The new england journal of medicine

original article

Trastuzumab plus Adjuvant Chemotherapy

for Operable HER2-Positive Breast Cancer
Edward H. Romond, M.D., Edith A. Perez, M.D., John Bryant, Ph.D.,
Vera J. Suman, Ph.D., Charles E. Geyer, Jr., M.D., Nancy E. Davidson, M.D.,
Elizabeth Tan-Chiu, M.D., Silvana Martino, D.O., Soonmyung Paik, M.D.,
Peter A. Kaufman, M.D., Sandra M. Swain, M.D., Thomas M. Pisansky, M.D.,
Louis Fehrenbacher, M.D., Leila A. Kutteh, M.D.,
Victor G. Vogel, M.D., Daniel W. Visscher, M.D., Greg Yothers, Ph.D.,
Robert B. Jenkins, M.D., Ph.D., Ann M. Brown, Sc.D., Shaker R. Dakhil, M.D.,
Eleftherios P. Mamounas, M.D., M.P.H., Wilma L. Lingle, Ph.D.,
Pamela M. Klein, M.D., James N. Ingle, M.D., and Norman Wolmark, M.D.

abstract
From the National Surgical Adjuvant
background
Breast and Bowel Project, Pittsburgh
We present the combined results of two trials that compared adjuvant chemotherapy (E.H.R., J.B., C.E.G., E.T.-C., S.P., S.M.S.,
with or without concurrent trastuzumab in women with surgically removed HER2- L.F., V.G.V., G.Y., A.M.B., E.P.M., N.W.);
University of Kentucky, Lexington (E.H.R.);
positive breast cancer. North Central Cancer Treatment Group,
methods Rochester, Minn. (E.A.P., V.J.S., T.M.P.,
L.A.K., D.W.V., R.B.J., S.R.D., W.L.L., J.N.I.);
The National Surgical Adjuvant Breast and Bowel Project trial B-31 compared doxoru- Mayo Clinic Jacksonville, Jacksonville, Fla.
bicin and cyclophosphamide followed by paclitaxel every 3 weeks (group 1) with the (E.A.P.); University of Pittsburgh, Pitts-
same regimen plus 52 weeks of trastuzumab beginning with the first dose of paclitaxel burgh (J.B., G.Y.); Mayo Clinic, Rochester,
Minn. (V.J.S., T.M.P., D.W.V., R.B.J., W.L.L.,
(group 2). The North Central Cancer Treatment Group trial N9831 compared three J.N.I.); Allegheny General Hospital, Pitts-
regimens: doxorubicin and cyclophosphamide followed by weekly paclitaxel (group A), burgh (C.E.G., N.W.); Sidney Kimmel Com-
the same regimen followed by 52 weeks of trastuzumab after paclitaxel (group B), and prehensive Cancer Center at Johns Hop-
kins University, Baltimore (N.E.D.), the
the same regimen plus 52 weeks of trastuzumab initiated concomitantly with paclitaxel Cancer Research Network, Plantation, Fla.
(group C). The studies were amended to include a joint analysis comparing groups 1 and (E.T.-C.); Angeles Clinic and Research Insti-
A (the control group) with groups 2 and C (the trastuzumab group). Group B was ex- tute, Santa Monica, Calif. (S.M.); Norris
Cotton Cancer Center, Dartmouth–Hitch-
cluded because trastuzumab was not given concurrently with paclitaxel. cock Medical Center, Lebanon, N.H.
(P.A.K.); Cancer Therapeutics Branch, Na-
results tional Cancer Institute, Bethesda, Md.
By March 15, 2005, 394 events (recurrent, second primary cancer, or death before re- (S.M.S.); Kaiser Permanente Medical Cen-
currence) had been reported, triggering the first scheduled interim analysis. Of these, ter Northern California, Vallejo (L.F.); On-
cology Associates of Cedar Rapids, Cedar
133 were in the trastuzumab group and 261 in the control group (hazard ratio, 0.48; Rapids, Iowa (L.A.K.); University of Pitts-
P<0.0001). This result crossed the early stopping boundary. The absolute difference in burgh Cancer Institute, Pittsburgh (V.G.V.);
disease-free survival between the trastuzumab group and the control group was 12 per- Wichita Community Clinical Oncology Pro-
gram, Wichita, Kans. (S.R.D.); Aultman
cent at three years. Trastuzumab therapy was associated with a 33 percent reduction in Health Foundation, Canton, Ohio (E.P.M.);
the risk of death (P=0.015). The three-year cumulative incidence of class III or IV con- and Genentech, South San Francisco, Calif.
gestive heart failure or death from cardiac causes in the trastuzumab group was 4.1 (P.M.K.). Address reprint requests to Dr.
Geyer at the Allegheny Cancer Center, 5th
percent in trial B-31 and 2.9 percent in trial N9831. Fl., 320 E. North Ave., Pittsburgh, PA
15212, or at [email protected].
conclusions
Drs. Romond and Perez contributed equally
Trastuzumab combined with paclitaxel after doxorubicin and cyclophosphamide im- to the article.
proves outcomes among women with surgically removed HER2-positive breast cancer. N Engl J Med 2005;353:1673-84.
(clinicaltrials.gov numbers, NCT00004067 and NCT00005970.) Copyright © 2005 Massachusetts Medical Society.

n engl j med 353;16 www.nejm.org october 20, 2005 1673

The New England Journal of Medicine

Downloaded from nejm.org on April 2, 2023. For personal use only. No other uses without permission.
Copyright © 2005 Massachusetts Medical Society. All rights reserved.
 The new england journal of medicine

