immediately before the initiation of the operative proce- Giampiero Capobianco, MD, PhD
dure. We agree with this consideration and think that Erich Cosmi, MD
ultrasonographic findings (such as uterine wall thickness, *Departments of Pharmacology and
fetal presentation or orientation, and the amniotic fluid Obstetrics and Gynecology
volume) can help the surgeon to characterize the inher- University of Sassari
ent risk to the fetal laceration injury at the cesarean Viale San Pietro 12
delivery also. Our medical management in the operating Sassari 07100, Italy
theater states always to perform obstetric examination E-mail: dessole@uniss.it
before the performance of cesarean delivery. In cases
of doubt about fetal presentation and/or orientation,
we do transabdominal and transvaginal ultrasonogra- Reference
phy. Then, if we observe an abnormal presentation, we
1. Dessole S, Cosmi E, Balata A, Uras L, Caserta D, Capobianco G,
use the precautions that we have reported.1 But, despite et al. Accidental fetal lacerations during cesarean delivery: experi-
this management, the risk of accidental fetal lacerations ence in an Italian level III university hospital. Am J Obstet Gynecol
during cesarean delivery may be reduced but not elimi- 2004;191:1673-7.
doi:10.1016/j.ajog.2005.02.125
American Journal of Obstetrics and Gynecology (2005) 193, 899
www.ajog.org
Correction
The article, ‘‘Metabolites of progesterone and the pregnane X receptor: A novel pathway regulating uternine
contractibility in pregnancy?’’ by Mitchell et al, published in the April 2005 (Volume 192, p 1304-15) contains an error
in Table I. The corrected table appears below.
Table I Inhibitory activity on oxytocin-induced contractions in rat myometrium in vitro and binding affinity for PXR for metabolites
of progesterone
Steroid Inhibition at 100 mM* PXR activationy
DMSO (control) 98.8 G 1.6 d
4-pregnen-3,20-dione (progesterone)z 45.0 G 5.0 1.2
4-pregnen-20a-ol,3-onez 63.4 G 6.1 1.3
4-pregnen-20b-ol,3-onez 40.9 G 8.2 0.3
5-pregnen-3b-ol, 20-one (pregnenolone) 91.3 G 5.5 2.3
5a-pregnan-20a-ol,3-one 75.2 G 9.2 6.7
5a-pregnan-3a-ol,20-onez 67.7 G 7.1 6.8
5a-pregnan-3a,20b-diol 61.4 G 11.4 1.4
5a-pregnan-3b,20a-diol 90.2 G 11.1 14.5
5b-pregnan-3,20-dionez 28.0 G 7.9 2.5
5b-pregnan-20a-ol,3-onez 32.0 G 10.0 4.2
5b-pregnan-3a-ol,20-onez 50.4 G 9.7 3.3
5b-pregnan-3b-ol,20-onez 49.6 G 7.3 1.7
5b-pregnan-3a,20a-diol 85.6 G 2.0 8.1
5b-pregnan-3a,20b-diol 93.5 G 13.0 1.1
5b-pregnan-3b,20a-diolz 25.2 G 5.9 2.0
5b-pregnan-3b,20b-diolz 23.7 G 5.3 0.7
RU486 (mifepristone, n Z 1) 19.5 1.2
* Concentrations 1-100 mM were used but only the 100 mM dose is shown (n Z 3 – 10 experiments for each metabolite except mifepristone).
y
Calculated as the EC50 concentration (mM) for activation of mouse PXR from the transient transfection assays (see text).
z
Denotes statistically significant inhibition by ANOVA (P ! 0.05). The bold font indicates the metabolite used in the present studies.
doi:10.1016/j.ajog.2005.07.084
Information
FOR READERS
Communications
Communications regarding editorial management should be elsevier.com. Reprints of single articles available online may be
addressed to the appropriate Editor. For mailing address, tele- obtained by purchasing Pay-Per-View access for $30 per article
phone and fax numbers, e-mail address, and detailed instructions on the journal website, www.ajog.org.
on manuscript submission, see Information for Authors in the first Copyright 2005 Mosby, Inc. All rights reserved.
issue of each volume and on our Web site at www.ajog.org. Electronic availability: Online from Elsevier Inc., at http://
Address business communications to Periodical Publisher, www.ajog.org and from Ovid Technologies, Inc, at (800)950-
Mosby, Inc, 11830 Westline Industrial Dr., St. Louis, MO 2035.
63146-3318. Microform edition available from ProQuest Information and
Customer service (orders, claims, online, change of address): Learning, 300 N Zeeb Rd, Ann Arbor, MI 48106-1346; tele-
Elsevier Periodicals Customer Service, 6277 Sea Harbor Drive, phone (734)761-4700 or (800)521-0600.
Orlando, FL 32887-4800. Tel: (800) 654-2452 (U.S. and Photocopying: Single photocopies of single articles may be
Canada); (407) 345-4000 (outside U.S. and Canada). Fax: (800) made for personal use as allowed by national copyright laws. Per-
225-4030 (U.S. and Canada); (407) 363-9661 (outside U.S. and mission of the Publisher and payment of a fee is required for all
Canada). E-mail: elspcs@elsevier.com. other photocopying, including multiple or systematic copying,
Address changes must be submitted four weeks in advance. copying for advertising or promotional purposes, resale, and all
Yearly subscription rates: United States and possessions: Indi- forms of document delivery. Special rates are available for educa-
vidual $212; Institution $470; Student/Resident: $106. All other tional institutions that wish to make photocopies for non-profit
countries (prices include airspeed delivery): Individual $260; Insti- educational classroom use.
tution $522; Student/Resident $130. Single Issues $44. To receive Permissions may be sought directly from Elsevier’s Rights De-
student/resident rate, orders must be accompanied by name of partment in Philadelphia, PA, USA: phone (+1) 215 239-3804,
affiliated institution, date of term and the signature of program/ fax (+1) 215 239-3805, e-mail healthpermissions@elsevier.com.
residency coordinator on institution letterhead. Orders will be Requests may also be completed on-line via the Elsevier homepage
billed at the individual rate until proof of status is received. (http://www.elsevier.com/locate/permissions).
Current prices are in effect for back volumes and back issues. In the USA, users may clear permissions and make payments
Further information on this journal is available from the Pub- through the Copyright Clearance Center, Inc., 222 Rose-
lisher or from this journal’s website (http://www.ajog.org). Infor- wood Drive, Danvers, MA 01923, USA; phone: (+1) (978)
mation on other Elsevier products is available through Elsevier’s 7508400, fax: (+1) (978) 7504744, and in the UK through the
website (http://www.elsevier.com). Copyright Licensing Agency Rapid Clearance Service (CLARCS),
Advertising information. Advertising orders and inquiries can 90 Tottenham Court Road, London W1P 0LP, UK; phone: (+44)
be sent to: USA, Canada, and South America, Elsevier Inc, John 20 7631 5555; fax: (+44) 20 7631 5500. Other countries may
Maney, 360 Park Avenue S, New York, NY 10021; phone: 212- have a local reprographic rights agency for payments.
633-3855; fax: 212-633-3820; e-mail: j.maney@elsevier.com. Clas- Derivative works: Subscribers may reproduce tables of contents
sified advertising orders and inquiries can be sent to Traci Peppers, or prepare lists of articles including abstracts for internal circula-
Elsevier, 360 Park Avenue South, New York, NY 10010; phone tion within their institutions. Permission of the Publisher is re-
(212)633-3766. Europe and the rest of the world, Julie Toop; quired for resale or distribution outside the institution.
phone +44 (0) 1865 843016; fax +44 (0) 1865 843976; e-mail Permission of the Publisher is required for all other derivative
media@elsevier.com. works, including compilations and translations.
Electronic storage or usage: Permission of the Publisher is re-
quired to store or use electronically any material contained in
Author inquiries this journal, including any article or part of an article.
For inquiries relating to accepted articles (including electronic Except as outlined above, no part of this publication may be re-
submission where available), please visit Elsevier’s Author Gate- produced, stored in a retrieval system or transmitted in any form
way at http://authors.elsevier.com. The Author Gateway also pro- or by any means, electronic, mechanical, photocopying, recording
vides the facility to track accepted articles and set up e-mail alerts or otherwise, without prior written permission of the Publisher.
to inform you of when an article’s status has changed, as well as Address permissions requests to: Elsevier Rights Department, at
detailed artwork guidelines, copyright information, frequently the fax and e-mail addresses noted above.
asked questions, and more. Please see Information for Authors Notice: No responsibility is assumed by the Publisher for any
for individual journal. Contact details for questions arising after injury and/or damage to persons or property as a matter of prod-
acceptance of an article, especially those relating to proofs, are ucts liability, negligence or otherwise, or from any use or opera-
provided after registration of an article for publication. tion of any methods, products, instructions or ideas contained

1 The paper used in this publication meets the requirements of in the material herein. Because of rapid advances in the medical
sciences, in particular, independent verification of diagnoses and
ANSI/NISO Z39.48-1992 (Permanence of Paper).
drug dosages should be made.
Reprints. For queries about author offprints, e-mail author- Although all advertising material is expected to conform to
support@elsevier.com. To order 100 or more reprints for educa- ethical (medical) standards, inclusion in this publication does
tional, commercial, or promotional use, contact the Commercial not constitute a guarantee or endorsement of the quality or
Reprints Department, Elsevier Inc., 360 Park Avenue South, New value of such product or of the claims made of it by its manu-
York, NY 10010-1710. Fax: (212) 462-1935; e-mail reprints@ facturer.
16A September 2005

