EUROPEAN JOURNAL OF CANCER x x x ( 2 0 1 1 ) x x x –x x x

available at www.sciencedirect.com

journal homepage: www.ejconline.com

Hormonal and reproductive factors in relation to melanoma

in women: Current review and meta-analysis

Sara Gandini a,*, Simona Iodice a, Els Koomen b, Alessandra Di Pietro c,

Francesco Sera d, Saverio Caini e
a
Division of Epidemiology and Biostatistics, European Institute of Oncology, Via Ripamonti 435, 20141 Milan, Italy
b
Department of Clinical Pharmacy and Toxicology, Leiden University Medical Center, Albinusdreef 2, 2300 RC Leiden, The Netherlands
c
Division of Sarcoma and Muscolo Cutaneous Melanoma, European Institute of Oncology, Via Ripamonti 435, 20141 Milan, Italy
d
Medical Research Council Centre of Epidemiology for Child Health, University College London Institute of Child Health, London, UK
e
Molecular and Nutritional Epidemiology Unit, ISPO, Scientific Institute of Tuscany, Via di San Salvi 12, 50135 Florence, Italy

A R T I C L E I N F O A B S T R A C T

Article history: A number of studies have focused on possible relationships between characteristics of
Available online xxxx female endocrine status and melanoma (CM) risk; however, the link between melanoma,
oral contraceptive (OC) and hormonal replacement therapy (HRT) use, and reproductive
Keywords: factors remains controversial. A comprehensive, systematic bibliographic search of the
Hormones medical literature was conducted to identify relevant studies. Random effects models were
Oestrogens used to summarise results. Subgroup, meta-regression and sensitivity analyses have been
Menopause carried out to explore sources of between-study variation and bias. We included thirty-six
Menarche observational studies published in the last 30 years. Summarising a total of 5626 melanoma
Parity cases, we did not find any significant melanoma risk associated with OC and HRT use. Sev-
Education eral reproductive factors were also investigated, summarising data on 16787 melanoma
Melanoma cases. We found a significantly increased melanoma risk for late age at first birth, and
Meta-analysis women with more than one child may be at a lower risk for melanoma; however, socio-eco-
nomic confounders were found to play a significant role in explaining this association. This
study confirmed no increased risk of CM with the use of oral contraceptives and hormone
replacement therapy: exogenous female hormones do not contribute to an increased risk of
CM. In contrast, significant associations of CM with parity and age at first pregnancy were
observed in this meta-analysis finds and warrant further research.
 2011 Elsevier Ltd. All rights reserved.

1. Introduction OC use and reports of nevi darkening and enlarging during

pregnancy.1–3
A number of epidemiological studies from the 1980s focused Additionally, animal experiments have shown increases
on possible relationships between characteristics of female in melanin production or melanocyte number in associa-
endocrine status and melanoma (CM) but substantial contro- tion with oestrogens and oestrogens/progesterone combi-
versy exists on potential hormonal effects on melanoma risk. nations.4,5
The original concerns arose from some case reports of CM in Moreover, the influence of female sex steroids on mela-
pregnancy and the observation of hyperpigmentation during noma is supported by several observational studies. Incidence

* Corresponding author: Tel.: +39 02 57489819; fax: +39 02 57489922.

E-mail address: sara.gandini@ieo.it (S. Gandini).
0959-8049/$ - see front matter  2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ejca.2011.04.023

Please cite this article in press as: Gandini S et al., Hormonal and reproductive factors in relation to melanoma in women: Current review and
meta-analysis, Eur J Cancer (2011), doi:10.1016/j.ejca.2011.04.023
2 EUROPEAN JOURNAL OF CANCER x x x ( 2 0 1 1 ) x x x –x x x

trend studies have shown that until the age of 45 years, CM tencies and conflicting results due to imprecise estimation
incidence rates in women exceeded those in men, after which of risk (e.g. due to small sample size), inadequate
the incidence rates in men rose markedly but levelled off in study design or adjustments for potential confounding
women, which may suggest hormonal or reproductive influ- factors.
ences.12–15 In this meta-analysis we will investigate associations be-
Although the average age at diagnosis of CM is around tween the incidence of cutaneous melanoma and the use of
60 years, about one-third of all CM in women occurs during exogenous hormones, oral contraceptives (OC), hormonal
their childbearing age and, in women aged 25–29 years, CM replacement therapy (HRT), reproductive factors, including
is the most common malignancy.16 Moreover, women with a age at menarche, fertility, use of fertility drugs, parity, meno-
history of breast cancer have been reported to be at higher pausal status and age at menopause.
risk for CM and vice versa.14,17–20
A number of studies from the 1980s focused on possible
relationships between characteristics of female endocrine 2. Materials and methods
status and the risk of CM. 21,22,39,40,41 The studies included
questions about the use of oral contraceptives (OC), hormone We planned, conducted and reported this systematic litera-
replacement therapy (HRT), parity and age at first child, age at ture search and review following MOOSE guidelines regarding
menarche, menopausal status and age at menopause or infer- meta-analysis of observational studies.28
tility drugs that reflect influences of exogenous and endoge-
nous hormones, respectively. However, the link between CM 2.1. Data sources and search strategy
and pregnancy, hormonal and reproductive factors remains
controversial. We reviewed published reports using validated search strate-
Cancers sensitive to female sex steroids are associated gies29,30 from these databases:
with several risk factors, 6–11 such as low parity, infertility,
early age at menarche, and late age at menopause. These fac- • PUBMED (http://www.ncbi.nlm.nih.gov/entrez/query.fcgi).
tors frequently coexist in infertile patients and some studies • Ovid MEDLINE database (1950 to July 2009)
suggested that different infertility causes could be involved • ISI Web of Science Science Citation Index Expanded (SCI
in cancer risk development.23 Furthermore, the influence of Expanded)
fertility drugs on malignant melanoma risk has not been
widely studied despite the wide use of this group of exoge- The following keywords and/or corresponding MESH
nous hormones over the last 30 years and their recognised ef- terms were used: melanoma, skin cancer, oral contraceptive,
fect on ovulation and endogenous hormone production.24 hormonal replacement therapy, oestrogens, pregnancy,
Given the increasing number of prescriptions of fertility drugs parity, fertility, menarche and menopause. Additional studies
among infertile couples over the last 30 years, the question of were collected through cross-referencing of reference lists of
whether fertility drugs may increase malignant melanoma the retrieved articles and preceding reviews and conference
risk is of great public health concern. abstracts on the topic. The search was limited to human stud-
The aim of this paper is to review these issues using meta- ies. No language or time restrictions were applied.
analyses and evidence-based approaches. Dose–response
models through meta-analytic approach allow investigation 2.2. Data extraction
of different types of exposure: duration, age at exposure and
time since first and last exposure. Data were extracted and cross-checked independently by two
Two meta-analyses and a pooled-analysis were published investigators (S.C. and S.G). Any disagreement was resolved
on OC use and a pooled-analysis on reproductive factors by consensus among them. The following information from
investigating the associations with CM risk.25–27 Since these the published papers was extracted and coded: authors, year
analyses, additional evidence has become available (include of publication, study period, type of study design, country,
Refs. of studies not included in the above analyses). Moreover, number of cases, number and sources of controls, features
previously published analyses did not include cohort studies, of included subjects or populations, use of matching in study
small studies and/or publications based on information from design, definitions of exposure, and adjustments used for sta-
postal questionnaires. tistical analyses.
To update previous meta-analyses and in order to also in- We used wide inclusion criteria in order to select studies
clude cohort studies, small studies and/or publications based with necessary information and to be able to investigate be-
on information from postal questionnaires, we carried out an tween-study variability:
evidence-based systematic review and meta-analysis of all
the literature to obtain an overall picture considering both (1) Study publications should contain the minimum infor-
exogenous and endogenous hormones, including evaluations mation necessary to obtain comparable risk estimates
on the effect of HRT use, menopausal status and menarche. and corresponding 95% confidence intervals (i.e. odds
Using dose–response models has allowed the investigation ratios or relative risks and a measure of uncertainty:
of different types of exposure: duration, age at exposure and standard errors, variance, confidence intervals or exact
time since first and last exposure. By including all possible P-value of the significance of the estimates).
studies we aim to avoid selection bias, publication bias and (2) The studies should be independent to avoid giving dou-
to quantitatively explore any possible similarities, inconsis- ble weight to a single study. In case of multiple reports

