new england

The
journal of medicine
established in 1812 june 5, 2014 vol. 370 no. 23

Once-Weekly Dalbavancin versus Daily Conventional Therapy

for Skin Infection
Helen W. Boucher, M.D., Mark Wilcox, M.D., George H. Talbot, M.D., Sailaja Puttagunta, M.D.,
Anita F. Das, Ph.D., and Michael W. Dunne, M.D.

A BS T R AC T

BACKGROUND
Dalbavancin, a lipoglycopeptide antibiotic agent that is active against gram-positive From the Division of Infectious Diseases
pathogens, has a long plasma half-life, allowing for once-weekly dosing. DISCOVER 1 and Geographic Medicine, Tufts Medical
Center and Tufts University School of
and DISCOVER 2 were identically designed noninferiority trials of dalbavancin for Medicine, Boston (H.W.B.); the Depart-
the treatment of acute bacterial skin and skin-structure infection. ment of Microbiology, Leeds Teaching
Hospital and University of Leeds, Old
Medical School, Leeds, United Kingdom
METHODS (M.W.); Talbot Advisors, Anna Maria, FL
We randomly assigned patients to receive dalbavancin intravenously on days 1 and 8 (G.H.T.); Durata Therapeutics, Branford,
or vancomycin intravenously for at least 3 days with the option to switch to oral CT (S.P., M.W.D.); and InClin, San Mateo,
CA (A.F.D.). Address reprint requests to
linezolid to complete 10 to 14 days of therapy. The primary end point, early clinical Dr. Boucher at Tufts Medical Center, 800
response, required the cessation of spread of infection-related erythema and the Washington St., Box 238, Boston, MA 02111,
absence of fever at 48 to 72 hours. Secondary end points at the end of therapy in- or at hboucher@tuftsmedicalcenter.org.
cluded clinical status and investigator’s assessment of outcome. N Engl J Med 2014;370:2169-79.
DOI: 10.1056/NEJMoa1310480
Copyright © 2014 Massachusetts Medical Society.
RESULTS
Analysis of the primary end point showed noninferiority of dalbavancin in both
DISCOVER 1 and DISCOVER 2. In the pooled analysis, 525 of 659 patients (79.7%)
in the dalbavancin group and 521 of 653 (79.8%) in the vancomycin–linezolid group
had an early clinical response indicating treatment success (weighted difference,
−0.1 percentage point; 95% confidence interval, −4.5 to 4.2). The outcomes were
similar in the analyses by study and the pooled analyses of clinical status at the end
of therapy and the investigator’s assessment of outcome. For patients infected with
Staphylococcus aureus, including methicillin-resistant S. aureus, clinical success was seen
in 90.6% of the patients treated with dalbavancin and 93.8% of those treated with
vancomycin–linezolid. Adverse events and study days with an adverse event were
less frequent in the dalbavancin group than in the vancomycin–linezolid group. The
most common treatment-related adverse events in either group were nausea, diar-
rhea, and pruritus.

CONCLUSIONS
Once-weekly intravenous dalbavancin was not inferior to twice-daily intravenous
vancomycin followed by oral linezolid for the treatment of acute bacterial skin
and skin-structure infection. (Funded by Durata Therapeutics; DISCOVER 1 and
DISCOVER 2 ClinicalTrials.gov numbers, NCT01339091 and NCT01431339.)

n engl j med 370;23 nejm.org june 5, 2014 2169

The New England Journal of Medicine
Downloaded from nejm.org on June 24, 2014. For personal use only. No other uses without permission.
Copyright © 2014 Massachusetts Medical Society. All rights reserved.
 The n e w e ng l a n d j o u r na l of m e dic i n e