American Journal of Obstetrics and Gynecology (2005) 193, e1ee10
www.ajog.org
The Following lists contains Diplomates Passing

the 2004 Ob/Gyn Oral Examination From
11/12/2004 to 01/14/2005
Certificates valid through 12/31/2010 Arky, Lawrence, MD, West Hartford, CT
Arovas, N. Elizabeth, MD, Irvine, CA
Report Date: 02/02/2005 As-Sanie, Sawsan, MD, Cary, NC
Ascencio, J. Krystal, MD, Baldwin, NY
Aafreen, Tahmina, MD, Watertown, WI Asghar, Pamela, MD, Valrico, FL
Abbassi, Derek C., MD, Berkeley, CA Ashkin, Eric, MD, Bethesda, MD
Abbruzzi, Kristin, MD, Bethlehem, PA Atkins, Katharyn Meredith, MD, Natick, MA
Abdelghani, Waleed M., MD, Hackensack, NJ Aung, Diana, MD, Saratoga, CA
Abdul-Mbacke, Makunda, MD, Columbus, NJ Avery, Anita R., MD, East Grand Rapids, MI
Abrams, Robert M., MD, Chicago, IL Aviles, Angel, MD, San Juan, PR
Abravesh, Soodabeh, MD, Lancaster, CA Awonuga, Awoniyi Olumide, MD, Staten Island, NY
Adams, Michelle Dawn, MD, Oakland, CA Awwad, Amal A., MD, Ypsilanti, MI
Adams, Yarameekah, MD, Corona, CA Axford, Paul Campbell, MD, Brockport, NY
Addy, Douglas M., MD, Columbia, SC
Adeniji, Adegoke, MD, Visalia, CA Bachman, Jolene, MD, Kailua, HI
Adkins, Tonya Marie, MD, Fredericksburg, VA Bahador, Afshin, MD, San Diego, CA
Agarwala, Neena, MD, State College, PA Baird, Darcy Page, MD, Livermore, CA
Agbasi, Nwogo, MD, Coudersport, PA Bajzak, Krisztina Ilona, MD, Raleigh, NC
Aguilar, Vivian del Carmen, MD, Belle Mead, NJ Baker, Brenda S., MD, Newton Center, MA
Ahsan, Arzou, MD, Oakland, CA Baldwin, Kimberly S., MD, Franklin Park, NJ
Ailawadi, Radhika K., MD, Nashville, TN Balk, Kimberly Margaret, MD, Gilbert, AZ
Akinnawonu, Kolawole Felix, MD, Mobile, AL Balli, Kevin C., MD, Loma Linda, CA
Al Khabbaz, Antoun, MD, Harlan, KY Bane, Susan, MD, Greenville, NC
Alan, Carol Brown, MD, Columbia, SC Banks-Jackson, Roslyn, MD, Stone Mountain, GA
Allemand, Carrie B., MD, Rochester, MN Barber, Colleen M., MD, Manchester, NH
Allen, Jennifer E., MD, McKinney, TX Barber, James, MD, Henderson, NV
Alozie-Arole, Chidinma Nwanmgbede, MD, MPH, Barrett, Diane Pettit, MD, Vienna, VA
Canton, MI Barsoom, Michael, MD, Omaha, NE
Alverson, Mitchell Wade, Jr., MD, Opelika, AL Bartels, Russell, MD, Phoenix, AZ
Amersi, Shamsah F., MD, Santa Monica, CA Bartholomew, Lynne, MD, Boston, MA
Amini, Dennis, MD, Bethesda, MD Barton, Charles Brent, MD, Barbourville, KY
Amui, Jewel, MD, Woodlynne, NJ Bashuk, Stephen Todd, MD, Alpharetta, GA
Andrews, Stacy, MD, Washington, DC Bason-Mitchell, Secily N., MD, San Francisco, CA
Angelats, Juan Carlos, MD, Minneapolis, MN Beatty, Heather, MD, Carmichael, CA
Annamalai, Akila, MD, San Jose, CA Beceiro, Anna, MD, San Antonio, TX
Anstine, Suzanne Marie, MD, Boise, ID Beckner, James P., MD, Kingsport, TN
Anupol, Noel M., MD, Milford, DE Beer, Rodolfo Scott, MD, Kenosha, WI
Arbona, Heidi D., MD, Westerville, OH Behnam, Nadereh, MD, Pomptonplains, NJ
Arena, Julianne, MD, Quincy, MA Bendayan, Jose S., MD, North Woodmere, NY
doi:10.1016/j.ajog.2005.08.008
e2 2004 Ob/Gyn Oral Examination Diplomates
Benevides, Colby M., MD, Cary, NC Burlison, C. Kevin, MD, Bolivar, MO