Please cite this article in press as: Gandini S et al., Hormonal and reproductive factors in relation to melanoma in women: Current review and
meta-analysis, Eur J Cancer (2011), doi:10.1016/j.ejca.2011.04.023
 EUROPEAN JOURNAL OF CANCER x x x ( 2 0 1 1 ) x x x –x x x 3

of the same study, we considered the estimates from 3. Results of meta-analysis

the most recent publication.
The analysis based on 36 independent publications between
If multiple risk estimates were available for one item with- 1977 and 2009 was conducted on two main groups of risk fac-
in a study, for example several estimates for different types of tors: exogenous hormones use and reproductive factors. To
exposure measures or sources of data (e.g. types of drug fertil- evaluate the effect of oestrogens we considered OC and HRT
ity or data from questionnaires and from GPs), we included use, time since first and last use, and age at first OC use. To as-
only one of those estimates, and chose to include the RR esti- sess the effect of reproductive factors we investigated parity,
mate showing the strongest association. In sensitivity analy- age at first pregnancy, age at menarche, menopausal status
ses, we evaluated these choices. and age at menopause, and history of problems of fertility.
We retrieved 25 estimates:

2.3. Definition of the outcome and exposures

– Twenty-two on OC and 10 on HRT ‘ever use’, time and
age since first and last exposure and duration (two
The main outcome of this systematic meta-analysis was risk
studies did not publish the estimate for ‘ever use’);
estimates for incidence of histological confirmed CM.
– Twenty-four on reproductive factors (18 on parity, 8 on
We retrieved exposure data for: ever OC and HRT users but
age at first pregnancy, 7 on evaluation for infertility, 3
also dose–response estimates for duration of use, age and
on fertility drugs, 6 on menopausal status and 3 on age
time since first use and since last use. We also retrieved data
at menopause, 7 on age at menarche).
for parity, age at first child (or pregnancy), menopausal status,
age at menopause, age at menarche, exams for fertility and
use of fertility drugs. Finally, we reviewed and extracted risk We found 6 cohort studies for exogenous hormones use
estimates for the association between pregnancy and CM and 7 for reproductive factors. Among the remaining 16
recurrence and mortality. case-control studies on oestrogens, 11 were population based;
among the 17 case-control studies on reproductive factors, 12
were population based. From Europe, we retrieved 10 esti-
2.4. Statistical analysis mates for exogenous hormones use and 9 for reproductive
factors (Tables 1 and 2).
All measures of association and the corresponding confidence Summary estimates for different types of exposure (‘ever’
intervals, adjusted for the maximum number of confounding versus ‘never’, ‘highest’ versus ‘lowest’ categories, dose–re-
variables available, were translated into log relative risk, and sponse and time-dependent evaluations) were presented in
corresponding variance, with the formula proposed by Table 3.
Greenland.31 Forest plots are presented for the risk factors and compar-
We used random effects models, taking into account be- isons with the highest number of studies: ‘ever use’ of OC and
tween-study and within study variability when more than HRT and parity.
one estimate from a single study was used. Summary relative
risks (SRR) were obtained from maximum likelihood estimate: 3.1. Exogenous hormones’ use
PROC MIXED in SAS.32
Homogeneity of effects across studies was assessed using This meta-analysis, evaluating a total of 5626 CM cases with
the Chi-square statistic and quantified by I2, which represents information on exogenous hormones’ use, confirmed that OC
the percentage of total variation across studies that is attrib-
utable to heterogeneity rather than chance.33,32
Since the Chi-square test has limited power, we consid-
ered statistically significant heterogeneity at the P = 0.10 level
of association. Between-study heterogeneity was investigated
by sub-group analyses and meta-regressions.34
For dose–response, duration and time-dependent esti-
mates we extracted RRs and frequencies for each category
of exposure and we used linear models, within each study,
taking into account correlation within RRs having the
same reference category. We obtained the summary RR
pooling the study-specific estimates by random effects
models.35
Sensitivity analysis was carried out to evaluate whether
results were influenced by a single study. A funnel-plot-based
approach was used for assessing publication bias evaluating
regression of log(RR) on the sample size, weighted by the in-
verse of the variance.36
We presented results exploring for any publication bias if
heterogeneity (I2) was not larger than 60% and if studies with
significant results were present. Fig. 1 – Forest plot of OC use and melanoma risk.