t rastuzumab, a monoclonal anti-

body targeting the extracellular domain of
the HER2 protein, was approved in 1998
as a first-line treatment in combination with pacli-
taxel for HER2-positive metastatic breast cancer.1
methods
eligibility and enrollment
Enrollment required a pathological diagnosis of
adenocarcinoma of the breast with immunohisto-
The benefit of this approach in patients with meta- chemical staining for HER2 protein of 3+ intensity
static disease and the poor prognosis of HER2-pos- or amplification of the HER2 gene on fluorescence
itive breast cancer2,3 motivated the National Can- in situ hybridization. Initially, both trials required
cer Institute (NCI) to sponsor two trials of adjuvant patients to have histologically proven, node-posi-
treatment with trastuzumab, led by the National tive disease; as of May 2, 2003, patients with high-
Surgical Adjuvant Breast and Bowel Project (NSABP) risk node-negative disease (defined as a tumor that
and the North Central Cancer Treatment Group was more than 2 cm in diameter and positive for
(NCCTG). estrogen receptors or progesterone receptors or
NSABP trial B-31, which began accrual in Febru- as a tumor that was more than 1 cm in diameter
ary 2000, compares four cycles of doxorubicin and and negative for both estrogen receptors and pro-
cyclophosphamide followed by paclitaxel (group 1) gesterone receptors) were eligible for trial N9831.
with the same chemotherapy plus 52 weeks of tras- Other requirements were adequate hematopoiet-
tuzumab beginning on day 1 of paclitaxel therapy ic, hepatic, and renal function and a left ventricu-
(group 2). NCCTG trial N9831 began enrollment lar ejection fraction (LVEF) that met or exceeded
in May 2000 and compares three regimens: four the lower limit of normal. Patients with clinical or
cycles of doxorubicin and cyclophosphamide fol- radiologic evidence of metastatic disease were ex-
lowed by weekly paclitaxel for 12 weeks (group A), cluded. Findings suggestive of metastasis were con-
four cycles of doxorubicin and cyclophosphamide firmed or refuted by additional radiologic evalua-
followed by 52 weeks of trastuzumab after the com- tion or biopsy. Complete resection of the primary
pletion of paclitaxel therapy (group B), and four tumor and axillary-node dissection were required
cycles of doxorubicin and cyclophosphamide fol- (negative sentinel-node biopsy was allowed in trial
lowed by 52 weeks of trastuzumab beginning on N9831). Patients were ineligible if they had angina
day 1 of paclitaxel therapy (group C). pectoris requiring antianginal medication, arrhyth-
The control groups of the trials, as well as group mia requiring medication, a severe conduction ab-
2 in trial B-31 and group C in trial N9831, differed normality, clinically significant valvular disease,
in terms of the scheduling of paclitaxel treatment cardiomegaly on chest radiography, left ventricular
and some aspects of hormonal therapy and radio- hypertrophy on echocardiography (trial B-31 only),
therapy but were otherwise identical. For this rea- poorly controlled hypertension, clinically signifi-
son, the NCI and the Food and Drug Administra- cant pericardial effusion (trial N9831 only), or a
tion approved a joint-analysis plan developed by the history of myocardial infarction, congestive heart
NSABP and NCCTG to combine data from group 1 failure, or cardiomyopathy.
and group A (referred to as the control group) for Participating institutions obtained the approval
comparison with group 2 and group C (referred to of their human investigations committee or insti-
as the trastuzumab group). Group B of trial N9831 tutional review board and filed assurances with the
was excluded because the protocol required tras- Department of Health and Human Services. Writ-
tuzumab to be administered after the completion ten informed consent was required for enrollment.
of chemotherapy. In trial B-31, treatment assignments were bal-
The plan required a first interim analysis after anced according to nodal status, the planned hor-
the occurrence of 355 events. Before the data were monal therapy, the type of surgery (lumpectomy vs.
locked, 2043 patients (of a planned total of 2700) mastectomy), the intended radiotherapy, and insti-
were enrolled in trial B-31 and 1633 patients (of a tution with the use of a biased-coin minimization
total of 2000 for the comparison of group A with algorithm.4 Trial N9831 used a dynamic allocation
group C) were enrolled in trial N9831. In April 2005, procedure that balanced the marginal distributions
the independent data-monitoring committees of of nodal status and hormone-receptor status be-
each trial recommended closing enrollment and tween groups.5 The procedures used for HER2 test-
releasing the results. ing in the two trials are provided in the Supple-

1674 n engl j med 353;16 www.nejm.org october 20 , 2005

The New England Journal of Medicine

Downloaded from nejm.org on April 2, 2023. For personal use only. No other uses without permission.
Copyright © 2005 Massachusetts Medical Society. All rights reserved.
 trastuzumab and adjuvant chemotherapy for early breast cancer

mentary Appendix (available with the full text of N9831 was amended on July 19, 2002, to permit
this article at www.nejm.org). treatment with any aromatase inhibitor in post-
menopausal patients with estrogen-receptor–posi-
treatment regimens tive or progesterone-receptor–positive tumors.
In trial B-31, treatment consisted of 60 mg of dox- Beginning January 14, 2003, trial B-31 permitted
orubicin per square meter of body-surface area and treatment with anastrozole in these patients.
600 mg of cyclophosphamide per square meter ev-
ery 21 days for four cycles, followed by 175 mg of assessment of cardiac function
paclitaxel per square meter every 3 weeks for four In both trials, LVEF was assessed before entry, after
cycles (group 1), or the same chemotherapy regi- the completion of doxorubicin and cyclophospha-
men plus trastuzumab, beginning with a loading mide therapy, and 6, 9, and 18 months after ran-
dose of 4 mg per kilogram of body weight, given domization. Trial B-31 required multiple gated ac-
with the first dose of paclitaxel, and followed by quisition scanning, whereas trial N9831 allowed
weekly doses of 2 mg per kilogram for 51 weeks multiple gated acquisition scanning or echocardi-
(group 2) (Fig. 1). Beginning on May 16, 2003, pac- ography. The initiation of trastuzumab required
litaxel could also be given weekly for 12 weeks at a an LVEF after doxorubicin and cyclophosphamide
dose of 80 mg per square meter at the investiga- therapy that met or exceeded the lower limit of nor-
tor’s discretion. Group A in trial N9831 used the mal and a decrease of less than 16 percentage points
same regimen of doxorubicin and cyclophospha- from baseline. Patients in whom clinically signifi-
mide as in trial B-31, followed by 12 weekly doses cant cardiac symptoms developed while they were
of paclitaxel at a dose of 80 mg per square meter. receiving doxorubicin and cyclophosphamide were
Group B received the same chemotherapy regimen, excluded from subsequent trastuzumab therapy.
followed by trastuzumab, beginning with a load- The six- and nine-month cardiac assessments were
ing dose of 4 mg per kilogram and followed by used to determine whether trastuzumab should be
weekly doses of 2 mg per kilogram for 51 weeks. continued in patients without cardiac symptoms. If
Group C received the same chemotherapy regimen the LVEF had declined 16 or more percentage
plus trastuzumab, beginning with a loading dose points from baseline or 10 to 15 percentage points
of 4 mg per kilogram, given with the first dose of from baseline to below the lower limit of normal,
paclitaxel, and followed by weekly doses of 2 mg trastuzumab was withheld for four weeks, at
per kilogram for 51 weeks (Fig. 1). which time the LVEF was reassessed. If the LVEF re-
In both trials, patients treated with lumpecto- mained below these levels or if the patient had
my were to receive whole-breast radiotherapy with symptomatic cardiac dysfunction while receiving
an optional boost to the tumor bed. Regional ra- trastuzumab, administration of the antibody was
diotherapy after lumpectomy or mastectomy was permanently discontinued.
optional in trial B-31 but required in trial N9831 for
patients with at least four positive nodes. In both role of the sponsor
studies, radiotherapy was initiated after the com- Both studies were conducted under a corporate
pletion of chemotherapy, and there was no irradia- research and development agreement between Gen-
tion of the internal-mammary nodes. Trastuzumab entech and the NCI. Genentech provided trastuz-
treatment was continued during radiotherapy. umab and partial funding support but did not par-
Women with estrogen-receptor–positive or pro- ticipate in the design of the studies or the collection
gesterone-receptor–positive tumors were to re- of data. The joint analysis was developed and ana-
ceive 20 mg of tamoxifen per day for five years. In lyzed by the NSABP and the NCCTG. The lead au-
trial N9831, hormonal therapy was given after thors wrote the manuscript, which was reviewed by
chemotherapy. In trial B-31, tamoxifen was initi- all authors. A draft was provided to Genentech for
ated on day 1 of the first cycle of doxorubicin and comment, but no changes in content or conclusions
cyclophosphamide until an amendment on Janu- were requested by Genentech. The authors vouch
ary 14, 2003, required hormonal therapy to be for the completeness and accuracy of the data.
started after chemotherapy, in response to the
findings of Southwest Oncology Group trial statistical analysis
8814.6 Following the report of the Arimidex, Ta- The primary end point was disease-free survival.
moxifen Alone or in Combination trial,7 trial Events determining disease-free survival were lo-