A
cute bacterial skin and skin- On the basis of the treatment effect of 19 to 27%
structure infections are among the most that was observed in these studies for the cessation
common reasons for the hospitalization of lesion spread and for fever resolution, a non­
of adults in the United States today.1 These infec- inferiority margin of 10 percentage points for the
tions are caused most often by Staphylococcus au­ primary efficacy outcome was deemed appropriate.
reus and streptococci.2 Methicillin-resistant S. aureus The trials discussed here were designed to test
(MRSA) accounts for many of these infections the efficacy of dalbavancin with the use of this
and presents a particular treatment challenge be- new primary end point and to assess its associa-
cause current therapies are limited by toxicity, tion with the historical standard of investigator-
resistance, or the lack of an oral formulation.3 assessed response to treatment. Both the primary
Associated medical costs are substantial.4 and secondary end points conform to the FDA
Dalbavancin (Durata Therapeutics) is a lipo- guidance and were agreed on with the FDA as
glycopeptide antibiotic agent with in vitro and in part of a Special Protocol Assessment.
vivo activity against gram-positive pathogens, in-
cluding a minimal inhibitory concentration (MIC) ME THODS
required to inhibit the growth of 90% of the iso-
lates (MIC90 ) for S. aureus of 0.06 μg per milliliter. TRIAL DESIGN
Dalbavancin has a terminal half-life of 2 weeks, DISCOVER 1 and DISCOVER 2 were double-
related in part to 93% protein binding, and is more blind, double-dummy, international, multicenter,
efficacious in animal models when adminis- randomized trials conducted from 2011 through
tered as larger, less frequent doses, as compared 2012 at 54 and 86 investigative sites, respectively.
with smaller, more frequent doses. Phase 1 trials The institutional review board or ethics committee
of once-weekly dosing showed mean total and at each study site approved the protocols, avail-
calculated free plasma concentrations that were able with the full text of this article at NEJM.org.
above the S. aureus bactericidal concentration for 7 All the patients provided written informed con-
to 14 days.5-7 Efficacy was shown in a “registra- sent before participation. The identical design of
tional” development program (i.e., one designed the two trials allowed pooling of the data to en-
and conducted to support regulatory review and hance information regarding adverse events and
approval for marketing) that included four trials efficacy variables of interest.
involving 1086 patients with skin infection.5,8,9 The diagnosis of acute bacterial skin and
The role of antibiotic treatment, in addition to skin-structure infection required the presence of
incision and drainage, in the treatment of small cellulitis, a major abscess, or a wound infection,
cutaneous abscesses is an unresolved question each associated with at least 75 cm2 of erythema.
that is under active investigation. Assessment of Eligible patients were adults who were thought
the antibiotic effect in the treatment of large ab- to require at least 3 days of intravenous therapy
scesses and cellulitis has been challenging, given who had one or more systemic signs of infection
the need for placebo-controlled data on which to within 24 hours before randomization, including
base a noninferiority margin for registrational an elevated body temperature (>38°C), a white-
clinical trials. Recent guidance from the Food cell count of more than 12,000 cells per cubic
and Drug Administration (FDA) requires that the millimeter, or more than 10% band forms on the
area of the skin infection be larger than histori- white-cell differential count. In addition to ery-
cal standards and that abscesses be of substan- thema, at least two of the following local signs
tial size and complexity.10 Central to this updated were required: purulent drainage or discharge,
guidance was a focus on identifying a quantifi- fluctuance, heat or localized warmth, tenderness
able, reproducible efficacy end point that showed on palpation, and swelling or induration. Patients
sensitivity to drug effect, thereby providing jus- who had received antibiotic treatment within 14
tification for a noninferiority margin.10 Historical days before randomization were excluded.
studies by Snodgrass and Anderson comparing The studies were designed by all the authors
antibiotic treatment with the standard of care and conducted according to the respective study
were cited by the FDA as a demonstration of a protocols by the sponsor (Durata Therapeutics) in
treatment effect over placebo with the use of both collaboration with the investigators. The sponsor
absence of fever and stabilization in the size of collected the data, monitored the study conduct,
the infected area during therapy as end points.11,12 and performed the statistical analyses. All the

2170 n engl j med 370;23 nejm.org june 5, 2014

The New England Journal of Medicine

Downloaded from nejm.org on June 24, 2014. For personal use only. No other uses without permission.
Copyright © 2014 Massachusetts Medical Society. All rights reserved.
 Dalbavancin vs. Conventional Ther apy for Skin Infection

authors had access to the data, assume respon- early clinical response indicating treatment suc-
sibility for the integrity and completeness of the cess) was defined as both cessation of spread of
reported data and analyses, and vouch for the the erythema associated with the infection (i.e.,
fidelity of this report to the study protocols. no increase in the surface area as compared with
The first draft of the manuscript was prepared baseline) and a temperature of 37.6°C or lower at
by the first author, with input from all the au- three consecutive readings performed 6 hours
thors. All the authors participated in revisions apart. Treatment success or failure was deter-
to the manuscript. mined on the basis of the above criteria after the
treatment was completed and therefore did not
RANDOMIZATION, TREATMENT, AND MONITORING influence treatment decisions by the investigator
Patients were assigned by means of a telephone during the trial. Patients with missing data for
randomization system to a treatment group in a surface area of infection or temperature (i.e.,
1:1 ratio, with a block size of four. Randomization those for whom treatment success could not be
was stratified according to infection type and determined) were considered to have treatment
presence or absence of fever such that no more failure in the primary intention-to-treat analysis.
than 30% of the enrolled patients had a major Secondary end points (including clinical status at
abscess and at least 25% had fever. the end of therapy, determined programmatical-
Patients received either dalbavancin at a dose ly, and clinical response at the end of therapy, as
of 1 g given intravenously over a period of 30 min- assessed by the investigator), criteria for treat-
utes on day 1, followed by 500 mg given intra- ment failure, and planned sensitivity analyses are
venously over a period of 30 minutes on day 8, described in the Methods section in the
or vancomycin at a dose of 1 g (or 15 mg per Supplementary Appendix, available at NEJM.org.
kilogram of body weight) given intravenously
over a period of 120 minutes every 12 hours for Safety
at least 3 days, with an option to switch to oral Adverse events and serious adverse events were
linezolid, at a dose of 600 mg every 12 hours, to recorded throughout the study period. Adverse
complete 10 to 14 days of therapy. The decision events emerging during treatment were those
to use vancomycin at a fixed dose rather than at with an onset or worsening severity at or after
a weight-based dose was made by clinicians at administration of the first dose of the study drug
the individual study sites on the basis of their through the long-term follow-up visit (day 70).
local standard of care. The comparator regimen Additional features of the study design are de-
was selected because of its efficacy and routine scribed in the Supplementary Appendix.
use in clinical practice.13
The vancomycin dose could be adjusted, accord- STATISTICAL ANALYSIS
ing to the local standard of care, by a pharmacist
The sample size for each study was calculated
who was aware of the study-drug assignment. with the use of the Farrington–Manning method.
The protocol recommended adjusting the dose Assuming a point estimate for early clinical re-
of either vancomycin or dalbavancin according sponse of 85% in each treatment group, a one-sided
to the ideal body weight for patients with renal alpha level of 0.025, and a noninferiority margin
insufficiency. Patients in the dalbavancin group of 10 percentage points, we calculated that a total
received a placebo infusion every 12 hours, plus sample of 556 patients would be required in order
an oral placebo if there was a switch to oral to provide the study with 90% power.
therapy. Study-drug treatment was continued for The primary analysis of the early end point was
10 to 14 days in each group. Patients could re- based on the intention-to-treat population, which
ceive inpatient treatment, outpatient treatment, consisted of all the patients who underwent ran-
or both, according to the investigator’s judg- domization. All the patients in the intention-to-
ment and appropriate logistic arrangements at treat population who received at least one dose
the study site. of a study drug constituted the safety population.
The patients in the clinical per-protocol population
ASSESSMENTS met all the inclusion criteria and none of the ex-
End Points clusion criteria, received the correct study drug,
The primary end point was measured at 48 to and met minimum dosing requirements. The pa-
72 hours of therapy. A successful outcome (i.e., tients in the microbiologic per-protocol popula-