Please cite this article in press as: Gandini S et al., Hormonal and reproductive factors in relation to melanoma in women: Current review and
meta-analysis, Eur J Cancer (2011), doi:10.1016/j.ejca.2011.04.023
4 EUROPEAN JOURNAL OF CANCER x x x ( 2 0 1 1 ) x x x –x x x

Table 1 – Included studies for oral contraceptive (OC) and hormone replacement therapy (HRT).
First author, Study Country N. cases N. contr.a Match Source of OC or HRT Adjust. for Adjust. for
PY design controls estimatesb pheno-photo sun-
typesc exposured

Beral, 197766 CC US 37 74 Yes Hosp OC

Adam, 198167 CC UK 169 507 Yes Pop OC
Bain, 198268 NCC US 141 2820 Pop. OC
Holly, 198369 CC US 87 863 Yes Pop OC, HRT
Beral, 198470 CC Australia 287 574 Yes Hosp OC, HRT
Holman, 198471 CC Australia 276 276 Yes Pop OC Yes Yes
Green, 198572 CC Australia 91 91 Yes Pop OC
Osterlind, 19883 CC Denmark 280 536 Yes Pop HRT Yes Yes
Adami, 198973 Cohort Sweden 31 23244 HRT
Zanetti, 199074 CC Italy 186 205 Yes Pop OC Yes Yes
Lê, 199275 CC France 91 149 Yes Hosp OCe Yes Yes
Palmer, 199276 CC US 615 2107 Yes Hosp OC Yes
Zaridze, 199277 CC Russia 96 96 Yes Hosp OC Yes
Holly, 199446 CC US 452 930 Yes Pop OC, HRT
Persson, 199678 Cohort Sweden 60 22597 HRT
Smith, 199879 CC US 308 233 Yes Pop OC Yes Yes
Westerdahl, 199680 CC Sweden 400 640 Yes Pop OCe Yes Yes
Feskanich, 199944 Cohort US 146 183693 OC Yes Yes
Young, 200181 NCC Australia 14 93 OC
Freedman, 200382 Cohort US 207 54045 OC, HRT Yes Yes
Naldi, 200583 CC Italy 316 308 Yes Hosp OC, HRT Yes Yes
Vessey, 200684 Cohort UK 94 17032 OC
Hannaford, 200785 Cohort UK 146 28762 OC
Lea, 200786 CC US 318 395 Yes Hosp OC, HRT Yes
Koomen, 200937 CC The 778 4072 Yes Pop OC, HRT
Netherlands
CC: case-control study. NCC: nested case-control study (case-cohort studies were codified as NCC). PY: publication year. Hosp: Hospital-based
controls. Pop: population-based controls. Adjust. RRs adjusted for confounders.
a
For cohort study we presented the overall study size.
b
Including ‘ever use’, duration, age at first use and time since first and last use.
c
Including naevi counts.
d
Including sunburns.
e
The authors reported estimates only for OC duration.

and HRT are not associated with a significant increase in CM showed an increase in risk of 10% for first pregnancy at
risk (Table 3). SRR for ‘ever use’ was: 1.04 (95%CI: 0.92, 1.18; het- 30 years of age, compared to pregnancy at 20 years of age or
erogeneity P = 0.03) for OC and 1.16 (95%CI: 0.93, 1.44; heteroge- earlier.
neity P = 0.05) for HRT. Dose–response models also indicated Among reproductive factors the significant heterogeneity
no significant increases in CM risk for OC or HRT use (for each was found only for parity and age at first birth (Chi-square
10 years of use of OC: 1.10, 95%CI: 0.88, 1.37; and for 5 years of P = 0.03 and 0.06, respectively).
use of HRT: 1.36, 95%CI: 0.65, 2.84). Furthermore, the summary
estimates of age at first OC use and time since first and last OC 3.3. Heterogeneity and sensitivity analyses
confirmed no effect of OC on CM risk.
Only four studies, evaluating OC, presented mean age of Investigation of heterogeneity was carried out for factors with
cases: three out of four were below 40 years, and only one the highest number of studies that allowed subgroup and
was 54 years.37 meta-regression analyses, looking at all the factors that could
Forest plots for OC and HRT ‘ever use’ are presented in have an impact on the between-study variability.
Figs. 1 and 2. In Table 4 we presented factors that explained some heter-
ogeneity. Study design and country significantly explained
3.2. Reproductive factors heterogeneity for parity. Cohorts and population based case-
control studies presented significantly lower estimates
Several reproductive factors were also investigated, summa- (P = 0.004) than hospital-based case-controls studies: for wo-
rising data on 16787 CM. Summary estimates for menopausal men with 2 children in population-based studies we obtained
status, age at menopause, age at menarche and exams for fer- a reduction of 11% (95%CI: 6–16%) compared to nulliparous
tility showed that these factors were not associated with CM women, whereas hospital-based studies suggest a significant
(Table 3). increase in risk. Furthermore, in European countries we
The only factor that at a first look seemed significantly observed significantly lower estimates than the other
associated with CM risk is age at first pregnancy which countries (P < 0.001); we estimated a significant 15% (95%CI:

Please cite this article in press as: Gandini S et al., Hormonal and reproductive factors in relation to melanoma in women: Current review and
meta-analysis, Eur J Cancer (2011), doi:10.1016/j.ejca.2011.04.023
meta-analysis, Eur J Cancer (2011), doi:10.1016/j.ejca.2011.04.023
Please cite this article in press as: Gandini S et al., Hormonal and reproductive factors in relation to melanoma in women: Current review and

Table 2 – Included studies for reproductive factors.

b
First author, PY Study Country N. cases N. contr. Match Source Age at Parity Age at first Fertility Menop. Adjust. for Adjust. for Adjust
design controls menarche pregnancy examsa pheno- sun- for educ.
photo typesc exposured

Holly, 198369 CC US 87 863 Yes Pop X X Yes

Holman, 198471 CC Australia 276 276 Yes Pop X X

EUROPEAN JOURNAL OF CANCER

Gallagher, 198587 CC Canada 361 361 Yes Pop X Yes Yes
Green, 198572 CC Australia 91 91 Yes Pop X X X Yes Yes
Osterlind, 19883 CC Denmark 280 536 Yes Pop X X X X Yes Yes
Brinton, 198988 Cohort US 4 2335 X
Wyshak, 198989 NCC US 18 72 Pop X
Zanetti, 199074 CC Italy 186 205 Yes Pop X X Yes Yes
Lê, 199275 CC France 91 149 Yes Hosp X X Yes Yes
Holly, 199446 CC US 452 930 Yes Pop X X
Holly, 199590 CC US 452 930 Yes Pop X Yes
Rossing, 199591 NCC US 12 135 Hosp X
Lambe, 199692 CC Sweden 4779 23888 Yes Pop X X
Westerdahl, 199680 CC Sweden 400 640 Yes Pop X Yes Yes
Modan, 199893 Cohort Israel 8 2496 X