n engl j med 353;16 www.nejm.org october 20, 2005 1675

The New England Journal of Medicine

Downloaded from nejm.org on April 2, 2023. For personal use only. No other uses without permission.
Copyright © 2005 Massachusetts Medical Society. All rights reserved.
 The new england journal of medicine

Trial B-31

8 With follow-up
18 Declined protocol
1024 Assigned 10 With follow-up pending
therapy
to group I: as of Feb. 15, 2005
doxorubicin and 872 With follow-up
cyclophosphamide 864 With follow-up
plus paclitaxel 1006 Accepted protocol
142 With follow-up pending
therapy
as of Feb. 15, 2005

1 With follow-up
1019 Assigned 4 Declined protocol
3 With follow-up pending
to group 2: therapy
as of Feb. 15, 2005
doxorubicin and
864 With follow-up
cyclophosphamide
plus paclitaxel 863 With follow-up
1015 Accepted
and trastuzumab 152 With follow-up pending
protocol therapy
as of Feb. 15, 2005
1679 Women
in combined
Trial N9831 control group
4 With follow-up
819 Assigned 13 Declined protocol
9 With follow-up pending
to group A: therapy
as of March 15, 2005
doxorubicin and
807 With follow-up
cyclophosphamide
plus paclitaxel 803 With follow-up
806 Accepted
3 With follow-up pending
protocol therapy
as of March 15, 2005

152 Assigned
to group A 1672 Women in
from Feb. 1 to combined
152 Excluded from trastuzumab group
Sept. 3, 2002,
joint analysis
while group C
was closed
to enrollment

981 Assigned
to group B:
doxorubicin and
981 Excluded from
cyclophosphamide
joint analysis
plus paclitaxel
followed by
trastuzumab

814 Assigned 0 With follow-up

5 Declined protocol
to group C: 5 With follow-up pending
therapy
doxorubicin and as of March 15, 2005
cyclophosphamide 808 With follow-up
plus concomitant 808 With follow-up
paclitaxel and 809 Accepted
1 With follow-up pending
trastuzumab protocol therapy
as of March 15, 2005

Figure 1. Enrollment, Patients, and the Timing of Chemotherapy and Trastuzumab in Trial B-31 and Trial N9831.

cal, regional, and distant recurrence; contralateral weeks vs. weekly), nodal status (0, 1 to 3, 4 to 9, or
breast cancer, including ductal carcinoma in situ; 10 or more positive nodes), and hormone-receptor
other second primary cancers; and death before re- status (estrogen-receptor–positive or progester-
currence or a second primary cancer. Comparison one-receptor–positive vs. estrogen-receptor–neg-
of the two groups was based on a log-rank test, ative and progesterone-receptor–negative). Dis-
stratified according to the study (trial B-31 vs. trial ease-free survival was estimated according to the
N9831), intended paclitaxel schedule (every three Kaplan–Meier method. The primary cohort includ-

1676 n engl j med 353;16 www.nejm.org october 20 , 2005

The New England Journal of Medicine

Downloaded from nejm.org on April 2, 2023. For personal use only. No other uses without permission.
Copyright © 2005 Massachusetts Medical Society. All rights reserved.
 trastuzumab and adjuvant chemotherapy for early breast cancer

ed all enrolled patients with follow-up data, ana- that would preclude the initiation of trastuzumab
lyzed according to the intention-to-treat principle had they been randomly assigned to the investiga-
(Fig. 1). All reported P values are two-sided. tional group.
Other end points were overall survival, time to
distant recurrence, death from breast cancer, con- Data Lock
tralateral breast cancer, and other second primary Patients in trial B-31 include those enrolled as of
cancers. The Supplementary Appendix defines these February 15, 2005, with follow-up as of that date.
end points. We counted deaths as having been Patients in trial N9831 include those enrolled by
caused by breast cancer if they took place after re- November 1, 2004, with follow-up as of March
currence or if they were attributed to breast cancer. 15, 2005.