n engl j med 370;23 nejm.org june 5, 2014 2171

The New England Journal of Medicine
Downloaded from nejm.org on June 24, 2014. For personal use only. No other uses without permission.
Copyright © 2014 Massachusetts Medical Society. All rights reserved.
 The n e w e ng l a n d j o u r na l of m e dic i n e

tion were the subset of patients in the clinical More than 85% of the patients had a baseline
per-protocol population who had at least one temperature of more than 38°C; the median size
gram-positive pathogen isolated at baseline. of the infected area was 351 cm2 in DISCOVER 1
For each study, the 95% confidence interval and 336 cm2 in DISCOVER 2, well above the
for the difference in success rates (dalbavancin protocol-mandated minimum size of 75 cm2. The
minus vancomycin–linezolid), with adjustment for criteria for the systemic inflammatory response
the stratification factor at randomization of the syndrome were met in 62% of the patients in
presence or absence of fever at baseline, was com- DISCOVER 1 and 43% of those in DISCOVER 2.
puted with the use of the method of Miettinen A total of 608 of 659 patients (92.3%) in the
and Nurminen.14 For the pooled analysis, the dalbavancin group and 601 of 653 (92.0%) in the
weighted difference in success rates was calcu- vancomycin–linezolid group completed the study
lated, and the 95% confidence interval, adjusted treatment. Patients receiving dalbavancin were
for study, was computed with the use of the more likely than those receiving vancomycin–­
same method. The analysis of clinical status at linezolid to have a shorter duration of blinded
the end of therapy in the clinical per-protocol treatment. Fewer patients in the dalbavancin group
population included an adjustment in the confi- received 14 days of therapy, as compared with
dence interval for both the presence or absence those in the vancomycin–linezolid group (31.0%
of fever and infection type. A determination of vs. 38.4%, P = 0.008), with more patients in the
noninferiority for the primary end point required dal­ba­vancin group receiving 10 days of therapy
that the lower limit of the 95% confidence inter- (20.8% vs. 15.3%, P = 0.01). Approximately 25%
val be greater than −10 percentage points; the of the patients received all their treatment as an
95% confidence intervals for the other assess- outpatient.
ments are provided for descriptive purposes. No
adjustments were made for multiple comparisons. OUTCOME
For adverse events, post hoc exploratory anal- In DISCOVER 1, an early clinical response indi-
yses of differences between the treatment groups cating treatment success was documented in
were conducted with the use of the Cochran– 240 of 288 patients (83.3%) in the dalbavancin
Mantel–Haenszel test, with adjustment for study, group and 233 of 285 (81.8%) in the vancomycin–­
or with the use of a Poisson regression for the linezolid group (difference, 1.5 percentage points;
total number of adverse events. Two-sided P val- 95% confidence interval [CI], −4.6 to 7.9) (Table 2).
ues are reported as descriptive statistics. Addi- In DISCOVER 2, an early clinical response indicat-
tional statistical methods are described in the ing treatment success occurred in 285 of 371 pa-
Supplementary Appendix. tients (76.8%) in the dalbavancin group and 288
of 368 (78.3%) in the vancomycin–linezolid group
R E SULT S (difference, −1.5 percentage points; 95% CI, −7.4
to 4.6). Since the lower limit of each 95% confi-
PATIENTS dence interval was greater than −10 percentage
Of the 1312 patients who underwent randomiza- points, dalbavancin was determined to be non­
tion, 9 did not receive the study drug, so 1303 inferior to vancomycin–linezolid in each trial.
patients were included in the safety population: In the pooled analysis, 525 of 659 patients
652 patients in the dalbavancin group and 651 (79.7%) in the dalbavancin group and 521 of 653
in the vancomycin–linezolid group (Fig. 1). The (79.8%) in the vancomycin–linezolid group had a
clinical per-protocol population included 570 pa- successful outcome at 48 to 72 hours (weighted
tients in the dalbavancin group and 545 in the difference, −0.1 percentage point; 95% CI, −4.5
vancomycin–linezolid group. to 4.2). The reasons for treatment failure were
The treatment groups were well balanced similar with each regimen, and missing tempera-
according to age, sex, and race (Table 1, and ture data (e.g., the temperature was not recorded
Table S1 in the Supplementary Appendix). Ap­ or was taken at a time outside the protocol-
prox­imately 15% of the patients had a history of specified time window) accounted for the largest
recent or current intravenous drug use, and 13% proportion of failures (Table S2 in the Supple­
had diabetes mellitus. Major abscess was slightly men­t ary Appendix).
more frequent in DISCOVER 1, as was cellulitis In a prespecified sensitivity analysis, dalbavan­
in DISCOVER 2. cin and vancomycin–linezolid had similar success