Benoit, Richard Michael, MD, New Bedford, MA Burt, Randall J., MD, Frisco, TX
Benson, Michael A., MD, East Hanover, NJ Burton, Tracy, MD, Boca Raton, FL
Berger, Catherine A., MD, Berkeley, CA Busch, Jeanne M., DO, Virginia Beach, VA
Berger, Daniel Andrew, DO, Mahwah, NJ Bustamante, Ernestine, MD, Phoenix, AZ
Berghahn, Laura J., MD, Madison, WI Butler, Kira Linell, MD, Frisco, TX
Bernard, Kimberly, MD, Ellicott City, MD Byerly, Lee Erich, MD, Big Spring, TX
Bernstein, Sara, MD, Wellington, FL
Best, Gail, MD, Glendale, CO Calinisan, Joan Hazel B., MD, Temecula, CA
Beyer, Derek, MD, Owatonna, MN Callahan, Cathleen, MD, Rochester, NY
Bhatnagar, Upasana, MD, Laurel, MD Carisio-Farber, Renee, MD, Hume, VA
Bhojwani, Navin C., MD, Charlotte, NC Carley, Michael E., MD, Dallas, TX
Biggio, Joseph R., Jr., MD, Birmingham, AL Carlson, Niti Bhalla, MD, Evans, GA
Birenbaum, Emily K., MD, Berkeley, CA Carmona-Keller, Diana Patricia, MD, Venetia, PA
Biter, Robert M., MD, Encinitas, CA Carney, Michele C., MD, Aurora, IL
Bizovi, Leslie J., MD, Brooklyn, NY Carrera-Leal, Benito, MD, Brownsville, TX
Block, Joanne Jean Vicari, MD, Canton, MI Carroll, C. Shannon, DO, Jackson, MS
Bloom, Brooke, MD, Sharon, MA Carter, Andrew W., MD, Scottsdale, AZ
Bloomgarden, Amy, MD, New York, NY Caruthers, Thomas Jefferson, Jr., MD, Atoka, TN
Blumberg, Isabel S., MD, New York, NY Cassagnol, Hans, MD, Mountain Top, PA
Bobo, Kimberly A., MD, Houston, TX Cha, Ae Seon, MD, Troy, OH
Bochantin, Tracy A., MD, East Peoria, IL Chadwick, Tuesday L., MD, Paris, TX
Booker, James J., IV, MD, Winter Haven, FL Chalmers, Robert W., MD, Ft. Wainwright, AK
Boreham, Muriel Keenze, MD, Coppell, TX Chan, John K., MD, Foster City, CA
Boruta, David Michael, II, MD, Whitefish Bay, WI Chan, Kent C., MD, Astoria, NY
Boucher, Danielle Anne, MD, Rockledge, FL Chancellor, Jeff D., MD, Waco, TX
Bouma, Curt L., MD, Claremont, CA Chang, Joe S., MD, Irving, TX
Bovone, Suzanne C., MD, Campbell, CA Chang, Lisbeth H., MD, Encino, CA
Bowden, Mirna D., MD, Whitefish, MT Charles, Hillary H., MD, Phoenix, AZ
Bower, Brian, MD, Huntington, WV Chauhan, Subodhsingh Rambabusingh, MD,
Boyd, Scott C., MD, Ft. Wayne, IN Charleston, WV
Boyd, W. David, MD, Jacksonville, FL Chauhdry, Tahir Akram, DO, Hornell, NY
Boykin, Diane, MD, Bowie, MD Chavez, Martin R., MD, Hillsborough, NJ
Boyle, Jennifer M., MD, Grand Lake, CO Chen, Dehan, MD, Paramus, NJ
Brady, Wesley A., MD, Dallas, TX Chesshir, Kimberly Ann, MD, Dallas, TX
Brandewie, Elizabeth V., MD, Sidney, OH Cheung, Mon-Lai, MD, Bakersfield, CA
Branson, Cheryl, MD, Los Altos, CA Chismar, Steven, MD, North Lima, OH
Brantley, Kara L., MD, Jackson, MS Chmait, Ramen H., MD, Tampa, FL
Braz-Martin, Susan E., MD, Rehoboth, MA Choi, Janet Mee-Kyung, MD, New York, NY
Brecht-Doscher, Aimee, MD, Ventura, CA Chong, Deborah, MD, Sacramento, CA
Bridge, Danielle Duquette, MD, DePere, WI Christie, Heather DiMaio, MD, Gainesville, FL
Bridges-White, Kimberly G., MD, Deptford, NJ Chu, Christina Shuwai, MD, Moorestown, NJ
Brock, Julia R., MD, Shelburne, VT Chu, Kristine, MD, Englewood Cliffs, NJ
Brock, Mitch, MD, St. Simons Island, GA Chung, Judith Hyunsuk, MD, Long Beach, CA
Brodkin, Tara, MD, Phoenix, AZ Chung, Susie Noel, MD, Baltimore, MD
Brossfield, Jeralyn Ellice, MD, Rancho Mirage, CA Clayton, Esther R., MD, Minneapolis, MN
Broth, Richard Elliot, MD, Cherry Hill, NJ Cly, Geoffrey Charles, MD, Fort Wayne, IN
Brown, Desrene Kerry-Ann, MD, Bluffton, OH Cobb, Kelly A., MD, Spartanburg, SC
Brown, Jennifer A., MD, APO, AP Cockrum, Holly, MD, North Little Rock, AR
Brown, Melissa Ann, MD, Salt Lake City, UT Cohn, Jennifer Lynn, MD, Novi, MI
Brown, Monique LaMae, MD, MBA, Tucson, AZ Cole, David Scott, MD, Ardsley, NY
Brown, Sheila Elaine, MD, Philadelphia, PA Cole, Karen F., MD, Jackson, MS
Brown, Stephen M., MD, Marcellus, NY Cole, Nichole Fleming, MD, Houston, TX
Brown, Susan Bartow, MD, Montville, NJ Coleman, Jenell S., MD, Berkeley, CA
Browning, Joseph Edward, MD, Farmville, VA Coleman, Joseph I., MD, Lake Wylie, SC
Burgos, Rafael, MD, Yorktown, VA Coley, Katherine Price, MD, Hilton Head Island, SC
2004 Ob/Gyn Oral Examination Diplomates e3
Collin, Jennifer Wiley, MD, Niskayuna, NY Drake, Brian E., MD, Portland, OR
Connell, Kathleen Anne, MD, Shelton, CT Draper, Joy, MD, Richmond, VA
Connelly, Diane A., MD, San Bernardino, CA Driggers, Rita Wesley, MD, Davidsonville, MD
Cooper, Jason Edward, MD, Houston, TX Duncan, Kelly D., MD, Renton, WA
Copenhaver, Catherine S., MD, Bethesda, MD Dunn, Jennifer Colleen, MD, Elmhurst, IL
Corbett, Anna Rachel, MD, Denver, CO Duplantier, Noel Marie, MD, Bay St. Louis, MS
Cordts, Vicki Lynne, MD, Cudahy, CA Durbin, Mary A., MD, Bay City, MI
Corovessis, Catherine C., MD, Katy, TX Durfee, Mary Kay, MD, Forest Lake, MN
Courban, Daniella, MD, Cambridge, MA Durkee, Stephen M., MD, Macon, GA
Cox, Carol S., MD, Tampa, FL Dyoco, Lorree L., MD, Pewaukee, WI
Crandall, Blane, MD, Clinton, OK Dzanic-Cemalovic, Naida, MD, Roosevelt Island, NY
Crawford, Janel, MD, El Dorado Hills, CA
Crews, James H., MD, Jackson, MS Eble, Amy Catherine Wiedo, MD, Manvel, TX
Cropper, Stephanie, MD, Pomona, CA Echols, Karolynn Teresa, MD, Lafayette, LA
Crowley, Kelli B., MD, Schererville, IN Eckert, Cynthia A., MD, Overland Park, KS
Crute, Kimberly A., MD, Savannah, GA Eckhardt, Elizabeth Ann, MD, St. Anthony, MN
Culver, James M., MD, Charlottesville, VA Eckman, Troy Douglas, MD, Macomb, IL
Cummings, Allegra, MD, New York, NY Eddib, Abeer A., MD, East Amherst, NY
Currens, Andrea F., MD, Asheville, NC El Gammal, Nahed Z., MD, Decatur, GA
Cutler, Wendy, MD, Austin, TX Elam, Teryl, MD, Anchorage, AK
Cwiak, Carrie, MD, Atlanta, GA Elias, Darryl, MD, Jennings, LA
Elliott, Tracy, MD, Plano, TX
Daftary, Gaurang Shirish, MD, New Haven, CT Eltayeb, Suha Elsersawi, MD, Ajax, ON
Dalloul, Mudar, MD, Brooklyn, NY Emberson, John W., MD, Ringgold, GA
Dalton, Vanessa, MD, Ann Arbor, MI Emerick, George J., MD, Wellington, FL
Davidson, Aaron, MD, Bayside, NY Enegess, Deborah M., MD, Pittsburgh, PA
Davis, Daniel Heckel, MD, Columbus, IN Engel, Chris L., MD, Denver, CO
Davis, Kent Stuart, MD, Seattle, WA Ensminger, Jason Leslie, MD, Danville, VA
Davis, Michael, MD, Portland, OR Enyedy, Tricia M., MD, Carlsbad, CA
Dayal, Molina Bhatnagar, MD, Rockville, MD Ertel, Sylvia, MD, Indianapolis, IN
De Riese, Cornelia, MD, Lubbock, TX Ester, John B., MD, Rice Lake, WI
De Vita, Laura, MD, Sandy Hook, CT Estes, Vineeta Ahluvalia, MD, Homewood, AL
DeCesare, Julie A. Zemaitis, MD, Gulf Breeze, FL Estrella, Eduardo F., MD, Guaynabo, PR
DeFalco, Lisa, DO, Fredon, NJ Etsitty, Edison Virgil, MD, Gallup, NM
DeFranco, Emily, DO, Lexington, KY
DeKay, Peter B., MD, Reno, NV Fahmy, Farris, MD, Rye Brook, NY
DeLong, Gregory A., MD, Millersburg, OH Famuyide, Abimbola O., MD, Rochester, MN
DeVore, Kimberly M., DO, APO, AE Farber, Drew Alexander, DO, Flemmington, NJ
Dean, Gillian, MD, New York, NY Fedor, Debra M., DO, Frederick, MD
Deem, Adrianne M., MD, Ft. Worth, TX Fernandez, Andrea S., MD, Winston Salem, NC
Deli, Barbara, MD, New York, NY Fernandez, Fabiola, MD, Forest Park, IL
Demasio, Kafui Alfreda, MD, MPH, North Castle, NY Ferry, Amanda Catherine, MD, Bay Village, OH
Dennard-Hall, Keisha, MD, Pittsburgh, PA Filor, Caroline F., MD, Schenectady, NY
Dennehy, Daniel Thomas, MD, Wallingford, CT Fine, Stacey Silverman, MD, Thousand Oaks, CA
Devlin, Jeanine Grillo, MD, Huntington Valley, PA Finegan, Mary M., MD, Sacramento, CA
DiLuigi, Andrea, MD, West Hartford, CT Finger, Laura S., MD, Shady Shores, TX
Dietrich, Yvonne M., M., Ocean Springs, MS Fiore, Joseph, MD, Hampton Cove, AL
Dillard, Naima Aleijah, MD, Greensboro, NC Fisher, Timothy J., MD, Pensacola, FL
Dillard, Tiffani Clark, MD, Champaign, IL Fitz, Bronwyn E. S., MD, Charlottesville, VA
Dinnall, Vanessa N., MD, New York, NY Flaherty, Kara, MD, Colchester, VT
Dizon, Jennifer J., MD, South Pasadena, CA Flippin-Trainer, Angela D., MD, Tallahassee, FL
Dobay, Kristin Josef, MD, Clarksville, TN Flora, Stefanie, MD, Bellevue, MI
Doll-Pollard, Anne Elizabeth, MD, Breese, IL Flowers, Coy A., MD, Jacksonville, NC
Donnelly, Jennifer H., MD, Tulsa, OK Flubacher, John Scott, DO, Winter Harbor, ME
Dougherty, John Joseph, MD, Franklin, VA Fogg, Darlene E., MD, Lebanon, TN
Downs, Levi Stanford, Jr., MD, Minneapolis, MN Foglia, Lisa M., MD, APO, AE