Timing of Analyses results

Definitive analysis was scheduled after 710 primary
end-point events had occurred to provide the study patients
with a statistical power of 90 percent to detect a 25 As of February 15, 2005, 2043 patients had been
percent reduction in the event rate, allowing for a enrolled in trial B-31; 1736 of these women had at
number of attenuating factors. The first interim least one follow-up evaluation (end-point follow-
analysis was to take place after 355 events had been up was not required for six months after random-
reported. Subsequent interim analyses were sched- ization). By November 1, 2004, 1633 patients had
uled to take place semiannually. Consideration was been enrolled in groups A and C of trial N9831, 1615
to be given to early reporting if disease-free survival of whom had follow-up data submitted by March
differed at the nominal 0.0001 level. If this bound- 15, 2005 (Fig. 1). Table 1 shows the characteris-
ary was not crossed, alpha spending function was tics of patients with follow-up. Except for 191 pa-
to be applied to the final test so that the overall type tients with node-negative breast cancer who were
I error rate would be 0.05. enrolled in trial N9831, the groups from each study
were similar.
Secondary Analyses
Cox models were fitted to adjust for nodal status compliance
(0 to 3, 4 to 9, or 10 or more positive nodes), tu- Chemotherapy
mor size (2.0 cm or less, 2.1 to 4.0 cm, or 4.1 cm or In the combined trials, 31 of 1843 patients as-
more), receptor status (estrogen-receptor–positive signed to the control group declined therapy, as did
or progesterone-receptor–positive vs. estrogen- 9 of 1833 women randomly assigned to the tras-
receptor–negative and progesterone-receptor–neg- tuzumab group. Of the patients who began chemo-
ative), age (39 years or younger, 40 to 49 years, 50 to therapy with doxorubicin and cyclophosphamide,
59 years, or 60 years of age or older), grade (poor 97.9 percent received four cycles; 2.7 percent of pa-
vs. other), histologic findings (ductal carcinoma vs. tients who completed doxorubicin and cyclophos-
other), and trial (B-31 vs. N9831). Forest plots were phamide treatment did not begin treatment with
constructed to show the effects according to the paclitaxel; of those who did, 94.7 percent complet-
study and subgroups of patients.8 Hazard ratios ed all cycles.
were computed and compared according to the
length of follow-up (one year or less, one to two Eligibility for Trastuzumab
years, two to three years, or more than three years Of 3497 patients who had an evaluation of LVEF af-
after randomization). ter doxorubicin and cyclophosphamide therapy,
233 (6.7 percent) had an LVEF that declined at least
Sensitivity Analysis 16 percentage points from baseline or that de-
The protocols required a secondary analysis that clined below the lower limit of normal or had car-
excluded patients who were ineligible, did not con- diac symptoms during such treatment that would
tinue therapy after receiving doxorubicin and cy- preclude the initiation of trastuzumab therapy.
clophosphamide, or were HER2-negative on central
review. Also excluded were patients with symptoms Discontinuation of Trastuzumab
of adverse cardiac effects or declines in LVEF dur- Of 1159 patients with an adequate LVEF after doxo-
ing doxorubicin and cyclophosphamide therapy rubicin and cyclophosphamide treatment who be-

n engl j med 353;16 www.nejm.org october 20, 2005 1677

The New England Journal of Medicine

Downloaded from nejm.org on April 2, 2023. For personal use only. No other uses without permission.
Copyright © 2005 Massachusetts Medical Society. All rights reserved.
 The new england journal of medicine

Table 1. Characteristics of the Patients.

All Patients
Characteristic Control Group Trastuzumab Group (N=3351)
Trial B-31 Trial N9831 Trial B-31 Trial N9831
(N=872) (N=807) (N=864) (N=808)
Ineligible — no. (%) 20 (2.3) 14 (1.7) 19 (2.2) 11 (1.4) 64 (1.9)
Age at randomization — no. (%)
≤39 yr 146 (16.7) 138 (17.1) 140 (16.2) 129 (16.0) 553 (16.5)
40–49 yr 304 (34.9) 274 (34.0) 306 (35.4) 272 (33.7) 1156 (34.5)
50–59 yr 294 (33.7) 272 (33.7) 280 (32.4) 261 (32.3) 1107 (33.0)
≥60 yr 128 (14.7) 123 (15.2) 138(16.0) 146 (18.1) 535 (16.0)
Histologically positive nodes — no. (%)
Unknown 0 4 (0.5) 0 0 4 (0.1)
0 0 102 (12.6) 0 89 (11.0) 191 (5.7)
1–3 494 (56.7) 386 (47.8) 496 (57.4) 403 (49.9) 1779 (53.1)
4–9 253 (29.0) 203 (25.2) 251 (29.1) 205 (25.4) 912 (27.2)
≥10 125 (14.3) 112 (13.9) 117 (13.5) 111 (13.7) 465 (13.9)
Estrogen-receptor status — no. (%)
Unknown 5 (0.6) 2 (0.2) 0 1 (0.1) 8 (0.2)
Negative 407 (46.7) 379 (47.0) 416 (48.1) 393 (48.6) 1595 (47.6)
Positive 460 (52.8) 426 (52.8) 448 (51.9) 414 (51.2) 1748 (52.2)
Progesterone-receptor status — no. (%)
Unknown 7 (0.8) 2 (0.2) 1 (0.1) 4 (0.5) 14 (0.4)
Negative 504 (57.8) 472 (58.5) 526 (60.9) 486 (60.1) 1988 (59.3)
Positive 361 (41.4) 333 (41.3) 337 (39.0) 318 (39.4) 1349 (40.3)
Tumor size — no. (%)
Unknown 19 (2.2) 10 (1.2) 12 (1.4) 4 (0.5) 45 (1.3)
≤2.0 cm 355 (40.7) 323 (40.0) 322 (37.3) 307 (38.0) 1307 (39.0)
2.1–4.0 cm 372 (42.7) 372 (46.1) 386 (44.7) 382 (47.3) 1512 (45.1)
≥4.1 cm 126 (14.4) 102 (12.6) 144 (16.7) 115 (14.2) 487 (14.5)
Tumor grade — no. (%)
Unknown 21 (2.4) 14 (1.7) 15 (1.7) 15 (1.9) 65 (1.9)
Good 23 (2.6) 8 (1.0) 20 (2.3) 13 (1.6) 64 (1.9)
Intermediate 256 (29.4) 212 (26.3) 239 (27.7) 217 (26.9) 924 (27.6)
Poor 572 (65.6) 573 (71.0) 590 (68.3) 563 (69.7) 2298 (68.6)
Planned hormonal therapy — no. (%)
Unknown 0 83 (10.3) 0 73 (9.0) 156 (4.7)
None 378 (43.3) 331 (41.0) 382 (44.2) 349 (43.2) 1440 (43.0)
Tamoxifen or aromatase inhibitor 494 (56.7) 393 (48.7) 482 (55.8) 386 (47.8) 1755 (52.4)
Intended paclitaxel schedule — no. (%)
Every 3 wk 806 (92.4) 0 805 (93.2) 0 1611 (48.1)
Weekly 66 (7.6) 807 (100) 59 (6.8) 808 (100) 1740 (51.9)