2172 n engl j med 370;23 nejm.org june 5, 2014

The New England Journal of Medicine

Downloaded from nejm.org on June 24, 2014. For personal use only. No other uses without permission.
Copyright © 2014 Massachusetts Medical Society. All rights reserved.
 Dalbavancin vs. Conventional Ther apy for Skin Infection

A DISCOVER 1

659 Patients were assessed for eligibility

573 Underwent randomization

288 Were assigned to dalbavancin 285 Were assigned to vancomycin–

treatment group linezolid treatment group

284 Received study drug 284 Received study drug

4 Did not receive study drug 1 Did not receive study drug

27 Prematurely discontinued study 28 Prematurely discontinued study

15 Were lost to follow-up 14 Were lost to follow-up
6 Had other reasons 7 Had other reasons
6 Withdrew consent 2 Withdrew consent
0 Died 5 Died

261 Completed the study 257 Completed the study

B DISCOVER 2

836 Patients were assessed for eligibility

739 Underwent randomization

371 Were assigned to dalbavancin 368 Were assigned to vancomycin–

treatment group linezolid treatment group

368 Received study drug 367 Received study drug

3 Did not receive study drug 1 Did not receive study drug

39 Prematurely discontinued study 35 Prematurely discontinued study

23 Were lost to follow-up 15 Were lost to follow-up
9 Withdrew consent 13 Withdrew consent
6 Had other reasons 4 Had other reasons
1 Died 3 Died

332 Completed the study 333 Completed the study

Figure 1. Screening, Randomization, and Completion of Treatment in the Trials.

rates with respect to a reduction in the size of the [88.1%], respectively). A number of additional
infected area of 20% or more at 48 to 72 hours prespecified sensitivity analyses examined the
(584 of 659 patients [88.6%] and 575 of 653 effect of missing temperature data and had

n engl j med 370;23 nejm.org june 5, 2014 2173

The New England Journal of Medicine
Downloaded from nejm.org on June 24, 2014. For personal use only. No other uses without permission.
Copyright © 2014 Massachusetts Medical Society. All rights reserved.
 Table 1. Characteristics of the Patients in the Intention-to-Treat Population.*

2174
Characteristic DISCOVER 1 DISCOVER 2 Both Trials
Vancomycin– Vancomycin– Vancomycin–
Dalbavancin Linezolid Dalbavancin Linezolid Dalbavancin Linezolid
(N = 288) (N = 285) (N = 371) (N = 368) (N = 659) (N = 653)
Age — yr
Mean 48.8 48.9 49.1 51.4 48.9 50.3
Range 18–84 18–84 18–85 18–84 18–85 18–84
Male sex — no. (%) 170 (59.0) 173 (60.7) 223 (60.1) 201 (54.6) 393 (59.6) 374 (57.3)
Race or ethnic group — no. (%)†
White 264 (91.7) 259 (90.9) 328 (88.4) 320 (87.0) 592 (89.8) 579 (88.7)
Black 16 (5.6) 19 (6.7) 13 (3.5) 17 (4.6) 29 (4.4) 36 (5.5)
Other 8 (2.8) 7 (2.5) 30 (8.1) 31 (8.4) 38 (5.8) 38 (5.8)
The

Region of enrollment — no. (%)

United States or Canada 123 (42.7) 121 (42.5) 115 (31.0) 114 (31.0) 238 (36.1) 235 (36.0)
Europe, South Africa, or Asia 165 (57.3) 164 (57.5) 256 (69.0) 254 (69.0) 421 (63.9) 418 (64.0)
Diabetes mellitus — no. (%) 43 (14.9) 30 (10.5) 35 (9.4) 62 (16.8) 78 (11.8) 92 (14.1)
Intravenous drug use — no. (%) 36 (12.5) 51 (17.9) 58 (15.6) 56 (15.2) 94 (14.3) 107 (16.4)
Infection type — no. (%)

n engl j med 370;23

Major abscess 72 (25.0) 86 (30.2) 90 (24.3) 87 (23.6) 162 (24.6) 173 (26.5)
Cellulitis 156 (54.2) 147 (51.6) 198 (53.4) 202 (54.9) 354 (53.7) 349 (53.4)
Wound or surgical-site infection 60 (20.8) 52 (18.2) 82 (22.1) 79 (21.5) 142 (21.5) 131 (20.1)

nejm.org
n e w e ng l a n d j o u r na l

Temperature ≥38°C — no./total no. (%) 243/284 (85.6) 242/284 (85.2) 306/365 (83.8) 310/365 (84.9) 549/649 (84.6) 552/649 (85.0)

The New England Journal of Medicine

of

White-cell count >12,000 per mm3 — 98/259 (37.8) 104/254 (40.9) 149/368 (40.5) 146/367 (39.8) 247/627(39.4) 250/621 (40.3)
no./total no. (%)

june 5, 2014
White-cell bands ≥10% — no./total no. (%) 63/238 (26.5) 66/244 (27.0) 48/241 (19.9) 42/234 (17.9) 111/479 (23.2) 108/478 (22.6)
SIRS — no./total no. (%)‡ 175/284 (61.6) 175/284 (61.6) 157/368 (42.7) 161/368 (43.8) 332/652 (50.9) 336/652 (51.5)

Copyright © 2014 Massachusetts Medical Society. All rights reserved.

m e dic i n e

Infection area — cm2§

Median 333 368 314 362 324 367
Range 26–3400 78–3675 85–5100 72–3922 26–5100 72–3922