Foiles, Andrea C., MD, Wilmington, NC Goulding, Dawn M., MD, Leonard, MI
Foster, Kimberly, M.D, Beavercreek, OH Gowell, Kerry P., MD, S. Dartmouth, MA
Fox, Michelle C., MD, Baltimore, MD Graupe, Menachem H., MD, Mequon, WI
Francis, Anita, MD, Anaheim Hills, CA Greely, Jocelyn Tinsley, MD, New Orleans, LA
Francis, Sean L., MD, Martinez, GA Griffin, David, MD, Advance, NC
Franka, John Chris, MD, San Antonio, TX Griffith, Anna, MD, Marquette, MI
Fraser, Christina V., MD, San Antonio, TX Grondahl, Robert E., MD, Sanford, NC
Fraser, Cynthia Heresi, MD, Canandagua, NY Grossman, Eric B., MD, Cherry Hill, NJ
Freeman, Kendal Trent, MD, Chicago, IL Grotegut, Chad A., MD, Lafayette Hill, PA
Freeman-Kwaku, Mala Adele, MD, Charlotte, NC Grounds, Carrie A., MD, Leawood, KS
Freije, William Alexander, MD, PhD, Los Angeles, CA Grover, Julie G., MD, Provo, UT
Freitas, Sarah A., MD, Waconia, MN Grube, Jennifer O., MD, Evergreen, CO
Friedman, Amir, MD, Independence, KS Guidry, TriCia N., MD, Lake Charles, LA
Friedman, Yaron, MD, Orinda, CA Gustilo-Ashby, Arlan M., MD, Shaker Heights, OH
Friend, Diana J., MD, Los Angeles, CA Guth, Walter Lee, MD, Lake Charles, LA
Froerer, Christian Donald, MD, Holladay, UT Gutierrez, Christina Elaine, MD, San Antonio, TX
Fudge, Marilyn W., MD, St. Petersburg, FL Gutierrez, Griselda C., MD, Redondo Beach, CA
Fullwood, Valeria T., MD, Conyers, GA Guzman, Melissa A., MD, Austin, TX
Funk, Catrina M.C., MD, Georgetown, OH
Haag, Terry A., DO, Clarion, PA
Gahn, David R., MD, Tahlequah, OK Habak, Patricia J., MD, Phoenix, AZ
Galla-Elizeus, Kathy Lynne, MD, Monroeville, PA Hager, Julie Strebel, MD, Oklahoma City, OK
Galusha, Andrea L., MD, Lewisville, TX Hagerty, Genevieve B., MD, Hilliard, OH
Gandhi, Neelam, MD, Carpentersville, IL Hakakha, Michele Milovina, MD, Beverly Hills, CA
Gandia, Justin, MD, Chicago, IL Hakim, Paul George, MD, Bloomfield Hills, MI
Garcia, Anna Marie, MD, Cortland, NY Hakimi, Daniel, MD, Clifton, NJ
Garcia, Carlos, MD, Pinecrest, FL Halbach, Melissa Menson, MD, Bogart, GA
Gardner, Sandy Lynn, MD, Cincinnati, OH Hall, Richard Rush, MD, Big Rapids, MI
Garner, Elizabeth I.O., MD, M.P.H., Newton, MA Hamid, Cherine A., MD, Palm Harbor, FL
Gaskin, Richelle S., MD, Baltimore, MD Hamilton, Monique Pierce, MD, New Orleans, LA
Gatcliffe, Troy A., MD, Dallas, TX Hammond, Lee Ann, MD, Colorado Springs, CO
Gebhart, John Bennett, MD, Rochester, MN Hanauer, Jennifer T., MD, Wantagh, NY
Gennarelli, Louis A., MD, New York, NY Hannah, J. Todd, MD, Bentonville, AR
Gerber, Dmitry, MD, New Rochelle, NY Hardison, Joshua L., MD, Henderson, NC
Gernhart, Sarah Vondrak, MD, Omaha, NE Hardy, Curtis Lee, Jr., MD, Brooklyn, NY
Ghiai, Afshan, MD, Oxnard, CA Hardy, Lori L., MD, Wheaton, IL
Ghosh, Sumit Bijoy, MD, Dallas, PA Harkness, Cameron Blair, MD, Loveland, OH
Giebmanns, Thomas, MD, Burlington, NC Harms, Konrad P., MD, Pearland, TX
Giedt, Amy L., MD, Fairway, KS Harper, Adrienne K., MD, San Diego, CA
Gillean, Julia R., MD, Rockwall, TX Harper, Julie, MD, Hernando, MS
Gillett-Elrington, Ann R., MD, Inkster, MI Harrell, Amy L., MD, College Station, TX
Giudice, Michael A., MD, Towson, MD Harris, Julie A., MD, Pine Bluff, AR
Glass, Justin Todd, MD, University City, MO Harris, Selena, MD, Hermosa Beach, CA
Glymph, Brian, MD, Nacogdoches, TX Harris, Terrence W., MD, Coral Springs, FL
Gnoy, Andrew W., MD, Bridgewater, NJ Hart, Stuart R., MD, Snellville, GA
Goldberg, Dahna L., MD, Clarksville, MD Hartshorn, Rachel Michelle, MD, Oakland, CA
Goldman, Debra L., MD, Cranston, RI Hashemi, Neda, MD, Centreville, VA
Goldman, Noah Adam, MD, Hoboken, NJ Hasling, Jill C., MD, Columbus, GA
Golpira, Elizabeth Baker, MD, Norfolk, VA Haver, Mary Claire, MD, Houston, TX
Gonima, Camilo A., MD, San Antonio, TX Hayden, Siobhan F., MD, Cooperstown, NY
Gonzalez, Allyson Ann, MD, Santa Monica, CA Heath, Christy Ellen James, MD, Birmingham, AL
Gor, Hetal B., MD, Tenafly, NJ Heaton, Alison A., MD, Birmingham, AL
Gorenberg, David Matthew, MD, Seattle, WA Heiberger, Heather Brenton, MD, Lake Forest, IL
Gorlitsky, Helen, MD, Marlton, NJ Heinberg, Eric M., MD, Providence, RI
Gottlieb, Aren, MD, New York, NY Held, Barbara, MD, Houston, TX
Gould, Natalie Sandra, MD, Denver, CO Helliwell, Jason, MD, Bakersfield, CA
Heltzer, Paul S., MD, Mt. Vernon, NY Jennings, Christopher, MD, Anderson, SC
Hennesy, Michael S., MD, Centennial, CO Jensen, Teresa G., MD, Fremont, CA
Henry, Kerrie M., MD, Ft. Walton Beach, FL Jerald, Heather L., MD, San Antonio, TX
Herbolsheimer, Heather, DO, Bowie, MD Jhang, Helen J. Park, MD, Eastchester, NY
Hernandez, Carmelo A., MD, Brevard, NC Jin, Wen Hui, MD, New York City, NY
Hernandez-Parkhurst, Annette M., MD, Tucson, AZ Jiricko, Audrey, MD, Farmington, UT
Hershberger, Jonathan, MD, Tampa, FL Jo, Jenny, MD, Atlanta, GA
Hibner, Michael, MD, Phoenix, AZ Johns, Joseph Bram, MD, Katy, TX
Hickey, Kimberly W., MD, Burtonsville, MD Johnson, Glenda, MD, Shreveport, LA
Hicks, Diana Carol, MD, Suffolk, VA Johnson, Kim, DO, Trenton, MI
Hill, Andrea, MD, Somerset, KY Johnson, Kristi A., MD, Wooster, OH
Hill, James Bernard, MD, Fayetteville, NC Johnson, Malinda K., MD, Andover, MN
Hillebrand, Linda, DO, Laverne, CA Johnson, Melissa, MD, Sturbridge, MA
Hinton, Emily, MD, Dallas, TX Johnson, Temeka L., MD, Pearl, MS
Ho, Eleanor Meng-Yao, MD, Atlanta, GA Johnson, Traci C., MD, Lilburn, GA
Ho, Phoebe Fei, MD, Puyallup, WA Jones, Andrea D., MD, Naperville, IL
Hogue, Thomas C., MD, Fayetteville, NC Jones, Karin D.S., MD, Riverside, CA
Hoilett-Barrett, Althea, MD, Watertown, NY Jones, Monica Brown, MD, Rochester, MN
Holcroft, Cynthia Jean, MD, Baltimore, MD Jones, Norma L., MD, San Dimas, CA
Hollingsworth, Jill A., MD, Indianapolis, IN Jones, Rachel Zloczover, MD, Wilmington, NC
Horowitz, Neil Stuart, MD, Boston, MA Jones-Monte, Kathleen R., MD, Henderson, NV
Hou, Dennis, MD, Hillborough, CA Jordan, Maureen O., MD, Colorado Springs, CO
Hough, Tobi M., MD, Madison, IN Judge, Karen Olson, MD, Seattle, WA
Householder, JeanMarie, MD, Fort Smith, AR Jurema, Marcus W., MD, Providence, RI
Hu, Michael Pao-Chun, MD, Chesterton, IN
Huang, Jennifer, MD, Holliswood, NY Ka, Hysoo, MD, New Lenox, IL
Huang, Louise, MD, Novato, CA Kagumba, Ada A., MD, Quincy, IL
Huang, Melinda, MD, Eastchester, NY Kahen, Tanaz, MD, Encino, CA
Huang, William, MD, New York, NY Kakarla, Nirupama, MD, Bellaire, TX
Huang, Wilson H., MD, Henderson, NV Kalish, Robin Beth, MD, New York, NY
Huebert, Allison L., MD, Jefferson City, MO Kamat, Aparna, MD, Houston, TX
Huertas, Otoniel, MD, Borger, TX Kamelle, Scott Ahmed, MD, Santa Rosa, CA
Hull, Janice L., MD, Inglewood, CA Kang, Cambria, MD, Santa Ana, CA
Hunt, Mary Elizabeth, MD, New York, NY Kang, Katherine Eunhwa Lee, MD, Fort Lee, NJ
Hunter, Lillian M., MD, Baltimore, MD Kaplan, Chaim E., MD, Brooklyn, NY
Huo, Lichun, MD, Woodland Hills, CA Kasper, Kelly M., MD, Zionsville, IN
Katz, Simone L., MD, Hayward, CA
Icatar, Julianne Yantachka, MD, Norwalk, CT Kaufman, Leah, MD, Glen Oaks, NY
Illuzzi, Jessica L., MD, Fairfield, CT Kaufman, Leesa Ann, MD, San Antonio, TX
Imhoff, Anna L., MD, Winston-Salem, NC Kavanagh, Colleen M., MD, Oakland, CA
Iskandar, Samy R., MD, Easley, SC Keeley, Christopher Courtney, MD, Roanoke, VA
Isom, Matthew J., DO, Harker Heights, TX Kelly-Layton, Tammy Rose, MD, Las Vegas, NV
Ivester, Thomas Steven, MD, Chapel Hill, NC Kerner, Nicole P., MD, Niskayuna, NY
Ivie, Jocelyn Q., MD, Las Vegas, NV Kerr, Mary Campbell, MD, Los Angeles, CA
Kessel, Allan D., MD, Ferndale, MI
Jabara, Sami Issam, MD, Harrisburg, PA Kesselman, Erica, MD, Pomeret Center, CT
Jacobson, Rebecca Eve, MD, Highland Park, IL Kessler, Michael, MD, Hartsdale, NY
Jacobstein, Julie M., MD, Lafayette Hill, PA Khabele, Dineo, MD, Nashville, TN
Jain, Manish, MD, West Bloomfield, MI Khoudary, Maryann, MD, Warren, NJ
Jarrell, April, MD, Spartanburg, SC Kickham, Jennifer Moore, MD, Cambridge, MA
Jarvis, Nicole, MD, Norman, OK Kim, Eleanore S., MD, Oakland, CA
Javernick-Hodges, Necole Ann, MD, Meridian, ID Kim, Martha B., MD, Portland, OR
Jazbec, Andrea M., MD, Denver, CO Kim-Ashchi, Sunwook, MD, Jacksonville, FL
Jebelli, Babak, MD, Redlands, CA Kimmel, Ann L., MD, Tempe, AZ
Jeffries, Lisa R., MD, Boulder, CO King, Amy L., MD, Myrtle Beach, SC
Jeng, Karen K., MD, Plano, TX King, Kathy A., MD, Oconomowoc, WI
King, Rebecca, MD, Sacramento, CA Leight, Melanie A., MD, Abingdon, VA