x x x ( 2 0 1 1 ) x x x –x x x
Smith, 199879 CC US 308 233 Yes Hosp X X X Yes Yes
Young, 200181 Cohort Australia 14 3186 X X X X
Freedman, 200382 Cohort US 153 54045 X X X Yes Yes Yes
Neale, 200594 Cohort Sweden 2285 1234967 X X
Althuis, 200595 Cohort US 42 8422 X
Naldi, 200583 CC Italy 316 308 Yes Hosp X X X Yes Yes Yes
Kaae, 200796 Cohort Denmark 5688 1725627 X
Lea, 200786 CC US 318 395 Yes Hosp X Yes Yes
Hannibal, 200897 NCC Denmark 112 1226 Pop X X
CC: case-control study. NCC: nested case-control study (case-cohort studies were codified as NCC). Hosp: Hospital-based controls. Pop: population-based controls. Menop: Menopausal status. PY:
publication year. Adjust. RRs adjusted for confounders. Ca. cases. Co. controls, but for cohort study we presented the overall study size. Educ. Educational factors.
a
Including use of fertility drugs.
b
Including age at menopause.
c
Including nevi counts.
d
Including sunburns.

5
6 EUROPEAN JOURNAL OF CANCER x x x ( 2 0 1 1 ) x x x –x x x

Table 3 – Summary relative risks estimates for CM in women.

N. studies Risk factors SRR (95%CI) I2

Oral contraceptive (OC)

20 Ever versus never 1.04 (0.92, 1.18) 42
17 Duration: (each year of use) 1.01 (0.99, 1.03) 45
3 Age at first use: ‘ P 25 y’ versus never 1.03 (0.91, 1.17) 0
3 Age at first use (each year of older age) 1.01 (0.99, 1.02) 0
4 Time since first use: ‘ P 20’ versus never 0.90 (0.46, 1.78) 74
4 Time since first use (each year since first use) 1.00 (0.97, 1.02) 74
6 Time since last use: ‘ P 10 y’ versus never 0.85 (0.53, 1.38) 76
6 Times since last use: (each year since last use) 0.99 (0.97, 1.02) 76
Hormonal replacement therapy (HRT)
10 ‘Ever use’ versus never 1.16 (0.93, 1.44) 47
4 Duration: (each year of use) 1.06 (0.92, 1.23) 63
Reproductive factors
7 Age at menarche: (each year of older age at menarche) 1.01 (0.99, 1.02) 19
7 Evaluation for fertility: ever versus never 1.23 (0.87, 1.75) 33
3 Use of fertility drugs 1.17 (0.57, 2.39) 17
8 Age at first pregnancy: (each year of older age at first pregnancy) 1.011 (1.008, 1.013) 51
18 Parity (one more child) 0.98 (0.91, 1.06) 55
6 Menopausal status: pre versus post 1.09 (0.78, 1.52) 0
3 Age at menopause: (each year of older age at menopause) 1.00 (0.97, 1.02) 34
RRs for +1 y are summary estimates from dose–response models. Median reference categories among studies were: 45 years for age at
menopause; 20 years for age at I pregnancy; 12 for age at menarche.

than adjusted estimates (P = 0.09): again the suggestion for an

association comes mainly from unadjusted estimates.
Sensitivity analysis showed that heterogeneity for ‘ever
use’ of OC is mainly due to a Russian hospital based case-
control study38 that published a highly significant protective
effect for OC ‘ever use’: OR = 0.04 (95%CI: 0.003, 0.53). Exclud-
ing this publication, the summary estimate does not change:
SRR = 1.05 (95%CI: 0.92, 1.19) and the heterogeneity decreased
considerably (Chi-Square P = 0.09 and I2 = 33).
If more than one estimate was presented by the authors, we
carried out sensitivity analyses to include the available risk
estimates. This did not change any of the conclusions. For
example, for fertility drug use we did not observe an increased
risk and no heterogeneity was present: SRR = 0.95 (95%CI: 0.43,
2.13) with I2 = 0. For all end-points evaluated, none of the fac-
tors was associated with between-study heterogeneity; how-
Fig. 2 – Forest plot of HRT use and melanoma risk. ever, it is interesting to note that SRR for HRT ‘ever use’
adjusted for sun exposure and/or phenotypical factors were
lower than estimates not adjusted for those factors:
11–18%) decrease in risk for women with 2 children in Euro- SRR = 1.05 (95%CI: 0.76, 1.44) for the 4 adjusted estimates and
pean countries compared to nulliparous women. No reduc- SRR = 1.22 (95%CI: 0.97, 1.52) for the unadjusted estimates,
tion was observed for the non-European countries. although this difference was not statistically significant
Looking at the types of adjustments we found that esti- (P = 0.45). On the other hand, the SRR for the 7 estimates on par-
mates for parity, adjusted for education, one of the most ity adjusted for sun exposure were virtually identical to the one
important confounder for parity, were significantly greater summarising the unadjusted estimates: SRR = 0.94 (95%CI:
(P = 0.03) suggesting no association with melanoma: the indi- 0.88, 1.01) and SRR = 0.97 (95%CI: 0.93, 1.00), respectively.
cation for an effect comes mainly from unadjusted estimates. No indication for publication bias was found (P = 0.25 for
Forest plot for parity is presented by subgroups consider- OC ‘ever use’; P = 0.87 for HRT use; P = 0.72 for age at first preg-
ing factors explaining between-study heterogeneity (Fig. 3). nancy and P = 0.78 for parity).
None of the factors investigated significantly influenced
risk estimates of age at first birth. 4. Conclusion
The only factor that explained some variability in the esti-
mates for OC (‘ever use’) was adjustments for confounders: In this meta-analysis, we could not confirm an increased CM
unadjusted relative risks were borderline significantly greater risk with the use of oral contraceptives and hormone replace-

Please cite this article in press as: Gandini S et al., Hormonal and reproductive factors in relation to melanoma in women: Current review and
meta-analysis, Eur J Cancer (2011), doi:10.1016/j.ejca.2011.04.023
 EUROPEAN JOURNAL OF CANCER x x x ( 2 0 1 1 ) x x x –x x x 7

Fig. 3 – Forest plot of parity and melanoma risk.

Table 4 – Heterogeneity analyses: summary relative risks estimates by subgroup analyses.