Downloaded from nejm.org on June 24, 2014. For personal use only. No other uses without permission.
* There were no significant differences in baseline characteristics between the treatment groups except for the proportion of patients with diabetes mellitus in DISCOVER 2 (P = 0.003).
† Race or ethnic group was self-reported. Other included Asian, American Indian or Alaskan Native, Native Hawaiian or Pacific Islander, and other.
‡ The systemic inflammatory response syndrome (SIRS) was defined as the presence of two or more of the following: temperature below 36°C or above 38°C, heart rate of more than
90 beats per minute, respiratory rate of more than 20 breaths per minute, and white-cell count below 4000 per cubic millimeter or above 12,000 per cubic millimeter or with more than
10% band forms.
§ The area of erythema was defined as the longest length times the widest width that was perpendicular to length.
 Dalbavancin vs. Conventional Ther apy for Skin Infection

Table 2. Primary and Secondary Efficacy End Points.*

Vancomycin– Absolute Difference

End Point Dalbavancin Linezolid (95% CI)
number/total number (percent) percentage points
Primary end point
DISCOVER 1 240/288 (83.3) 233/285 (81.8) 1.5 (−4.6 to 7.9)
DISCOVER 2 285/371 (76.8) 288/368 (78.3) −1.5 (−7.4 to 4.6)
Both trials 525/659 (79.7) 521/653 (79.8) −0.1 (−4.5 to 4.2)
Sensitivity analysis
DISCOVER 1 259/288 (89.9) 259/285 (90.9) −1.0 (−5.7 to 4.0)
DISCOVER 2 325/371 (87.6) 316/368 (85.9) 1.7 (−3.2 to 6.7)
Both trials 584/659 (88.6) 575/653 (88.1) 0.6 (−2.9 to 4.1)
Secondary end point
Clinical status 517/570 (90.7) 502/545 (92.1) −1.5 (−4.8 to 1.9)
Sensitivity analysis of clinical status† 533/570 (93.5) 517/545 (94.9) −1.4 (−4.2 to 1.4)
Investigator’s assessment of outcome 547/570 (96.0) 527/545 (96.7) −0.7 (−3.0 to 1.5)

* The primary end point was the success rate at 48 to 72 hours after the initiation of therapy (i.e., early clinical response)
in the intention-to-treat population. The sensitivity analysis of the primary end point was the success rate, defined as a
reduction in the infection area of at least 20% at 48 to 72 hours after the initiation of therapy, in the intention-to-treat
population. The secondary end points were evaluated in a pooled analysis and included success rates at the end of therapy
in the clinical per-protocol population. For the pooled analysis, the weighted difference in success rates was calculated.
† The degree of fluctuance or localized heat or warmth had to be improved from baseline.

similar results. Among patients with bacteremia type, underlying illness, and severity of infection
at baseline who had a follow-up blood culture, (Table 3, and Table S4 in the Supplementary Ap­
23 of 23 patients (100%) in the dalbavancin group pen­dix); outcomes in the individual studies were
and 12 of 14 (85.7%) in the vancomycin–linezolid similar to those in the pooled analysis. In a pooled
group had negative blood cultures at the end of analysis of clinical status at day 14 in the clinical
therapy. per-protocol population, success was seen in
Clinical status indicating treatment success at 90.7% of patients in the dalbavancin group and
the end of treatment in the clinical per-protocol 92.1% of those in the vancomycin–linezolid
population was documented in a similar propor- group. Resolution of pain occurred at similar
tion of patients in each group (Table S3 in the rates in each treatment group (Table S5 in the
Supplementary Appendix). The most common Supplementary Appendix).
reason for failure was a lack of complete resolu- The MIC90 of dalbavancin was 0.06 μg per
tion of skin warmth at the infection site, which milliliter for the 511 S. aureus isolates and 0.06 μg
was seen more frequently in the dalbavancin per milliliter for the 77 Streptococcus pyogenes iso-
group than in the vancomycin–linezolid group lates in the two studies combined; all S. aureus
in DISCOVER 1 and at similar rates in the two isolates had a vancomycin MIC of 1 μg per mil-
treatment groups in DISCOVER 2. Analyses in liliter or less. In the pooled analysis of mono­
the intention-to-treat population had similar re- microbial S. aureus infections in the microbiologic
sults. The rate of clinical response indicating per-protocol population, 90.6% of the patients
treatment success according to the investigator’s treated with dalbavancin and 93.8% of those
assessment at the end of therapy was numeri- treated with vancomycin–­linezolid had a success-
cally higher than the programmatically deter- ful clinical outcome. Dal­ bavancin therapy was
mined clinical-status outcomes, with similar associated with a successful clinical outcome in
rates of success in each treatment group. 89.2% of patients with MRSA infection and in
Treatment outcomes in the two groups were 91.5% of those with methicillin-susceptible S. au­
similar when analyzed according to infection reus infection (Table S4 in the Supplementary Ap-

n engl j med 370;23 nejm.org june 5, 2014 2175

The New England Journal of Medicine
Downloaded from nejm.org on June 24, 2014. For personal use only. No other uses without permission.
Copyright © 2014 Massachusetts Medical Society. All rights reserved.
 The n e w e ng l a n d j o u r na l of m e dic i n e