Kirsch, Daniel Joseph, MD, Omaha, NE Leonard, Caroline J., MD, St. Paul, MN
Kitagawa, Gregory Y., MD, Olmstead Twp, OH Leonard, Stacy Laraine, MD, Texarkana, TX
Kleinberg, Marc, MD, Chicago, IL Leung, Michael P., MD, Houston, TX
Kleiss, Kimberly McCann, MD, Atlanta, GA Levi, Andrew J., MD, Easton, CT
Klingele, Christopher Joseph, MD, Rochester, MN Levin, Daniel Emil, MD, Hazlehurst, MS
Knight, Teresa L., MD, Clayton, MO Levy, Chanan, MD, Lutherville, MD
Knutson, Christina M., MD, Boise, ID Lewis, Andrew J., MD, Winston-Salem, NC
Koellermeier, Michelle Marie, MD, Nenah, WI Lewis, Arlene D., MD, Fredricksburg, VA
Kohanowski, Erika Perl, MD, Mexico, MO Lewis, Dawnette Ann-Marie, MD, MPH, Bronx, NY
Kohl-Thomas, Belinda M., MD, Temple, TX Lewis-Boardman, Mary Beth, MD, Carmel, NY
Kolberg, Amy M., MD, Yankton, SD Li, B. Charles, MD, Memphis, TN
Kongkasuwan, Kimberly R., MD, Ghent, WV Lin, Kathleen, MD, Philadelphia, PA
Koscica, Karen Lynn, D.O., Neptune, NJ Lin, Nancy, MD, Portland, OR
Koutoulas, Antigoni, MD, Boxford, MA Lin, Suzanne H., DO, Lewis Center, OH
Kovac, Christine, MD, Georgetown, TX Linares, Claudio E., MD, Oregon, OH
Kovacs, Peter, MD, Budapest Lindemann, Matt, MD, Anchorage, AK
Kramer, Susan J.L., MD, Highland Park, IL Lindley, Elisa M., MD, Rancho Mirage, CA
Kruger, Janine K., MD, Middleton, WI Lipschitz, Lisa, MD, San Diego, CA
Kruskol, Bryan Mitchell, DO, South Elgin, IL Lipscomb, Lewis D., Jr., MD, Southern Pines, NC
Kulsakdinun, Pamorn, MD, South Barrington, IL Littles, Xercerla Adrenna, MD, Mesquite, TX
Kusic, Michael B., MD, Falls Church, VA Lock, W. Scott, MD, Palo Alto, CA
Kwan, Mindy, MD, New York, NY Lockett, Tammy M., MD, Brooklyn, NY
Kwiecien, Marni S., MD, Tigard, OR Lockey, Renee, MD, Austin, TX
Kwon, Christina Hyun Sook, MD, New York, NY Loewen, Natalie K., MD, Salt Lake City, UT
Lograno, Paul, MD, Smithtown, NY
LaCour, Delese E., MD, Baltimore, MD Lopez, Sandra, MD, Chula Vista, CA
Laasch, Cassie, MD, Royal Oak, MI Loudon, Holly C., MD, New York City, NY
Lachance, Deborah L., MD, Grand Forks, ND Louis, Martha, MD, Glen Cove, NY
Lai, Amy Y., MD, New Hyde Park, NY Lovelady, Alecia, MD, Atlanta, GA
Lalwani, Sasmira, MD, Cambridge, MA Loveland, Joan Elizabeth, MD, Washington, DC
Lam, Garrett Ka Keung, MD, Scottsdale, AZ Lowder, Laura Hines, MD, Charlotte, NC
Lambrou, Nicholas Constantine, MD, Lowre, Cheri A., MD, Studio City, CA
Coral Gables, FL Lu-Ferguson, Ming X., MD, Metuchen, NJ
Lamvu, Georgine M., MD, Durham, NC Lucas, Maureen C., MD, Augusta, ME
Landes, Jennifer Mae, DO, Pennington, NJ Luckett, Marc A., MD, Creve Coeur, MO
Lane, Felicia L., MD, Newport Beach, CA Lukacz, Emily Spencer, MD, La Jolla, CA
Larkin, Molly A., MD, Chadds Ford, PA Lyell, Deirdre Judith, MD, Palo Alto, CA
Lashbrook, Daphne L., MD, Norman, OK Lynch, Sean M., MD, Belmont, NC
Lathi, Ruth Bunker, MD, Palo Alto, CA Lyons, Karen S., MD, Glasgow, KY
Lawrence, Debra Celeste, MD, Conway, AR
Lawrence, Melissa, MD, Kailua, HI MacLaurin, Nancy A., MD, Durham, NC
Lawson, Glasine O., MD, St. Clair, MI Maddison, Sarah Daniel, MD, Raleigh, NC
Lawton, Robyn Denise, MD, Placentia, CA Maddox, Jennifer Mills, MD, Birmingham, AL
Le, Khanh Nha, MD, Fairfax, VA Mahoney, Mary K., MD, Minneapolis, MN
LeBlanc, Joy Paul, MD, Pearland, TX Mahr, Dominique, MD, Los Angeles, CA
Lee, Alice, MD, New York, NY Mainguy, Sarah Byfield, MD, Minneapolis, MN
Lee, Angie Y., MD, Pompton Lakes, NJ Mann, Sylvia L. X., MD, Stevenson Ranch, CA
Lee, Ding-Ding Kelly, MD, Brooklyn, NY Marave, Jerome Sambrano, MD, Cockeysville, MD
Lee, Dorothy, MD, Great Falls, MT Mark, Alice G., MD, Jamaica Plain, MA
Lee, Katrina L., MD, Plano, TX Marks, Nicholas R., MD, Littleton, NH
Lee, Lily L., MD, Manhattan Beach, CA Marroquin, Bridget M., MD, Timberlake, NC
Lee, Lydia K., MD, Roslindale, MA Marshall, Kimberly A., MD, West Burlington, IA
Lee, Nina Y., MD, Antioch, CA Martin, Amy G., MD, Addison, TX
Lee, Sondra B., MD, Los Angeles, CA Martin, Jerry K., Jr., MD, Oxford, MS
Leff, Ricky Phillip, MD, Melbourne, FL Martin, Julie Ann, MD, Olathe, KS
Martin, Todd D., MD, Lincoln, NE Moulton, James Richard, II, MD, Atlanta, GA
Marvin, Judy L., MD, Spokane, WA Moylan, Laura Beth, MD, Dover, DE
Marzano, David A., MD, Ann Arbor, MI Mozayeni, Pantea, MD, Azusa, CA
Mason, Romy E., MD, Denver, CO Muench, Michael V., MD, Point Pleasant, NJ
Massa, Bonni Stacy, MD, San Francisco, CA Mullin, Deanna Eleanor, MD, Roseville, CA
Mathis, Robert T., MD, Columbus, GA Mullin, Patrick Michael, MD, FPO, AP
Mathison-Ezieme, Linda Joy, MD, Harvey, LA Murdock, Tammy J., MD, San Antonio, TX
Mattern, Shannon E.T., MD, High Point, NC Murray, Catrina, MD, South Bend, IN
Matthews, Louise Suzanne, MD, Winnetka, IL Murray, Kristin M., MD, Anchorage, AK
Mazarei, Nahid, MD, Reston, VA Murthy, Kamaljeet Purhar, MD, Lafayette Hill, PA
McCarthy, Lizbeth, MD, Louisville, CO Murthy, Rachel, MD, Athens, GA
McCartin, Richard T., MD, Kapolei, HI
McClure, Kathryn A., MD, Gulfport, MS Naim, Arjang, MD, Los Angeles, CA
McCullough, Michael, MD, Clarksville, TN Nazareth, Sonja F., MD, Irvine, CA
McDonald, Tedd Mikel, MD, North Richland Hills, TX Neal, Laura, MD, Dubuque, IA
McElrath, Thomas Frederick, MD, PhD, Nelson, Erin L., MD, Fayetteville, NC
Sherborn, MA Nelson, Keith H., MD, Winterville, NC
McElroy, Tara M., MD, Mayfield Heights, OH Nelson, Renee, MD, San Diego, CA
McGee, Carmen, MD, Atlanta, GA Nelson, Virginia Carney, D.O., Lauderdale, MS
McGhee, Vida L., MD, Munster, IN Nemunaitis-Keller, Jennifer A., MD, Peddleton, IN
McIntosh, Elizabeth, MD, Savannah, GA Nevins, Juliet Marcia, MD, Sewell, NJ
McIntosh, Kimberly Jaye, MD, Snellville, GA Ng, Eliza, MD, New York, NY
McKinney, Gisele, MD, Williamston, NC Nguyen, Anh T., MD, Belmont, CA
McKnight, Eric J., MD, Hudson, OH Nguyen, Chau Dang-Thi, MD, Sugarland, TX
McLawhorn, Netasha D.S., MD, Henderson, NC Nguyen, Tan-Loc, MD, Warner Robins, GA
McNeive, Daniel F., MD, St. Louis, MO Nichols, John M., DO, South Jordan, UT
Medley, Tamara C., MD, Grants Pass, OR Nichols, Mark E., MD, Arlington, TX
Mehta, Minal G., MD, Irvine, CA Nieves-Neira, Wilberto, MD, New Brunswich, NJ
Meigs, Amory H., MD, Philadelphia, PA Nokari, Daed, MD, Brooklyn, NY
Meinig, Martin L., MD, Fairfax, VA North, Michelle M., MD, Marshfield, MA
Mele, Michele M., MD, Philadelphia, PA Nugent, Andrea Scharfe, MD, Baltimore, MD
Melnik, Marc David, MD, Whittier, CA Nusz, Jean H., MD, Cincinnati, OH
Memarzadeh, Sanaz, MD, Los Angeles, CA Nwankpa-Keshinro, Ngozi I., MD, Brooklyn, NY
Mendelssohn, Andrea R., MD, Castro Valley, CA Nwosa, Nkemdilim, MD, Silver Springs, MD
Merhi, Nahla O., MD, New Lenox, IL
Merideth, Melissa A., MD, Bethesda, MD O’Connor, Anne Marie, MD, Espanola, NM
Mestad, Renee E., MD, Gresham, OR O’Connor, Siobhan Marie, MD, Pittsburgh, PA
Metherell, James F., MD, Greenville, SC O’Regan, Neil Jeffrey, MD, Sanford, ME
Meyer, Leslie S., MD, Fredericksburg, VA O’Reilly, Geralyn Camacho, MD, Seattle, WA
Michael, Kristi C., MD, Shreveport, LA O’Shea, Maura Kathleen, MD, Akron, OH
Miguel, Breno Loureiro, MD, Houston, TX Oakes, Richard Edsel, II, MD, Bridgeport, WV
Minto, Oletha R., MD, Aiken, SC Oh, Chien, MD, Columbia, MD
Miser, Alyson Deas, MD, Alexandria, VA Ohm, Mark, MD, Royal Oak, MI
Misra, Suman L., MD, Bloomfield Hills, MI Olaru, Gabriela Ana, MD, Port Washington, NY
Mitchell, Christopher Ray, MD, Rome, GA Olatinwo, Moshood O., MD, McDonough, GA
Mitchell, Leah S., MD, Lexington, KY Olvey, Kendall, MD, New Bern, NC
Moein, Sudabeh, MD, Palos Verdes, CA Ombalsky, Joseph, MD, Westerly, RI
Mohamed, Fazil A., II, MD, Hamer, SC Onibokun, Adedayo, MD, Watertown, NY
Mohammed, Decca, MD, Parsippony, NY Orris, John Joseph, DO, Chester Springs, PA
Molina-Millet, Leonardo, MD, San Juan, PR Ortega, Esteban, MD, Brooklyn, NY
Moncla, Paul Richard, MD, Henderson, NV Ortiz, B. Hannah, MD, Winter Park, FL
Mondestin, Myriam A.J., MD, Monroe, NJ Osborn, Alvadore Perry, DO, Urbandale, IA
Morgan, Natalie A., MD, Omaha, NE Osei, Solomon A., MD, Brooklyn, NY
Morgan, Sandra L., MD, Alpena, MI
Morrison, Jeffrey D., DO, Newport News, VA Packard, Alison Swift, MD, Jamaica Plain, MA
Moseman, Cher P., MD, Colorado Springs, CO Paczos, Tamera Ann, MD, Buffalo, NY
Page, Sarah M., MD, Beavercreek, OH Potts, Michael Earl, MD, Rogers, AR
Painter, Lauren J., MD, Columbia, SC Powell, Lana D., MD, Ann Arbor, MI
Pal, Lubna, MD, Orange, CT Powers, Christopher J., MD, El Paso, TX
Pali, Rozafa Luca, MD, Yonkers, NY Preen, Amy E., D.O., APO, AE
Palta, Vidya Nagaraj, MD, Port Washington, NY Priebe, Kimberly, MD, Durango, CO
Pan, Vivien Lin, MD, Ladera Ranch, CA Proctor, Rachael E., MD, Saint Joseph, MI
Panzarella, Cristin S., MD, Metairie, LA Provost, Holly, MD, Opelousas, LA
Paolilli, Joanna, MD, East Williston, NY Pugmire, Jonathan E., MD, Salem, OR
Papandrea, Lora L., MD, Utica, NY
Pare, Emmanuelle, MD, Philadelphia, PA Quezada, Oscar Rogiero, MD, El Paso, TX
Parekh, Mitesh, MD, Danville, PA Quimby, Jennifer C., MD, Silverdale, WA
Parker, Hannah, MD, Jamaica Plain, MA
Parmar, Madhu, MD, Clayton, NC Radpour, Chris, MD, Chattanooga, TN
Paroski, Sylvie H., MD, Coppell, TX Raid-Witthoeft, Laura J., MD, Dubuque, IA
Parungao, Lee Jennifer, MD, Chicago, IL Raiford, Karen G., MD, Decatur, AL
Pastoriza, Rajan A., MD, Jackson, MI Rajendran, Geetha Prabhavathy, MD, Stanfordville, NY
Patel, Amit I., MD, Litchfield Park, AZ Raju, Tanuja, MD, Folsom, CA
Patel, Ashlesha A., MD, Chicago, IL Raman, Rita S., MD, Mountain View, CA
Patel, Mayur V., MD, Edison, NJ Rampaul, Reginald C., MD, San Jose, CA
Patel, Shefali, MD, Cranford, NJ Rapp, Terri, MD, Lancaster, PA
Patton, Theresa M., MD, Dallas, TX Rardin, Charles Roswell, MD, Providence, RI
Paykel, Jacquelyn M. McLees, MD, Chesapeake, VA Rasbach, Adrienne, MD, Noblesville, IN
Pearlman, Larisa Marcie, MD, Smyrna, GA Raup, Aimee L., DO, Rockledge, FL
Pegram, George Vernon, III, MD, Ft. Payne, AL Reddy, Madhuri G., MD, Council Bluffs, IA
Pejovic, Tanja, MD, Portland, OR Reddy, Michelle Hutchison, MD, Seal Beach, CA
Pellini, Timothy M., MD, Columbus, GA Reddy, Radha B., MD, West Chester, OH
Peng, Noel, MD, Dallas, TX Redick, Dana L., MD, Charlottesville, VA
Perlman, Darcey Bromberg, MD, Carlsbad, CA Redman, Mark Edward, MD, West Bloomfield, MI
Perlman, Rachel K., MD, Menlo Park, CA Reed, Sharman Lynn, MD, Denver, CO
Perry, Shonda Y., MD, Wesley Chapel, FL Reese, Heather, MD, Winchester, VA
Persaud-Kraut, Arlene N., MD, Overland Park, KS Reffigee, Lester G., MD, Santa Maria, CA
Peterson, Matthew, MD, Blowing Rock, NC Refuerzo, Jerrie Selga, MD, Lathrup Village, MI
Pflum, Jeannie, D.O., Santa Rosa, CA Rice, Sharon M., MD, Florence, MA
Phair, Neva, MD, Boulder, CO Richlin, Spencer S., MD, Westpor, CT
Picardo, Carla, MD, Durham, NC Richter, Amy R,, MD, Port Jefferson, NY
Pieczonka, Sheila M., DO, West Seneca, NY Richter, Paula D., MD, San Clemente, CA
Pierre, Gamilah N., MD, Frankfort, IL Rindfuss, Richard T., MD, New York, NY
Pierre-Louis, Edna N., MD, Winterhaven, FL Roberts, Keith McDuffie, MD, Richmond, VA
Pinkerton, Nicole L., MD, Colorado Springs, CO Roberts, Shannon, MD, Jackson, WY
Piquion, Johann M., MD, Pittsford, NY Rodgers, Bennye D., Jr., MD, Zachary, LA
Pitt, Kara A., MD, Sherborn, MA Rodriguez, Consuelo, MD, Scarsdale, NY
Plastino, Kristen A., MD, San Antonio, TX Rosca, Anca, MD, Jersey City, NJ
Platt, Julie Suzanne, MD, Indianapolis, IN Rose, Shayla S., MD, Lake Saint Louis, MO
Plociennik, Krzysztof Zbigniew, MD, Stowe, VT Roth, Diana, MD, Kings Point, NY
Plotkin, Mindy Michelle, DO, Fremont, CA Rowland, Daniel Lee, MD, Mesa, AZ
Poggi, Sarah Hougen, MD, Chevy Chase, MD Runnels, Alexandra, MD, San Antonio, TX
Pollack, Jason A., MD, Phoenix, AZ Rushing, Rodney Scott, MD, Portland, OR
Pollydore, June Hazel, MD, Ferdinand, IN Russell, Jonathan R., MD, Tomball, TX
Pomeroy, Lisa Echt, MD, Warwick, RI Russell-McKesey, Camille A., MD, El Dorado, KS
Ponder, Jerome V., MD, Fayetteville, NC Russo, Joie D., DO, Tarzana, CA
Pool, Kathryn S., MD, Columbus, OH Ryan, Catherine A., MD, Boulder, CO
Porter, Marya, MD, Gretna, LA Ryntz, Timothy E., MD, Boston, MA
Pothuri, Bhavana, MD, New York, NY
Pott-Grinstein, Elisabeth Anna, MD, Saadat, Peyman, MD, Beverly Hills, CA
Lighthouse Point, FL Sabatini, Marie M., MD, South Nyack, NY
Potter, Megan Baldwin, MD, Griffin, GA Sabin, Sarah F., MD, Cincinnati, OH
Saint-Louis, Hedwige, MD, Birmingham, AL Skudlarek-Prete, Caroline Maria, DO, Newport, RI