gan treatment with trastuzumab and have com- (1.9 percent), a confirmed asymptomatic decline in
pleted therapy, 364 (31.4 percent) discontinued the LVEF in 164 (14.2 percent), symptoms of conges-
treatment before 52 weeks. Reasons for discontin- tive heart failure or other adverse cardiac effect in
uation were recurrence in the case of 22 patients 54 (4.7 percent), noncardiac adverse effect or death

1678 n engl j med 353;16 www.nejm.org october 20 , 2005

The New England Journal of Medicine

Downloaded from nejm.org on April 2, 2023. For personal use only. No other uses without permission.
Copyright © 2005 Massachusetts Medical Society. All rights reserved.
 trastuzumab and adjuvant chemotherapy for early breast cancer

in 27 (2.3 percent), patient-initiated discontinua- vival rate at three years was 94.3 percent in the tras-
tion in 70 (6.0 percent), and other reasons in 27 tuzumab group and 91.7 percent in the control
(2.3 percent). group (absolute difference, 2.5 percentage points;
95 percent confidence interval, 0.1 to 5.0 percent-
disease-free survival, overall survival, age points); at four years, the respective rates were
and time to distant recurrence 86.6 percent and 91.4 percent (absolute difference,
The median follow-up was 2.0 years (2.4 years in 4.8 percentage points; 95 percent confidence inter-
trial B-31 and 1.5 years in trial N9831). There were val, 0.6 to 9.0 percentage points).
261 events in the control group and 133 events in Distant metastases were reported in 193 patients
the trastuzumab group. The hazard ratio for a first in the control group and 96 in the trastuzumab
event in the trastuzumab group, as compared with group. The hazard ratio for a first distant recur-
the control group, was 0.48 (95 percent confidence rence was 0.47 in the trastuzumab group as com-
interval, 0.39 to 0.59; P<0.0001) (Fig. 2A). The pared with the control group (95 percent confi-
percentages of patients alive and disease-free at dence interval, 0.37 to 0.61; P<0.0001) (Fig. 3). At
three years were 75.4 percent in the control group three years, 90.4 percent of women in the trastuzu-
and 87.1 percent in the trastuzumab group (abso- mab group were free of distant recurrence, as com-
lute difference, 11.8 percentage points; 95 percent pared with 81.5 percent of women in the control
confidence interval, 8.1 to 15.4 percentage points). group (absolute difference, 8.8 percentage points;
At four years, the respective percentages were 67.1 95 percent confidence interval, 5.5 to 12.1 percent-
percent and 85.3 percent (absolute difference, 18.2 age points); the respective rates at four years were
percentage points; 95 percent confidence interval, 89.7 percent and 73.7 percent (absolute difference,
12.7 to 23.7 percentage points). This difference 15.9 percentage points; 95 percent confidence in-
crossed the early stopping boundary. terval, 11.1 to 20.8 percentage points).
There were 62 deaths in the trastuzumab group,
as compared with 92 deaths in the control group sites of first reported events
(hazard ratio, 0.67; 95 percent confidence interval, The benefit of trastuzumab was evident at both lo-
0.48 to 0.93; P=0.015) (Fig. 2B). The absolute sur- cal–regional and distant sites (Table 2). The num-

A B
100 100 Trastuzumab
94.3% (62 deaths)
Trastuzumab 91.4%
(133 events) Control
90 87.1% 90 (92 deaths)
85.3% 91.7%
Disease-free Survival (%)

Overall Survival (%)

80 80
Control
(261 events) 86.6%

70 75.4% 70
67.1%

60 60
P<0.0001 P=0.015
Hazard ratio, 0.48 Hazard ratio, 0.67
50 50
0 0
0 1 2 3 4 5 0 1 2 3 4 5
Years after Randomization Years after Randomization
No. at Risk 3351 2379 1455 801 133 0 No. at Risk 3351 2441 1571 908 165 0
Control 1679 1162 689 374 59 0 Control 1679 1200 766 448 83 0
Trastuzumab 1672 1217 766 427 74 0 Trastuzumab 1672 1241 805 460 82 0

Figure 2. Kaplan–Meier Estimates of Disease-free Survival (Panel A) and Overall Survival (Panel B).
The hazard ratios are for the comparison of the trastuzumab group with the control group.

n engl j med 353;16 www.nejm.org october 20, 2005 1679

The New England Journal of Medicine

Downloaded from nejm.org on April 2, 2023. For personal use only. No other uses without permission.
Copyright © 2005 Massachusetts Medical Society. All rights reserved.
 The new england journal of medicine

breast cancer treated with trastuzumab.9-11 In both

Trastuzumab
100
(N=1672; 96 events)
trials, the incidence of isolated brain metastases as
Patients Free of Distant Recurrence (%)