N. studies Subgroups SRR 95%CI P-value

Parity (+1 child)

4 Hospital-based CC 1.10 (1.00, 1.22) 0.004
14 Pop. based CC or cohorts 0.94 (0.92, 0.97)
9 In Europe 0.92 (0.90, 0.94) <0.001
9 Other countries 1.00 (0.96, 1.04)
12 RRs unadjusted for education 0.93 (0.91, 0.95) 0.003
6 Adjusted for education 1.00 (0.96, 1.04)
Oral contraceptive (‘ever use’)
15 Adjusted RRsa 1.00 (0.88, 1.14) 0.09
5 Crude RRs 1.32 (0.97, 1.79)
CC: case-control study. P-values for the difference between groups obtained from meta-regression analysis.
a
Adjustment for confounding factors.

ment therapy. To conclude: exogenous female hormones do viewed all the evidence using meta-analyses to summarise
not significantly contribute to the risk of developing CM. Simul- the estimates and understand variations in results. We
taneously, this meta-analysis shows significant associations wanted to update previous meta-analyses on OC and
between CM risk and parity as well as age at first pregnancy. pooled-analysis on reproductive factors and present an over-
However, this may be caused by socio-economic differences. view of the evidence about the effects of endogenous and
65
Further research is required among women with more than exogenous hormones on the development of CM.
one child and early pregnancy who may be at lower risk for Two previous meta-analyses and a pooled analysis of
CM either causally or due to confounding biases. case-control studies concluded that OC use does not affect a
woman’s risk of CM.25–27
5. Discussion Our summary estimates, updating the previous meta-
analyses, confirmed their conclusions: no increased risk of
A number of studies suggest a potential role of female hor- CM was found with ever-use of OC, duration of use, age at first
mones in the pathogenesis of melanoma among women. use, and time since first and last use.
The studies included questions about the use of OC and post- The main limitation of this analysis on CM and OC use is
menopausal oestrogen therapy, parity and age at menarche, that we could not take into account oral contraceptive formu-
menopause, and first childbirth, which reflect influences of lations used: during the 80s OCs deeply differed from the ones
exogenous and endogenous hormones. This analysis re- used later on, e.g. low oestrogen, triphasic. However, in the

Please cite this article in press as: Gandini S et al., Hormonal and reproductive factors in relation to melanoma in women: Current review and
meta-analysis, Eur J Cancer (2011), doi:10.1016/j.ejca.2011.04.023
8 EUROPEAN JOURNAL OF CANCER x x x ( 2 0 1 1 ) x x x –x x x

Nurses Health Study44 an elevated non-significant risk of mel- with fertility problems, and possibly taking some treatments.
anoma was found in relation to the current but not past use of Findings from the present meta-analysis confirmed previous
oral contraceptives and when we investigated the influence of results: fertility investigations/treatments were not found to
publication year we did not find any significant trend. be significantly associated with increased CM risk.
In order to verify if an increased RR for use of OC could be The issue of the relationship between hormones and can-
identified for specific subgroups of patients, e.g. those diag- cer in humans is a rather wide field. Several publications have
nosed at a younger age, but mean age of cases in our meta- led to the hypothesis that reproductive events or hormonal
analysis was 41 years. exposures could explain gender differences in cancer suscep-
For HRT use we were able to obtain summary estimates for tibility and mortality. Increasing parity is associated with re-
‘ever use’ and duration of use, and we did not find significant duced risks of breast and ovarian cancer, in the general
associations; however, we were not able to obtain summary female population and also in high-risk women.48–50 A protec-
RRs for the time since menopause and age at start of therapy. tive effect of parity for lung cancer has been observed in sev-
Vandenbroucke45 commented on the different results obtained eral countries.51–54
from clinical trials and observational studies on HRTand breast Prior studies have also reported inverse associations be-
cancer risk and suggested that the time since menopause tween increasing parity, younger age at first birth and older
should be the key point. In fact, shorter time since menopause age at first birth and the risk of pancreatic cancer.55–57
increased the risk of HRT. The authors of a population-based Some clinical studies58,59 showed that pigmentary and
study carried out in the San Francisco Bay Area did not find nevi changes occur during pregnancy, while experimental
any trend effect on melanoma risk for time elapsed between animal studies confirmed increased pigmentation, melanocy-
menopause and first exogenous hormones use.46 tic proliferation, and tumour growth followed oestrogen
Furthermore, if there is a risk it should be found in obser- administration.25,47,60,61
vational studies, summarised in this meta-analysis, because Observational studies suggest that the pattern of age-
HRT is usually started close to menopause. Therefore, this incidence rates of melanoma in women resembles that of
should not be a major issue. breast cancer: they are higher in women than in men espe-
The previous pooled-analysis47 summarising results of cially before age 45 years, and afterwards, differently from
case-controls studies on pregnancy history showed that age men, rates of increase slow down.43 Furthermore, a higher
at first birth and parity may play a role in risk of CM in wo- risk of breast cancer among women with a history of
men: they found a significant trend (P = 0.02) for number of melanoma and excess melanoma risk among breast cancer
live births (OR = 0.94, 95%CI: 0.89–0.99) and detected a signifi- cases have been reported in several studies.18–20,62–64
cant reduced risk of melanoma among women with both ear- This study confirmed no increased risk of CM for use of
lier age at first birth (<20 years versus P25) and higher parity oral contraceptives and hormone replacement therapy: exog-
(P5 children), adjusting for socio-economic status. enous female hormones do not contribute significantly to in-
For reproductive factors in the present work, we summa- creased risk of CM. On the other hand, this meta-analysis
rised information from eight studies that published data on found significant associations of CM with parity and age at
age at pregnancy and we found that a 10-year increase in first pregnancy that requires further research.
age at first pregnancy statistically increases the risk for CM Our findings supporting the hypotheses that pregnancy re-
by about 10%. Furthermore, investigating between-study lated factors could be causally linked with CM risk require fur-
heterogeneity of risk estimates among the eighteen studies ther research.
presenting estimates for parity, this work shows that the
studies with the best study design (cohort studies and popu- Conflict of interest statement
lation-based case-control studies) significantly differ from
the others, suggesting a protective effect of parity. Similar None declared.
differences were found when examining the country where
the studies were conducted: studies carried out in Europe Appendix A. Supplementary data
suggest a significant reduction in risk of about 15% for
women with 2 children compared to nulliparous women. Supplementary data associated with this article can be found,
All these results are suggestive of an apparent protective in the online version, at doi:10.1016/j.ejca.2011.04.023.
effect of multiparity that could have an immunological
background.47 This negative association, however, may also R E F E R E N C E S
reflect confounding by socio-economic status. In fact, heter-
ogeneity analysis showed the significant effect of adjustment 1. Conybeare RC. Malignant melanoma and pregnancy: report of
for education on estimates for parity. Residual confounding 3 cases. Obstet Gynecol 1964;24:451–4.
could also be due to sun exposure due to more modest sun- 2. Byrd Jr BF, McGamity WJ. The effect of pregnancy on the
bathing habits among mothers of many children; however, clinical course of malignant melanoma. South Med J
estimates adjusted for sun exposure were virtually identical 1954;47(3):196–200.
to the unadjusted ones. 3. Osterlind A, Tucker MA, Stone BJ, Jensen OM. The Danish
case-control study of cutaneous malignant melanoma. III.
A previous pooled-analysis of case-controls47 also investi-
Hormonal and reproductive factors in women. Int J Cancer
gated associations with fertility treatments and found no ef- 1988;42(6):821–4.
fect. Three studies published estimates on the effect of
drugs for fertility and seven studies presented RRs for women