Salamat Saberi, Naghmeh, MD, Foster City, CA Slade, Shenee, MD, San Jose, CA
Salazar, Charles, MD, Macon, GA Slager, Barbara Neily, MD, Gorham, ME
Salinger, Darren Scott, MD, Miami, FL Slepian, Jodi, MD, Wynnewood, PA
Sam, Lanalee Araba, MD, Coconut Creek, FL Small, Maria Jacqueline, MD, New Haven, CT
Sams, Joseph, MD, Jonesboro, AR Smith, Britte D., MD, Fayetteville, AR
Sanders, Marni, MD, Port Washington, NY Smith, Dayna J., MD, Elmsford, NY
Sanders, Stephen M., MD, Cedar City, UT Smith, Erica N., MD, West Bloomfield, MI
Sanford, Daniel Keith, MD, Olney, IL Smith, J. Sid, MD, Tuscaloosa, AL
Santiesteban, Joanna L., MD, Buena Vista, VA Smith, Patricia Amanda, MD, Houston, TX
Sarvis, April M., MD, Royal Oak, MI Smith, Sarah, MD, Pawleys Island, SC
Sawaged, Khalid S., DO, West Paterson, NJ Snowise, Saul, MD, Boulder, CO
Sayat, Jason G., MD, Columbus, OH Snyder, Amy, MD, Marlborough, CT
Schad, Todd A., MD, Prairie du Sac, WI Sohn-Woo, Christine, MD, Rosedale, MD
Schapker, Scott A., MD, Phoenix, AZ Sole, Kristina, MD, Cleveland Heights, OH
Schaub, Allyn G., MD, Chicago, IL Solnik, Meir Jonathon, MD, Newport Beach, CA
Schaufelberger, Adriana O., MD, Hales Corners, WI Solomon, Molham M., MD, Larchmont, NY
Scheve, Dawn, MD, Renton, WA Somersel, Gavin, MD, Albany, NY
Schimp, Veronica Lee, DO, Grosse Pointe Farms, MI Sonoda, Yukio, MD, New York, NY
Schindler, Lynnett, MD, Virginia Beach, VA Sorensen, Lisa Moser, MD, Santa Rosa, CA
Schubert, Daniel, MD, Eagle River, AK Sorkin, Leonid, MD, New York, NY
Schubert, Frank P., MD, Indianapolis, IN Soultanakis, Emmanuel Nicholas, MD, Shelburne, VT
Schulz, Jodi, MD, Hinesville, GA Souter, Irene C., MD, Dover, MA
Schwartz, Benjamin Michael, MD, Babylon, NY Spence, Mark Rohan, MD, Miami, FL
Schwertner, Belinda, MD, Georgetown, TX Spielvogel, Kenneth, MD, Myrtle Beach, SC
Scott, Jodi L., MD, Sioux Falls, SD Stallings, Shawn Patrick, MD, Signal Mountain, TN
Seale, Lindsey N., MD, Springdale, AR Stanfield, Amy Brame, MD, Durham, NC
Seals, David Edward, MD, Jersey City, NJ Stank, Melissa, MD, Superior, WI
Sehgal, Geeta, DO, Newton, NJ Steele, Sharon, MD, Lexington, KY
Seidel, Geoffrey K., MD, Cary, NC Stegman, Nicole G., MD, Sleepy Hollow, NY
Seiler, Tanya E., MD, Victoria, TX Stein, James Christopher, MD, Springfield, MO
Semple, Lisa, DO, Savannah, GA Sternberg, Holly, MD, Atlanta, GA
Sexton, Michael, DO, Pensacola, FL Stiller, Jeffrey D., MD, Shelbyville, IN
Shah, Aashish K., MD, Friendswood, TX Stoltz, Michelle, MD, Greeley, CO
Shah, Biren P., MD, Orland Park, IL Straughn, John Michael, Jr., MD, Birmingham, AL
Shahim, Ebrahim Shaun, MD, New York, NY Streitman, David Collier, MD, Vacaville, CA
Shepard, Deborah Kay, MD, Fenton, MI Strittmatter, Dennis S., MD, Mt. Sinai, NY
Shi, Susan Y., MD, New York, NY Styler, Marianne, MD, Teaneck, NJ
Shipman, Suzanne D., MD, Oregon City, OR Sukalich, Sara, MD, West Henrietta, NY
Shivvers, Stephan Allan, MD, Dallas, TX Sullivan, Anthony, MD, Colorado Springs, CO
Shobeiri, Seyed Abbas, MD, Edmond, OK Sun, Andrew Nicholas, MD, Eatontown, NJ
Shores, Sarah Bernice, DO, St. Louis, MO Sutler, Denise W., MD, Edgewood, KY
Shvartsman, Hyun Sook, MD, Redlands, CA Swaby, Stanley Stephen, MD, Yonkers, NY
Silas, Glen H., MD, Fairfax, VA Swenson, Kerry L., MD, Chicago, IL
Silverman, Karianne, MD, Long Beach, CA Sword, Andrew James, MD, Pittsburgh, PA
Siman, Deborah Cecilia, MD, Coral Gables, FL Swyers, Angela H., MD, Atlanta, GA
Simon, Beth J., MD, New York, NY Szekely, Joseph C., MD, Farmington, NM
Simons, Pamela D., MD, MBA, Colorado Springs, CO Szurkus, Dennis, MD, FPO, AE
Simpson, Cristina Paloma Carreno, MD,
San Antonio, TX Taghechian, Elizabeth Ramak, MD, Marietta, GA
Sims, Emma J., MD, The Woodlands, TX Tahtawi, Samira S., MD, Wilson, NC
Singh, Deepjot Kaur, MD, Redondo Beach, CA Talreja, Reena R., MD, Virginia Beach, VA
Sipe, Christopher Sheldon, MD, Iowa City, IA Tamura, Kimi, MD, Panorama City, CA
Sironi, Rindo R., MD, Butte, MT Tasillo, Dawn S., MD, Holden, MA
Skeete, Delisa A., MD, Ft. Lauderdale, FL Taylor, Christine, MD, Godfrey, IL
Skolnick, Sara A., MD, Austin, TX Taylor, Karen E., MD, Corona del Mar, CA
Tchuisse, Lucien-Maximin Yanou, MD, Walsh, Leigh, MD, Louisville, KY