90.4% first events was higher in the trastuzumab group

89.7%
90 than in the control group (21 vs. 11 in trial B-31
Control
81.5% and 12 vs. 4 in trial N9831). Since patients in trial
80 (N=1679; 193 events)
73.7%
B-31 were followed for additional recurrences be-
yond the first distant event, we could determine
70 whether the imbalance was due to masking of the
incidence of brain metastases in the control group
60 as a result of earlier failures in other organs. In trial
P<0.0001
Hazard ratio, 0.47 B-31, brain metastases as a first or subsequent
50 event were diagnosed in 28 patients in the trastuz-
0 umab group, as compared with 35 patients in the
0 1 2 3 4 5
control group (hazard ratio, 0.79; P=0.35). The
Years after Randomization
imbalance in brain metastases as first events can
Figure 3. Kaplan–Meier Estimates of Freedom from Distant Recurrence. therefore be attributed to earlier failures at other
The hazard ratios are for the comparison of the trastuzumab group with the distant sites among patients in the control group.
control group.
additional secondary end points
Table 3 summarizes additional secondary end
ber of contralateral breast cancers was insufficient points, including the time to recurrence, death from
to evaluate the effect of trastuzumab. There was a breast cancer, the occurrence of contralateral breast
reduction in nonbreast second primary cancers in cancer, and the occurrence of other second primary
trial B-31 in the trastuzumab group as compared cancers. Sites of second cancers are summarized
with the control group, but the results of trial in Table 4.
N9831 did not confirm (or refute) this trend.
Six patients in the control group died without estimated treatment benefit after
recurrence or second cancers, as did eight patients adjustment for additional characteristics
in the trastuzumab group. These included three A Cox model was fitted to disease-free survival to
treatment-related deaths that occurred in patients control for treatment assignment, nodal status,
who received paclitaxel and trastuzumab, one as a pathological tumor size, hormone-receptor status,
result of cardiomyopathy and two as a result of in- age, tumor grade, histologic appearance of the tu-
terstitial pneumonitis. mor, and trial. Adjustment for these factors mini-
An increased frequency of brain metastases has mally altered the effect of trastuzumab, as compared
been reported among patients with metastatic with that of control therapy (hazard ratio for a first

Table 2. Sites of First Events.

Patients Trial B-31 Trial N9831

Control Group Trastuzumab Group Control Group Trastuzumab Group

number of patients
All patients with follow-up 872 864 807 808
Patients alive and event-free 701 781 717 758
Patients with any first event 171 83 90 50
Local or regional recurrence 35 15 22 12
Distant recurrence 111 60 63 30
Contralateral breast cancer 6 2 0 1
Other second primary cancer 15 2 3 3
Death with no evidence of disease 4 4 2 4

1680 n engl j med 353;16 www.nejm.org october 20 , 2005

The New England Journal of Medicine

Downloaded from nejm.org on April 2, 2023. For personal use only. No other uses without permission.
Copyright © 2005 Massachusetts Medical Society. All rights reserved.
 trastuzumab and adjuvant chemotherapy for early breast cancer

Table 3. Summary of Efficacy End-Point Analyses.*

Trastuzumab Control Hazard Ratio

End Point Group Group Total (95% CI) P Value

no. of patients
Disease-free survival (primary end point) 133 261 394 0.48 (0.39–0.59)† <0.0001
Time to recurrence 117 235 352 0.47 (0.38–0.59) <0.0001
Time to distant recurrence 96 193 289 0.47 (0.37–0.61) <0.0001
Overall survival 62 92 154 0.67 (0.48–0.93)‡ 0.015
Death from breast cancer 53 79 132 0.66 (0.47–0.94) 0.02
Contralateral breast cancer 4 6 10 0.64 (0.18–2.27) 0.48
Other second primary cancer 5 20 25 0.24 (0.09–0.64) 0.002

* All P values were two-sided. CI denotes confidence interval.

† The hazard ratio is for a first event.
‡ The hazard ratio is for death.

event, 0.46; 95 percent confidence interval, 0.37 variation in hazard rates over time
to 0.56; P<0.001). The number of positive nodes, Figure 4 in the Supplementary Appendix includes
pathological tumor size, hormone-receptor status, plots of the hazard rates for first events, first recur-
and tumor grade were significant predictors of dis- rence, distant recurrence, and death as a function
ease-free survival. There was no evidence that the of time since randomization. After the first year,
benefit of trastuzumab differed significantly be- the hazard rate for distant recurrence rose sharply
tween the two studies (P=0.38) (see Figure 1 in the in the control group; in the trastuzumab group, it
Supplementary Appendix). first rose and then decreased markedly after year 2.

subgroups adverse cardiac events

A Forest plot of the hazard ratios for first events is The principal adverse event associated with tras-
shown in Figure 2 in the Supplementary Appendix. tuzumab therapy among patients with prior expo-
Only in subgroups in which there were a negligible sure to anthracycline is cardiac dysfunction.12,13 In
number of events (negative nodes and a low tumor trial B-31, of the patients who remained free of car-
grade) did the 95 percent confidence interval fail diac symptoms during doxorubicin and cyclophos-
to exclude 1. Plots of disease-free survival accord- phamide therapy and who had LVEF values that
ing to nodal and estrogen-receptor status are also met requirements for the initiation of trastuzumab
shown in Figure 3 in the Supplementary Appendix. therapy, the cumulative incidence of New York
Heart Association class III or IV congestive heart
sensitivity analysis failure or death from cardiac causes at three years
A secondary analysis excluded ineligible patients, was 0.8 percent in the control group (4 patients
those who had received only doxorubicin and cy- had congestive heart failure, and 1 died from cardi-
clophosphamide, those found on central testing to ac causes) and 4.1 percent in the trastuzumab
be HER2-negative, and those who had symptomat- group (31 patients had congestive heart failure). Of
ic cardiac dysfunction during therapy with doxoru- the 31 women in the trastuzumab group who had
bicin and cyclophosphamide or reductions in LVEF congestive heart failure, 27 have been followed for
that would preclude treatment with trastuzumab. at least six months after the onset of heart failure,
As compared with the control group, the trastuzu- and only 1 reported persistent symptoms of heart
mab group had improved disease-free survival (haz- failure at the most recent follow-up visit. Details of
ard ratio for a first event, 0.45; P<0.0001), a longer the cardiac effects of trastuzumab in trial B-31 are
time to recurrence (hazard ratio, 0.45; P<0.0001), reported elsewhere.14 In trial N9831, the three-
a longer time to distant recurrence (hazard ratio, year cumulative incidence of New York Heart Asso-
0.46; P<0.0001), and improved overall survival ciation class III or IV congestive heart failure or
(hazard ratio for death, 0.61; P=0.01). death from cardiac causes was 0 percent in the

n engl j med 353;16 www.nejm.org october 20, 2005 1681

The New England Journal of Medicine

Downloaded from nejm.org on April 2, 2023. For personal use only. No other uses without permission.
Copyright © 2005 Massachusetts Medical Society. All rights reserved.
 The new england journal of medicine