Please cite this article in press as: Gandini S et al., Hormonal and reproductive factors in relation to melanoma in women: Current review and
meta-analysis, Eur J Cancer (2011), doi:10.1016/j.ejca.2011.04.023
 EUROPEAN JOURNAL OF CANCER x x x ( 2 0 1 1 ) x x x –x x x 9

4. Walker MJ, Beattie CW, Patel MK, Ronan SM, Das Gupta TK. 27. Pfahlberg A, Hassan K, Wille L, Lausen B, Gefeller O.
Estrogen receptor in malignant melanoma. J Clin Oncol Systematic review of case-control studies: oral contraceptives
1987;5(8):1256–61. show no effect on melanoma risk. Public Health Rev 1997;25(3–
5. Chaudhuri PK, Walker MJ, Briele HA, Beattie CW, Gupta TK. ):309–15.
Incidence of estrogen receptor in benign nevi and human 28. Stroup DF, Berlin JA, Morton SC, et al. Meta-analysis of
malignant melanoma. JAMA 1980;244(8):791–3. observational studies in epidemiology: a proposal for
6. Mogren I, Stenlund H, Hogberg U. Long-term impact of reporting. Meta-analysis of observational studies in
reproductive factors on the risk of cervical, endometrial, epidemiology (MOOSE) group. JAMA 2000;283(15):2008–12.
ovarian and breast cancer. Acta Oncol 2001;40(7):849–54. 29. Haynes RB, Kastner M, Wilczynski NLHedges Team.
7. Kelsey JL, Bernstein L. Epidemiology and prevention of breast Developing optimal search strategies for detecting clinically
cancer. Annu Rev Public Health 1996;17:47–67. sound and relevant causation studies in EMBASE. BMC Med
8. Grady D, Ernster VL. Hormone replacement therapy and Inform Decis Mak 2005;5:8.
endometrial cancer: are current regimens safe? J Natl Cancer 30. Shojania KG, Bero LA. Taking advantage of the explosion of
Inst 1997;89(15):1088–9. systematic reviews: an efficient MEDLINE search strategy. Eff
9. Eliassen AH, Hankinson SE. Endogenous hormone levels and Clin Pract 2001;4(4):157–62.
risk of breast, endometrial and ovarian cancers: prospective 31. Greenland S. Quantitative methods in the review of
studies. Adv Exp Med Biol 2008;630:148–65. epidemiologic literature. Epidemiol Rev 1987;9:1–30.
10. Newcomb PA, Trentham-Dietz A, Hampton JM, et al. Late age 32. van Houwelingen HC, Arends LR, Stijnen T. Advanced
at first full term birth is strongly associated with lobular methods in meta-analysis: Multivariate approach and meta-
breast cancer. Cancer 2010. regression. Stat Med 2002;21(4):589–624.
11. Reeves GK, Pirie K, Green J, et alMillion Women Study 33. Higgins JP, Thompson SG. Quantifying heterogeneity in a
Collaborators. Reproductive factors and specific histological meta-analysis. Stat Med 2002;21(11):1539–58.
types of breast cancer: prospective study and meta-analysis. 34. Thompson SG, Higgins JP. How should meta-regression
Br J Cancer 2009;100(3):538–44. analyses be undertaken and interpreted? Stat Med
12. Lee JA, Storer BE. Excess of malignant melanomas in women 2002;21(11):1559–73.
in the British isles. Lancet 1980;2(8208–8209):1337–9. 35. Greenland S, Longnecker MP. Methods for trend estimation
13. Lee JA, Storer BE. Further studies on skin melanomas from summarized dose-response data, with applications to
apparently dependent on female sex hormones. Int J Epidemiol meta-analysis. Am J Epidemiol 1992;135(11):1301–9.
1982;11(2):127–31. 36. Macaskill P, Walter SD, Irwig L. A comparison of methods to
14. de Vries E, Houterman S, Janssen-Heijnen ML, et al. Up-to- detect publication bias in meta-analysis. Stat Med
date survival estimates and historical trends of cutaneous 2001;20(4):641–54.
malignant melanoma in the south-east of the Netherlands. 37. Koomen ER, Joosse A, Herings RM, et al. Does use of
Ann Oncol 2007;18(6):1110–6. estrogens decrease the breslow thickness of melanoma of the
15. Marrett LD, Nguyen HL, Armstrong BK. Trends in the skin? Oral contraceptives and hormonal replacement therapy.
incidence of cutaneous malignant melanoma in New South Melanoma Res 2009;19(5):327–32.
Wales, 1983–1996. Int J Cancer 2001;92(3):457–62. 38. Zaridze D, Mukeria A, Duffy S. Risk factors for skin melanoma
16. Bataille V, de Vries E. Melanoma – part 1: epidemiology, risk in Moscow (letter). Int J Cancer 1992;52:159–61.
factors, and prevention. BMJ 2008;337:a2249. 39. Lens MB, Rosdahl I, Ahlbom A, et al. Effect of pregnancy on
17. de Vries E, Nijsten TE, Visser O, et al. Superior survival of survival in women with cutaneous malignant melanoma. J
females among 10, 538 Dutch melanoma patients is Clin Oncol 2004;22(21):4369–75.
independent of Breslow thickness, histologic type and 40. O’Meara AT, Cress R, Xing G, Danielsen B, Smith LH.
tumour site. Ann Oncol 2008;19(3):583–9. Malignant melanoma in pregnancy. A population-based
18. Koh HK, Sober AJ, Carey RA. Possible association between evaluation. Cancer 2005;103(6):1217–26.
malignant melanoma and breast cancer. Arch Dermatol 41. Stensheim H, Moller B, van Dijk T, Fossa SD. Cause-specific
1987;123(6):712–3. survival for women diagnosed with cancer during pregnancy
19. Volk N, Pompe-Kirn V. Second primary cancers in breast or lactation: a registry-based cohort study. J Clin Oncol
cancer patients in Slovenia. Cancer Causes Control 2009;27(1):45–51.
1997;8(5):764–70. 43. Lens MB, Dawes M. Global perspectives of contemporary
20. Harvey EB, Brinton LA. Second cancer following cancer of the epidemiological trends of cutaneous malignant melanoma. Br
breast in Connecticut, 1935–82. Natl Cancer Inst Monogr J Dermatol 2004;150(2):179–85.
1985;68:99–112. 44. Feskanich D, Hunter DJ, Willett WC, et al. Oral contraceptive
21. Slingluff CL Jr., Reintgen DS, Vollmer RT, Seigler HF. Malignant use and risk of melanoma in premenopausal women. Br J
melanoma arising during pregnancy. A study of 100 patients. Cancer 1999;81(5):918–23.
Ann Surg 1990;211(5):552–7 [discussion 558–9]. 45. Vandenbroucke JP. The HRT controversy: observational
22. Travers RL, Sober AJ, Berwick M, et al. Increased thickness of studies and RCTs fall in line. Lancet 2009;373(9671):1233–5.
pregnancy-associated melanoma. Br J Dermatol 46. Holly EA, Cress RD, Ahn DK. Cutaneous melanoma in women:
1995;132(6):876–83. ovulatory life, menopause, and use of exogenous estrogens.
23. Cetin I, Cozzi V, Antonazzo P. Infertility as a cancer risk factor Cancer Epidemiol Biomarkers Prev 1994;3(8):661–8.
– a review. Placenta 2008;29(Suppl. B):169–77. 47. Karagas MR, Zens MS, Stukel TA, et al. Pregnancy history and
24. Sovino H, Sir-Petermann T, Devoto L. Clomiphene citrate and incidence of melanoma in women: a pooled analysis. Cancer
ovulation induction. Reprod Biomed Online 2002;4(3):303–10. Causes Control 2006;17(1):11–9.
25. Karagas MR, Stukel TA, Dykes J, et al. A pooled analysis of 10 48. Braem MG, Onland-Moret NC, van den Brandt PA, et al.
case-control studies of melanoma and oral contraceptive use. Reproductive and hormonal factors in association with
Br J Cancer 2002;86(7):1085–92. ovarian cancer in the Netherlands cohort study. Am J
26. Gefeller O, Hassan K, Wille L. Cutaneous malignant Epidemiol 2010;172(10):1181–9.
melanoma in women and the role of oral contraceptives. Br J 49. Milne RL, Osorio A, Ramon y Cajal T, et al. Parity and the risk
Dermatol 1998;138(1):122–4. of breast and ovarian cancer in BRCA1 and BRCA2 mutation