Orange Park, FL Wang, Anne, MD, Chattanooga, TN
Tedjarati, Sean Shaheen, MD, Tampa, FL Wang, George Hai, MD, Monterey Park, CA
Terkildsen, Mary Faith C., MD, Highland Park, IL Wang, Lynn, MD, Newark, DE
Tewari, Krishnansu Sujata, MD, Irvine, CA Wang Natarajan, Margaret Pei-Ju, MD,
Thatcher, Kay C., MD, La Vista, NE San Marino, CA
Thibault, Amy B., MD, Essex Junction, VT Ward, Alisa K., MD, Richardson, TX
Thoma, Whitney Wolfe, MD, Chapin, SC Warner, Amy, MD, Mount Pleasant, SC
Thomas, Jeanette M., MD, Bloomington, MN Warner, Hiral R.P., MD, Indianapolis, IN
Thomas, Sandra, DO, Center Valley, PA Warner, Janet P., MD, Tucson, AZ
Thompson, Amy M., MD, Lubbock, TX Warner, Joan M., MD, Phoenix, AZ
Thompson, Jennifer Michele, MD, Texarkana, TX Warren, Lori M., MD, Elk Grove Village, IL
Thompson, Marcel Dwaine, DO, Harker Heights, TX Waterstone, Melissa B., MD, Paramus, NJ
Tidmore, Joi Robinson, MD, E. Cleveland, OH Weatherford, Shannon R., MD, Richmond, VA
Tijerino, Linda, MD, St. Petersburg, FL Webster, Gwen, MD, Plano, TX
Tillmanns, Todd David, MD, Memphis, TN Weinrich, Matthew A., MD, Waconia, MN
Tolaymat, Lama Linda, M.D, M.P.H, Jacksonville, FL Weiss, Joshua Lawrence, MD, Irving, TX
Tornatore, Jean M., MD, New York, NY Wells, Dylan R., MD, Memphis, TN
Tortoriello, Drew Vito, MD, Jersey City, NJ Wepman, Carolyn Jo, MD, Louisville, CO
Tran-Hoppe, Ngoc Quynh, MD, Edison, NJ Weyhrich, Darin L., MD, Boise, ID
Traynor, Jeffrey Dennis, MD, Walnut Creek, CA White, Karyn E., MD, Atlanta, GA
Truitt, Amy Elizabeth, MD, Metairie, LA White, Katharine S., MD, Mount Pleasant, SC
Tsatsas, Manolis T., MD, New York, NY Wiedel, Lisa M., MD, Irvine, CA
Tsong, Shirley W., MD, Eatontown, NJ Wieland-LeBel, Laura Danielle, MD,
Tsui, Timothy, MD, Los Angeles, CA Simpsonville, SC
Tu, Amy, MD, Bellingham, WA Wiggins, Shawn M., MD, Flushing, MI
Tugwell, Terence Randolph, MD, Monroe, LA Williams, Judith, MD, Collierville, TN
Turner, Francesca M., DO, Chicago, IL Wilson, Richard W., MD, Gilbert, AZ
Tweedy, Kathryn Ann, MD, Minneapolis, MN Winn, Virginia Dragone, MD, San Francisco, CA
Winter, William Edward, III, MD, San Antonio, TX
Urban, Michele Marie, MD, Salisbury, MD Wise, Elizabeth N., DO, Erie, PA
Urribarri, Nisseth Josefina, MD, Coral Springs, FL Wolfe, Michelle D., MD, Southfield, MI
Usadi, Rebecca S., MD, Simpsonville, SC Wong, Lily, MD, New York, NY
Ushigome, Mae M., MD, Santa Monica, CA Wood, Duke J., DO, Amory, MS
Wood, Kay L., MD, San Dimas, CA
Vale-Moreno, Yari, MD, Forest Hills, NY Woodring, Thomas C., MD, Jackson, MS
Van Fossen, Katharine Daniels, MD, Germantown, WI Woods, Grayson N., MD, Nashville, TN
Van Savage, Katherine, MD, Holden, MA Woolard, Dianne S., MD, Cape Girardeau, MO
Van Tuyle, Jessica L., MD, Richmond, VA Worster, Tasha, MD, Portland, ME
Vance, Amy Weaver, MD, Asheville, NC Worth, Jaqueline M., MD, New York, NY
Vargas, Juan Edmundo, MD, San Francisco, CA Wotherspoon, Barbara, MD, Germantown, TN
Varkey, Taz, MD, Los Angeles, CA Wright, Cameual N., MD, Schererville, IN
Ventolini, Gary, MD, Centerville, OH Wright, Kimberly E., MD, Naperville, IL
Vineyard, David D., MD, Nacogdoches, TX Wu, Sidney S., MD, New York, NY
Visintine, John, III, MD, Rosebud, SD Wuebker, Heather E., MD, St. Louis, MO
Vo-Hill, Hien T., MD, Plover, WI Wynne, Gracia Lenore, MD, Houston, TX
Von Wald, Tiffany, MD, Belmont, MA
Vongngarm, Saovaros, MD, Las Vegas, NV Yuan, Frances Huichi, MD, Kensington, CA
Wade, Christine T., MD, Pontiac, MI Zakhary, Atef Saad, MD, Tampa, FL
Wagar, David A., MD, Newport News, VA Zapantis, Gregory, MD, Syosset, NY
Wagner, Andrew S., MD, Saginaw, MI Zimmerman, Brett, DO, Toledo, OH
Wahib, Samir Ali, DO, Fairlawn, OH Zimmerman, Carolyn Delores, MD, Glenside, PA
Waliser, Thomas J., MD, Denver, CO Zinner, Tracy Rebecca, MD, Brookline, MA
Walker, Linda K., MD, Washington, DC Zollicoffer, Carl Donnell, MD, Port St. Lucie, FL
Wallace, Patricia A., MD, Lakeforest, CA Zoneraich, Nathaniel, MD, Vienna, VA