centage points). The reduction was similar among

Table 4. Incidence of Second Primary Cancers.
women with hormone-receptor–negative tumors
Control Trastuzumab and women with hormone-receptor–positive tu-
Site or Type of Second Primary Cancer Group Group Total mors. No subgroups that did not appear to benefit
no. of patients from trastuzumab therapy were identified. Since
Esophagus 1 1 2
only 191 women with node-negative breast cancer
were included in these studies and only 3 had had
Stomach 0 1 1
an event at the time of the analysis, we cannot com-
Colon 1 0 1
ment on the effect of trastuzumab in this sub-
Pancreas 1 0 1 group.
Lung 2 1 3 The addition of trastuzumab reduced the mor-
Bone 1 0 1 tality rate by one third (P=0.015). Among eligible
Melanoma 3 0 3 patients who continued treatment after doxorubi-
Contralateral breast 6 4 10
cin and cyclophosphamide and who were HER2-
positive on central testing, the relative reduction in
Endometrium 1 1 2
the mortality rate associated with trastuzumab was
Ovary 3 0 3
39 percent (P=0.01). Although relatively little fol-
Thyroid 4 0 4 low-up information is available beyond three years,
Multiple myeloma 1 0 1 current data rule out a risk of distant recurrence
Acute myelogenous leukemia 2 1 3 among trastuzumab-treated women of greater than
Total 26 9 35 27 per 1000 women per year, in contrast to a risk of
90 per 1000 women per year in the control group
(see Figure 4 in the Supplementary Appendix). Since
control group and 2.9 percent in the trastuzumab the risk of distant recurrence should remain ap-
group (20 patients had congestive heart failure, preciable in the control group for some time, a sub-
1 of whom died of cardiomyopathy). stantial additional survival benefit related to trastuz-
umab can be anticipated.
other adverse events The effect of trastuzumab was substantial in both
During treatment with paclitaxel alone or with trials (see Figure 1 in the Supplementary Appen-
trastuzumab, there was little imbalance between dix), a finding that is noteworthy given differences
treatment groups in the incidence of any Common in the paclitaxel schedule and the timing of hor-
Toxicity Criteria version 2.0 category except for a monal therapy. The benefit of trastuzumab was ev-
higher incidence of left ventricular dysfunction in ident at local or regional and distant sites. Although
the trastuzumab group. However, rare cases of in- isolated brain metastases were more common first
terstitial pneumonitis were reported that in some events in the trastuzumab group than in the control
cases appeared to be related to trastuzumab thera- group, the imbalance can be attributed to the occur-
py. In trial B-31, four patients in the trastuzumab rence of earlier failures at other distant sites in the
group had interstitial pneumonitis, one of whom control group.
died. In the N9831 trial, five patients in the trastuz- Data from trial B-31 suggested that trastuzu-
umab group had grade 3+ pneumonitis or pulmo- mab reduced the incidence of nonbreast second
nary infiltrates, one of whom died. primary cancers, an unanticipated effect requiring
independent verification. There was no obvious
discussion pattern of site or histologic type (Table 4).
The primary concern regarding the safety of
The addition of trastuzumab to paclitaxel after a trastuzumab is the increased risk of cardiac dys-
regimen of doxorubicin and cyclophosphamide re- function associated with past or concurrent an-
duced the rates of recurrence by half among wom- thracycline treatment.12,13 In both studies, the cu-
en with HER2-positive breast cancer. The absolute mulative three-year incidence of congestive heart
decreases in distant recurrence were 8.8 percent- failure increased by about 3 percentage points with
age points after three years and 15.9 percentage the addition of trastuzumab. Most episodes oc-
points after four years, although the latter value curred during trastuzumab treatment, but addition-
had a wide confidence interval (11.1 to 20.8 per- al follow-up will be needed to define the long-term

1682 n engl j med 353;16 www.nejm.org october 20 , 2005

The New England Journal of Medicine

Downloaded from nejm.org on April 2, 2023. For personal use only. No other uses without permission.
Copyright © 2005 Massachusetts Medical Society. All rights reserved.
 trastuzumab and adjuvant chemotherapy for early breast cancer

cardiotoxicity of trastuzumab. Clearly, appropriate the rate associated with chemotherapy alone.16
selection and careful cardiac monitoring of pa- Since only 26 percent of patients received taxanes
tients are essential. Trastuzumab did not increase in the HERA trial, comparison of those results with
the overall frequency or severity of noncardiac ad- ours may be problematic. Therefore, further fol-
verse effects associated with the chemotherapy regi- low-up of groups B and C in trial N9831 is neces-
mens, but we did see rare cases of interstitial pneu- sary for an adequate evaluation of the efficacy of
monitis in patients receiving trastuzumab during or concurrent as compared with sequential adminis-
shortly after the paclitaxel phase of treatment. Two tration of trastuzumab.
cases were fatal. Supported by Public Health Service grants (U10-CA-12027, U10-
Trial N9831 was also designed to address the ef- CA-69651, U10-CA-37377, and U10-CA-69974 to the NSABP; U10-
CA-25224 to the NCCTG; U10-CA-021115 to the Eastern Coopera-
ficacy of trastuzumab initiated concurrently with tive Oncology Group; U10-CA-32102 to the Southwest Oncology
paclitaxel as opposed to sequentially (group C vs. Group; and U10-CA-31946 to Cancer and Leukemia Group B) from
group B), but this comparison requires substan- the NCI, Department of Health and Human Services. Genentech,
South San Francisco, Calif., provided trastuzumab and partial fund-
tially longer follow-up than was needed for the as- ing for both trials. The Breast Cancer Research Foundation, New
sessment of concurrent trastuzumab therapy in the York, provided partial funding to Dr. Perez.
joint analysis. However, after reviewing the results Drs. Romond, Perez, Bryant, Geyer, Tan-Chiu, Paik, Kaufman,
and Mamounas report having received compensation for time
of the first joint interim efficacy analysis, the data- served on Genentech Breast Cancer Advisory Boards; Drs. Romond,
monitoring committee overseeing trial N9831 re- Tan-Chiu, Bryant, Kaufman, and Mamounas, lecture fees from Gen-
quested an unplanned comparison of groups B and entech; and Drs. Perez and Kaufman, grant support from Genen-
tech. Dr. Klein is an employee of Genentech and holds equity stock.
C and subsequently recommended disclosure of We are indebted to Dr. Barbara C. Good, director of scientific
the results. Though early, the comparison suggest- publications for the NSABP, for editorial assistance; to all the wom-
ed delayed administration of trastuzumab may be en participating in NSABP trial B-31 and NCCTG trial N9831 for
their contribution to this research effort; to the investigators of the
less effective than concurrent administration.15 Eastern Cooperative Oncology Group, the Southwest Oncology
Recent data from the Herceptin Adjuvant (HERA) Group, Cancer and Leukemia Group B, the NCCTG, and the NSABP
Trial showed that treatment with trastuzumab be- for their dedicated efforts in support of these trials; and to Dr. Jeff
Abrams, for support in the development and approval of the joint-
gun after the completion of chemotherapy sub- analysis plan.
stantially reduced the rate of recurrence relative to