Table 3. Additional Secondary Analyses of Treatment Success.*

Dalbavancin Vancomycin–Linezolid
Variable (N = 652) (N = 651)
number/total number (percent)
Clinical response according to infection type
Cellulitis
At 48–72 hr 281/354 (79.4) 269/349 (77.1)
At end of therapy 294/324 (90.7) 276/301 (91.7)
Major abscess
At 48–72 hr 133/163 (81.6) 149/173 (86.1)
At end of therapy 125/133 (94.0) 133/139 (95.7)
Traumatic wound or surgical-site infection
At 48–72 hr 111/142 (78.2) 103/131 (78.6)
At end of therapy 98/113 (86.7) 93/105 (88.6)
Investigator-assessed clinical response at end of therapy
according to baseline pathogen†
Staphylococcus aureus 187/191 (97.9) 171/177 (96.6)
Methicillin-resistant S. aureus   72/74 (97.3)   49/50 (98.0)
Streptococcus pyogenes    19/19 (100.0)   12/13 (92.3)
Clinical response at end of therapy according to diabetes
mellitus status at baseline
Diabetes mellitus   60/71 (84.5)   67/76 (88.2)
No diabetes mellitus 457/499 (91.6) 435/469 (92.7)
Clinical response at end of therapy according to SIRS status
at baseline
SIRS 257/296 (86.8) 263/290 (90.7)
No SIRS 260/274 (94.9) 239/255 (93.7)

* The success rates at 48 to 72 hours were assessed in the intention-to-treat population, and the success rates at the end
of therapy were assessed in the clinical per-protocol population of patients.
† The success rates at the end of therapy were assessed in the subgroup of patients with monomicrobial infection in the
microbiologic per-protocol population.

pendix). Successful outcomes according to the who was unaware of the treatment assignment.
investigator’s assessment for all these pathogens An adverse event led to the discontinuation of
were numerically higher than the programmati- the study treatment in 2.1% of the patients in
cally determined outcomes, but the rates were the dalbavan­cin group and 2.0% of those in the
similar in the two treatment groups. vancomycin–linezolid group.
The most common treatment-related adverse
SAFETY events in the dalbavancin and vancomycin–linezolid
Adverse events were reported in fewer patients groups were nausea (in 2.5% and 2.9% of pa-
treated with dalbavancin than in those treated with tients, respectively), diarrhea (in 0.8% and 2.5%;
vancomycin–linezolid (Table 4). The total num- P = 0.02), and pruritus (in 0.6% and 2.3%; P = 0.01);
ber of adverse events reported per patient in the these events, plus headache, were also the most
dalbavancin group was also lower, resulting in common adverse events due to any cause (treat-
fewer days with an adverse event in the dalba- ment-related or not) (Table S6 in the Supple­men­
vancin group than in the vancomycin–linezolid tary Appendix). An infusion site–related reaction
group. Most adverse events were thought to be was seen in 9 patients (1.4%) in the dalbavancin
unrelated to the study treatment and were mild; group and 11 (1.7%) in the vancomycin–linezolid
adverse events were assessed by the investigator, group; flushing was seen in 1 patient (0.2%) and

2176 n engl j med 370;23 nejm.org june 5, 2014

The New England Journal of Medicine

Downloaded from nejm.org on June 24, 2014. For personal use only. No other uses without permission.
Copyright © 2014 Massachusetts Medical Society. All rights reserved.
 Dalbavancin vs. Conventional Ther apy for Skin Infection

Table 4. Adverse Events.

Vancomycin–
Dalbavancin Linezolid
Variable (N = 652) (N = 651) P Value*
Any adverse event
Any event — no. of patients (%) 214 (32.8) 247 (37.9) 0.05
Total no. of events 540 645 0.05
Treatment-related adverse event†
Any event — no. of patients (%) 80 (12.3) 89 (13.7) 0.45
Total no. of events 139 183 0.02
Serious adverse event — no. of patients (%)
Any event 17 (2.6) 26 (4.0) 0.16
Treatment-related event† 2 (0.3) 4 (0.6) 0.41
Death — no. (%)‡ 1 (0.2) 7 (1.1) 0.03
Treatment-limiting adverse event — no. of patients (%)§ 14 (2.1) 13 (2.0) 0.85
Most common treatment-related adverse event — no. of
patients (%)¶
Nausea 16 (2.5) 19 (2.9) 0.62
Diarrhea 5 (0.8) 16 (2.5) 0.02
Pruritus 4 (0.6) 15 (2.3) 0.01

* The P value was calculated with the use of the Cochran–Mantel–Haenszel test, with adjustment for study. The P values
for the total number of adverse events and total number of drug-related adverse events were calculated by means of
Poisson regression.
† The investigator, who was unaware of the treatment assignment, assessed whether the adverse event was related to
treatment.
‡ One patient in the dalbavancin group died at day 32 from sepsis and a prior fracture. In the vancomycin–linezolid group,
two patients died from cardiopulmonary failure and one each from pulmonary emboli, congestive heart failure, acute
heart failure, and systemic lupus erythematosus; one patient died suddenly.
§ Treatment-limiting adverse events were those that led to the premature discontinuation of the study drug.
¶ The most common adverse events were defined as those that occurred in more than 2% of the patients in either treat-
ment group. A patient may have had more than one event.