Please cite this article in press as: Gandini S et al., Hormonal and reproductive factors in relation to melanoma in women: Current review and
meta-analysis, Eur J Cancer (2011), doi:10.1016/j.ejca.2011.04.023
10 EUROPEAN JOURNAL OF CANCER x x x ( 2 0 1 1 ) x x x –x x x

carriers. Breast Cancer Res Treat 2010;119(1):221–32. Med J Aust 1985;142(8):446–8.

50. Chiaffarino F, Pelucchi C, Parazzini F, et al. Reproductive and 73. Adami HO, Persson I, Hoover R, Schairer C, Bergkvist L. Risk of
hormonal factors and ovarian cancer. Ann Oncol cancer in women receiving hormone replacement therapy. Int
2001;12(3):337–41. J Cancer 1989;44(5):833–9.
51. Kreuzer M, Gerken M, Heinrich J, Kreienbrock L, Wichmann 74. Zanetti R, Franceschi S, Rosso S, Bidoli E, Colonna S.
HE. Hormonal factors and risk of lung cancer among women? Cutaneous malignant melanoma in females: the role of
Int J Epidemiol 2003;32(2):263–71. hormonal and reproductive factors. Int J Epidemiol
52. Kubik AK, Zatloukal P, Tomasek L, Petruzelka L. Lung cancer 1990;19(3):522–6.
risk among Czech women: a case-control study. Prev Med 75. Le MG, Cabanes PA, Desvignes V, et al. Oral contraceptive use
2002;34(4):436–44. and risk of cutaneous malignant melanoma in a case-control
53. Brenner AV, Wang Z, Kleinerman RA, et al. Menstrual and study of French women. Cancer Causes Control
reproductive factors and risk of lung cancer among Chinese 1992;3(3):199–205.
women, Eastern Gansu province, 1994–1998. J Epidemiol 76. Palmer JR, Rosenberg L, Strom BL, et al. Oral contraceptive
2003;13(1):22–8. use and risk of cutaneous malignant melanoma. Cancer
54. Paulus JK, Asomaning K, Kraft P, et al. Parity and risk of lung Causes Control 1992;3(6):547–54.
cancer in women. Am J Epidemiol 2010;171(5):557–63. 77. Zaridze D, Mukeria A, Duffy S. Risk factors for skin melanoma
55. Fernandez E, La Vecchia C, D’Avanzo B, Negri E. Menstrual in Moscow. Int J Cancer 1992;52(1):159–61.
and reproductive factors and pancreatic cancer risk in 78. Persson I, Yuen J, Bergkvist L, Schairer C. Cancer incidence
women. Int J Cancer 1995;62(1):11–4. and mortality in women receiving estrogen and estrogen-
56. Karlson BM, Wuu J, Hsieh CC, Lambe M, Ekbom A. Parity and progestin replacement therapy – long-term follow-up of a
the risk of pancreatic cancer: a nested case-control study. Int J Swedish cohort. Int J Cancer 1996;67(3):327–32.
Cancer 1998;77(2):224–7. 79. Smith MA, Fine JA, Barnhill RL, Berwick M. Hormonal and
57. Kreiger N, Lacroix J, Sloan M. Hormonal factors and reproductive influences and risk of melanoma in women. Int J
pancreatic cancer in women. Ann Epidemiol 2001;11(8):563–7. Epidemiol 1998;27(5):751–7.
58. Sanchez NP, Pathak MA, Sato S, et al. Melasma: a clinical, 80. Westerdahl J, Olsson H, Masback A, Ingvar C, Jonsson N. Risk
light microscopic, ultrastructural, and immunofluorescence of malignant melanoma in relation to drug intake, alcohol,
study. J Am Acad Dermatol 1981;4(6):698–710. smoking and hormonal factors. Br J Cancer 1996;73(9):1126–31.
59. Foucar E, Bentley TJ, Laube DW, Rosai J. A histopathologic 81. Young P, Purdie D, Jackman L, Molloy D, Green A. A study of
evaluation of nevocellular nevi in pregnancy. Arch Dermatol infertility treatment and melanoma. Melanoma Res
1985;121(3):350–4. 2001;11(5):535–41.
60. Lopez RE, Bhakoo H, Paolini NS, et al. Effect of estrogen on 82. Freedman DM, Sigurdson A, Doody MM, Rao RS, Linet MS. Risk
the growth of B-16 melanoma. Surg Forum 1978;29:153–4. of melanoma in relation to smoking, alcohol intake, and other
61. Beattie CW, Ronan SG, Amoss Jr MS. Estrogens influence the factors in a large occupational cohort. Cancer Causes Control
natural history of Sinclair swine cutaneous melanoma. Cancer 2003;14(9):847–57.
Res 1991;51(8):2025–8. 83. Naldi L, Altieri A, Imberti GL, et al. Cutaneous malignant
62. Bradford PT, Freedman DM, Goldstein AM, Tucker MA. melanoma in women. Phenotypic characteristics, sun
Increased risk of second primary cancers after a diagnosis of exposure, and hormonal factors: a case-control study from
melanoma. Arch Dermatol 2010;146(3):265–72. Italy. Ann Epidemiol 2005;15(7):545–50.