American Journal of Obstetrics & Gynecology (AJOG) - Sep05

Uploaded by

Copyright:

Available Formats

American Journal of Obstetrics & Gynecology (AJOG) - Sep05

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

American Journal of Obstetrics & Gynecology (AJOG) - Sep05

Uploaded by

Copyright:

Available Formats

Founded in 1920

TABLE OF CONTENTS September 2005

High-dose methadone maintenance in pregnancy: Maternal and neonatal outcomes 606

Contents continued on page 5A

Fetal response to maternal methadone administration 611

RESEARCH METHODS: STATE OF THE SCIENCE

Contents continued on page 6A

GENERAL OBSTETRICS AND GYNECOLOGY: GYNECOLOGY

First glimpse of the functional benefits of clitoral hood piercings 675

Arcus tendineus fascia pelvis: A further understanding 677

A national probability survey of American Medical Association gynecologists and 693

GENERAL OBSTETRICS AND GYNECOLOGY: OBSTETRICS

Chorioamnionitis with a fetal inflammatory response is associated with higher 708

Early discharge from obstetrics-pediatrics at the Hospital de Valme, with 714

Contents continued on page 8A

Impact of maternal-fetal surgery for myelomeningocele on the progression of 727

Isolated fetal pyelectasis and chromosomal abnormalities 732

Is zygosity or chorionicity the main determinant of fetal outcome 757

The involvement of Rho-associated kinases in agonist-dependent contractions of 815

Novel peptides prevent alcohol-induced spatial learning deficits 825

Heat shock protein-70 and 4-hydroxy-2-nonenal adducts in human 836

Cloning and cellular expression of aquaporin 9 in ovine fetal membranes 841

10A September 2005

An excerpt from the American Journal of Obstetrics and Gynecology

Clinicopathologic features of six cases of primary cervical lymphoma 866

Port site ischemic necrosis: An unforeseen complication of laparoscopic surgery 873

Spontaneous epidural hematoma of the spine in pregnancy 875

Contents continued on page 12A

September 2005 11A

Prenatal diagnosis of amniotic sheets by magnetic resonance imaging 881

Posterior reversible leukoencephalopathy in a case of postpartum eclampsia 885

Twin-to-twin transfusion syndrome at 11 weeks of gestation 887

Spontaneous closure of the hymen during pregnancy 889

LETTERS TO THE EDITORS

12A September 2005

Cytokines, preeclampsia, and uterine denervation? 896

Accidental fetal lacerations during cesarean delivery: Experience in an Italian 897

2004 Board Certificates e1

September 2005 13A

Pediatric presence at cesarean section: Justified or not?

Department of Neonatal Medicine, Royal Prince Alfred Hospital, Sydney, Australia

Table I Total cohortddemographic, presentation, and resuscitation details

Table II SVD and elective CSdeffect of fetal distress on resuscitation

Table III SVD and elective CSdeffect of presentation on resuscitation

High-dose methadone maintenance in pregnancy:

Table Infant outcomes by maternal methadone dose group

Fetal response to maternal methadone administration

decline, women at trough levels were experiencing

obscured by the nature of the design, subsequent

RESEARCH METHODS: STATE OF THE SCIENCE

A new series of articles on research methodology

* Reprint requests: George A. Macones, MD, MSCE, Division of

Issues in clinical trial design: Stopping a trial early

recommend calculating the conditional power assuming Summary

Table Comparison of Conventional and Large and Simple Clinical Trials

Research agenda for preterm birth: Recommendations

economic costs are staggering. Although only 9% of all

gation should include consideration of the broad range of

Centers for Disease Control (CDC), and Society of