4 patients (0.6%), respectively. The majority of A serious adverse event was reported in 17 of
infusion-related adverse events in patients in the 652 patients (2.6%) in the dalbavancin group and
dalbavancin group did not occur on administra- in 26 of 651 (4.0%) in the vancomycin–linezolid
tion day 1 or 8 but were related to the presence group (P = 0.16) (Table S8 in the Supplementary
of the indwelling catheter that was required for Appendix). Treatment-related serious adverse events
placebo infusions in the trial. were cellulitis and anaphylactoid reaction in 1 pa­
The median durations of adverse events were tient each in the dalbavancin group and cellu­
4.0 days (range, 1 to 101) in patients receiving litis, gastrointestinal disorder, toxic neph­rop­
dalbavancin and 3.0 days (range, 1 to 86) in those athy, and acute renal failure in 1 patient each in
receiving vancomycin–linezolid; the mean (±SD) the vancomycin–linezolid group. One patient
durations were 8.7±12.7 days and 8.7±12.6 days, (0.2%) in the dalbavancin group died, as com-
respectively. Adverse events at 28 days or later pared with 7 (1.1%) in the vancomycin–linezolid
after the initiation of treatment occurred in group (P = 0.03).
40 of 652 patients (6.1%) in the dalbavancin
group and in 59 of 651 (9.1%) in the vancomycin–­ DISCUSSION
linezolid group. Representative laboratory val-
ues obtained from testing performed according Our results show that for the treatment of acute
to the protocol are shown in Table S7 in the Sup- bacterial skin and skin-structure infection, the
plementary Appendix. efficacy of dalbavancin administered once weekly

n engl j med 370;23 nejm.org june 5, 2014 2177

The New England Journal of Medicine
Downloaded from nejm.org on June 24, 2014. For personal use only. No other uses without permission.
Copyright © 2014 Massachusetts Medical Society. All rights reserved.
 The n e w e ng l a n d j o u r na l of m e dic i n e

was not inferior to that of a conventional twice- signs of skin infection, particularly warmth,
daily antibiotic regimen. Our trial both met the fluctuance, and induration. The regulatory defi-
early primary end point of cessation of the spread nitions of major abscess, cellulitis, and wound
of infection and the absence of fever and also infection may not align with practice-based cri-
showed a consistent treatment effect regarding a teria. In addition, adherence to the twice-daily
reduction in the area of infection of 20% or more. vancomycin–linezolid regimen in the context of
Findings were consistent at the new early time a clinical trial may be greater than the adher-
point and the traditional later time points. ence observed in typical clinical practice.
Similarly, results were robust in patients with In conclusion, the results of these trials showed
major abscess, cellulitis, or wound infection; in the noninferiority of dalbavancin administered
those with S. aureus, including MRSA, or Strep. once weekly as compared with vancomycin–­
pyogenes infection; and in those treated as an out- linezolid administered twice daily for the treat-
patient. Patients included in the study were ill ment of acute bacterial skin and skin-structure
enough to require intravenous therapy and hospi- infection in seriously ill patients.
tal referral or admission and had a high rate of
fever with a larger median area of skin infection Supported by Durata Therapeutics.
Dr. Boucher reports receiving fees for serving on advisory
than that in patients included in two registra- boards from Basilea Pharmaceutica, Durata Therapeutics, Merck,
tional trials of acute bacterial skin and skin- Paratek Pharmaceuticals, Rib-X Pharmaceuticals, Theravance,
structure infection.15,16 Approximately 50% of the and Hospira/TheraDoc and fees for serving on an adjudication
committee from Merck. Dr. Wilcox reports receiving consulting
patients in our study met the criteria for the sys- fees from Actelion Pharmaceuticals, Astellas Pharma, AstraZeneca,
temic inflammatory response syndrome. Cerexa, Merck, Nabriva Therapeutics, Novacta Biosystems, Novar-
Patients treated with dalbavancin had fewer tis, Optimer Pharmaceuticals, Pfizer, Roche, Sanofi Pasteur,
Summit, the Medicines Company, VHsquared, Abbott Laborato-
adverse events than those treated with vancomy- ries, bioMérieux, and the European Tissue Symposium, lecture
cin–linezolid. The duration of treatment-related fees from Astellas Pharma, AstraZeneca, Cerexa, Merck, and
adverse events was similar for the two regimens. Pfizer, lecture fees through his institution from Alere, and grant
support from Actelion Pharmaceuticals, Astellas Pharma, Cerexa,
Late-onset events, a potential concern for a drug Merck, Summit, Abbott Laboratories, bioMérieux, Da Volterra,
with a long half-life, were infrequent and were and the European Tissue Symposium. Dr. Talbot reports receiv-
observed at a similar rate in the two treatment ing fees through Talbot Advisors for serving on advisory boards
from Actelion Pharmaceuticals, Astellas Pharma, Cubist Pharma-
groups. Infusion-related reactions related to dal- ceuticals, Durata Therapeutics, FAB Pharma, GlaxoSmithKline,
bavancin administered over a period of 30 min- and Kalyra Pharmaceuticals, fees for serving on the board of
utes were not more frequent than those associ- directors of Nabriva Therapeutics, fees for serving on an inde-
pendent data monitoring committee from Pfizer, consulting
ated with vancomycin administered over a period fees from Achaogen, Actelion Pharmaceuticals, Basilea Pharma-
of 120 minutes. Although the safety profile ob- ceutica, Cempra Pharmaceuticals, Cerexa, Durata Therapeutics,
served in these studies is consistent with the FAB Pharma, FujiFilm Pharmaceuticals, Johnson & Johnson,
Kalidex Pharmaceuticals, Merada Pharmaceuticals, Nabriva Thera-
profile in previous clinical trials,5,9 the true rate peutics, Paratek Pharmaceuticals, Sanofi, Sentinella, Shionogi,
of adverse events, especially rare ones, can be and Theravance, lecture fees from Bayer and Merck, and owning
established only after more extensive clinical use. stock or stock options in Achaogen, Calixa Therapeutics, Cempra
Pharmaceuticals, Cerexa, Durata Therapeutics, Kalyra Pharma-
Our trial had limitations. The new regulatory ceuticals, and Nabriva Therapeutics. Drs. Puttagunta and Dunne
end point mandates assessment of the primary report being employees of Durata Therapeutics and owning
efficacy end point at an early time point. The stock in Pfizer and Durata Therapeutics. Dr. Das reports receiving
consulting fees through her employer from Achaogen, Durata
analyses of traditional secondary end points Therapeutics, Trius Therapeutics, Cempra Pharmaceuticals,
confirmed those of the primary end point. Some Cerexa, Nabriva Therapeutics, Paratek Pharmaceuticals, and
differences between the clinical-status assess- Cubist Pharmaceuticals. No other potential conflict of interest
relevant to this article was reported.
ment and the investigator’s overall assessment Disclosure forms provided by the authors are available with
resulted from the evaluation of the subjective the full text of this article at NEJM.org.