63. Mellemkjaer L, Friis S, Olsen JH, et al. Risk of second cancer 84. Vessey M, Painter R. Oral contraceptive use and cancer.
among women with breast cancer. Int J Cancer Findings in a large cohort study, 1968–2004. Br J Cancer
2006;118(9):2285–92. 2006;95(3):385–9.
64. Stracci F, D’Alo D, Cassetti T, Scheibel M, La Rosa F. Incidence 85. Hannaford PC, Selvaraj S, Elliott AM, et al. Cancer risk among
of multiple primary malignancies in women diagnosed with users of oral contraceptives: cohort data from the royal
breast cancer. Eur J Gynaecol Oncol 2009;30(6):661–3. college of general practitioner’s oral contraception study. BMJ
65. MacKie RM, Hole DJ. Incidence and thickness of primary 2007;335(7621):651.
tumours and survival of patients with cutaneous malignant 86. Lea CS, Holly EA, Hartge P, Lee JS, et al. Reproductive risk
melanoma in relation to socioeconomic status. BMJ factors for cutaneous melanoma in women: a case-control
1996;312(7039):1125–8. study. Am J Epidemiol 2007;165(5):505–13.
66. Beral V, Ramcharan S, Faris R. Malignant melanoma and oral 87. Gallagher RP, Elwood JM, Hill GB, et al. Reproductive factors,
contraceptive use among women in California. Br J Cancer oral contraceptives and risk of malignant melanoma: Western
1977;36(6):804–9. Canada melanoma study. Br J Cancer 1985;52(6):901–7.
67. Adam SA, Sheaves JK, Wright NH, et al. A case-control study 88. Brinton LA, Melton 3rd LJ, Malkasian Jr GD, Bond A, Hoover R.
of the possible association between oral contraceptives and Cancer risk after evaluation for infertility. Am J Epidemiol
malignant melanoma. Br J Cancer 1981;44(1):45–50. 1989;129(4):712–22.
68. Bain C, Hennekens CH, Speizer FE, et al. Oral contraceptive 89. Wyshak G, Frisch RE, Albright NL, Albright TE, Schiff I.
use and malignant melanoma. J Natl Cancer Inst Reproductive factors and melanoma of the skin among
1982;68(4):537–9. women. Int J Dermatol 1989;28(8):527–30.
69. Holly EA, Weiss NS, Liff JM. Cutaneous melanoma in relation 90. Holly EA, Aston DA, Cress RD. Cutaneous melanoma in
to exogenous hormones and reproductive factors. J Natl women II. Phenotypic characteristics and other host-related
Cancer Inst 1983;70(5):827–31. factors. Am J Epidemiol 1995;141:934–42.
70. Beral V, Evans S, Shaw H, Milton G. Oral contraceptive use and 91. Rossing MA, Daling JR, Weiss NS, Moore DE, Self SG. Risk of
malignant melanoma in Australia. Br J Cancer cutaneous melanoma in a cohort of infertile women.
1984;50(5):681–5. Melanoma Res 1995;5(2):123–7.
71. Holman CD, Armstrong BK, Heenan PJ. Cutaneous malignant 92. Lambe M, Thorn M, Sparen P, Bergstrom R, Adami HO.
melanoma in women: exogenous sex hormones and Malignant melanoma: reduced risk associated with early
reproductive factors. Br J Cancer 1984;50(5):673–80. childbearing and multiparity. Melanoma Res 1996;6(2):
72. Green A, Bain C. Hormonal factors and melanoma in women. 147–53.

Please cite this article in press as: Gandini S et al., Hormonal and reproductive factors in relation to melanoma in women: Current review and
meta-analysis, Eur J Cancer (2011), doi:10.1016/j.ejca.2011.04.023
 EUROPEAN JOURNAL OF CANCER x x x ( 2 0 1 1 ) x x x –x x x 11

93. Modan B, Ron E, Lerner-Geva L, Blumstein T, et al. Cancer 96. Kaae J, Andersen A, Boyd HA, Wohlfahrt J, Melbye M.
incidence in a cohort of infertile women. Am J Epidemiol Reproductive history and cutaneous malignant melanoma: a
1998;147(11):1038–42. comparison between women and men. Am J Epidemiol
94. Neale RE, Darlington S, Murphy MF, et al. The effects of 2007;165(11):1265–70.
twins, parity and age at first birth on cancer risk in 97. Hannibal CG, Jensen A, Sharif H, Kjaer SK. Malignant
Swedish women. Twin Res Hum Genet 2005;8(2): melanoma risk after exposure to fertility drugs: results from a
156–62. large Danish cohort study. Cancer Causes Control
95. Althuis MD, Scoccia B, Lamb EJ et al. Melanoma, thyroid, 2008;19(7):759–65.
cervical, and colon cancer risk after use of fertility drugs. Am J
Obstet Gynecol 2005;193(3 Pt 1):668–74.

Please cite this article in press as: Gandini S et al., Hormonal and reproductive factors in relation to melanoma in women: Current review and
meta-analysis, Eur J Cancer (2011), doi:10.1016/j.ejca.2011.04.023