References

1. Jenkins TC, Knepper BC, Sabel AL, et al. Clinical practice guidelines by the Infec- 3. Daum RS. Skin and soft-tissue infec-
Decreased antibiotic utilization after im- tious Diseases Society of America for the tions caused by­methicillin-resistant Staphy­
plementation of a guideline for inpatient treatment of methicillin-resistant Staphy- lococcus aureus. N Engl J Med 2007;357:380-
cellulitis and cutaneous abscess. Arch In- lococcus aureus infections in adults and 90. [Erratum, N Engl J Med 2007;357:1357.]
tern Med 2011;171:1072-9. children. Clin Infect Dis 2011;52(3):e18-e55. 4. Jenkins TC, Sabel AL, Sarcone EE,
2. Liu C, Bayer A, Cosgrove SE, et al. [Erratum, Clin Infect Dis 2011;53:319.] Price CS, Mehler PS, Burman WJ. Skin

2178 n engl j med 370;23 nejm.org june 5, 2014

The New England Journal of Medicine

Downloaded from nejm.org on June 24, 2014. For personal use only. No other uses without permission.
Copyright © 2014 Massachusetts Medical Society. All rights reserved.
 Dalbavancin vs. Conventional Ther apy for Skin Infection

and soft-tissue infections requiring hos- therapy for catheter-related bloodstream Society of America, the American Society
pitalization at an academic medical cen- infection caused by gram-positive patho- of Health-System Pharmacists, and the
ter: opportunities for antimicrobial stew- gens. Clin Infect Dis 2005;40:374-80. Society of Infectious Diseases Pharmacists.
ardship. Clin Infect Dis 2010;51:895-903. 9. Seltzer E, Dorr MB, Goldstein BP, Perry Clin Infect Dis 2009;49:325-7. [Erratum,
5. Jauregui LE, Babazadeh S, Seltzer E, M, Dowell JA, Henkel T. Once-weekly dalba- Clin Infect Dis 2009;49:1465.]
et al. Randomized, double-blind compari- vancin versus standard-of-care antimicro- 14. Miettinen O, Nurminen M. Compara-
son of once-weekly dalbavancin versus bial regimens for treatment of skin and tive analysis of two rates. Stat Med 1985;
twice-daily linezolid therapy for the treat- soft-tissue infections. Clin Infect Dis 2003; 4:213-26.
ment of complicated skin and skin struc- 37:1298-303. 15. Corey GR, Wilcox MH, Talbot GH,
ture infections. Clin Infect Dis 2005;41:1 10. Guidance for industry: acute bacterial Thye D, Friedland D, Baculik T. CANVAS 1:
407-15. skin and skin structure infections: devel- the first Phase III, randomized, double-blind
6. Dorr MB, Jabes D, Cavaleri M, et al. oping drugs for treatment. Silver Spring, study evaluating ceftaroline fosamil for the
Human pharmacokinetics and rationale MD: Food and Drug Administration, 2010. treatment of patients with complicated skin
for once-weekly dosing of dalbavancin, a 11. Snodgrass WR, Anderson T. Sulpha- and skin structure infections. J Anti­microb
semi-synthetic glycopeptide. J Antimicrob nilamide in the treatment of erysipelas. Chemother 2010;65:Suppl 4:iv41-iv51.
Chemother 2005;55:Suppl 2:ii25-ii30. Br Med J 1937;2:1156-9. 16. Prokocimer P, De Anda C, Fang E,
7. Nord CE, Rasmanis G, Wahlund E. 12. Idem. Prontosil in erysipelas. Br Med Mehra P, Das A. Tedizolid phosphate vs
Effect of dalbavancin on the normal intes- J 1937;2:101-4. linezolid for treatment of acute bacterial
tinal microflora. J Antimicrob Chemother 13. Rybak MJ, Lomaestro BM, Rotschafer skin and skin structure infections: the
2006;58:627-31. JC, et al. Vancomycin therapeutic guide- ­ESTABLISH-1 randomized trial. JAMA 2013;
8. Raad I, Darouiche R, Vazquez J, et al. lines: a summary of consensus recom- 309:559-69.
Efficacy and safety of weekly dalbavancin mendations from the Infectious Diseases Copyright © 2014 Massachusetts Medical Society.

n engl j med 370;23 nejm.org june 5, 2014 2179

The New England Journal of Medicine
Downloaded from nejm.org on June 24, 2014. For personal use only. No other uses without permission.
Copyright © 2014 Massachusetts Medical Society. All rights reserved.