ap p e n d i x
The following institutions and principal investigators enrolled at least 25 patients in the B-31 trial or N9831 trial: Trial B-31 — Kaiser Per-
manente, Northern California Region, Vallejo, Calif., L. Fehrenbacher; University of Pittsburgh, Pittsburgh, V.G. Vogel; Atlanta Regional
Community Clinical Oncology Program (CCOP), Atlanta, T.E. Seay; Colorado Cancer Research Program, CCOP, Denver, E.R. Pajon; Metro
Minnesota CCOP, St. Louis Park, Minn., P.J. Flynn; Franklin Square Hospital Center, Baltimore, J.L. Zapas; Kaiser Permanente, San Diego,
Calif., J. Polikoff; Dayton CCOP, Dayton, Ohio, H.M. Gross; Christiana Care Health Services CCOP, Newark, Del., D.D. Biggs; Southeast
Cancer Control Consortium CCOP, Winston-Salem, N.C., J.N. Atkins; Huntsman Cancer Institute, Salt Lake City, Utah, R.D. Noyes; Puget
Sound Oncology Consortium, Seattle, R.B. Clarfeld; Columbus CCOP, Columbus, Ohio, J.P. Kuebler; Northwest CCOP, Tacoma, Wash.,
L.K. Colman; Scripps Clinic, La Jolla, Calif., J.F. Kroener; Illinois Oncology Research Association CCOP, Peoria, J.W. Kugler; Evanston
Northwestern Healthcare CCOP/Kellogg Cancer Center, Evanston, Ill., D. Merkel; Kansas City CCOP, Kansas City, Mo., W.T. Stephenson;
Montana Cancer Consortium CCOP, Billings, P.W. Cobb; Trial N9831 — Indiana University Cancer Center, Indianapolis, P.J. Loehrer; Johns
Hopkins University, Baltimore, A.A. Forastiere; Vanderbilt University, Nashville, D.H. Johnson; Northern New Jersey CCOP, Hackensack,
R.J. Rosenbluth; University of Chicago Medical Center, Chicago, G. Fleming; Dana–Farber Cancer Institute, Boston, G.P. Canellos; Duke
University Medical Center, Durham, N.C., J. Crawford; Mount Sinai School of Medicine, New York, L.R. Silverman; Memorial Sloan-Ketter-
ing Cancer Center, New York, C. Hudis; Loyola University Medical Center, Maywood, Ill., P.J. Stiff; Ohio State University Medical Center, Co-
lumbus, C.D. Bloomfield; Georgetown University Medical Center, Washington, D.C., E. Gelmann; Moffitt Cancer Center, Tampa, Fla., J.A.
Kish; Mayo Clinic, Rochester, Minn., S.R. Alberts; Toledo CCOP, Toledo, Ohio, P.L. Schaefer; Metro Minnesota CCOP, St. Louis Park,
Minn., P.J. Flynn; Wichita CCOP, Wichita, Kans., S. Dakhil; Ann Arbor CCOP, Ann Arbor, Mich., P.J. Stella; Missouri Valley Cancer Consor-
tium CCOP, Omaha, Nebr., J.A. Mailliard.

references
1. Slamon DJ, Leyland-Jones B, Shak S, et 3. Paik S, Bryant J, Tan-Chiu E, et al. HER2 nostic factors in the controlled clinical trial.
al. Use of chemotherapy plus a monoclonal and choice of adjuvant chemotherapy for Biometrics 1975;31:103-15.
antibody against HER2 for metastatic breast invasive breast cancer: National Surgical 6. Albain KS, Green S, Ravdin P, et al. Ad-
cancer that overexpresses HER2. N Engl J Adjuvant Breast and Bowel Project Protocol juvant chemohormonal therapy for primary
Med 2001;344:783-92. B-15. J Natl Cancer Inst 2000;92:1991-8. breast cancer should be sequential instead
2. Slamon DJ, Clark GM, Wong SG, Levin 4. White SJ, Freedman LS. Allocation of of concurrent: initial results from North
WJ, Ullrich A, McGuire WL. Human breast patients to treatment groups in a controlled American Intergroup Trial 0100 (SWOG-
cancer: correlation of relapse and survival clinical study. Br J Cancer 1978;37:849-57. 8814). Proc Am Soc Clin Oncol 2002;21:37A
with amplification of the HER-2/neu onco- 5. Pocock SL, Simon R. Sequential treat- abstract.
gene. Science 1987;235:177-82. ment assignment with balancing for prog- 7. Baum M, Buzdar AU, Cuzick J, et al.

n engl j med 353;16 www.nejm.org october 20, 2005 1683

The New England Journal of Medicine

Downloaded from nejm.org on April 2, 2023. For personal use only. No other uses without permission.
Copyright © 2005 Massachusetts Medical Society. All rights reserved.
 trastuzumab and adjuvant chemotherapy for early breast cancer

Anastrozole alone or in combination with 11. Lin NU, Bellon JR, Winer EP. CNS metas- for patients with node-positive HER2-over-
tamoxifen versus tamoxifen alone for adju- tases in breast cancer. J Clin Oncol 2004;22: expressing breast cancer. J Clin Oncol (in
vant treatment of postmenopausal women 3608-17. press).
with early breast cancer: first results of the 12. Seidman A, Hudis C, Pierri MK, et al. 15. Perez EA, Suman VJ, Davidson N, et al.
ATAC randomised trial. Lancet 2002;359: Cardiac dysfunction in the trastuzumab clin- HER2 testing by local, central and reference
2131-9. [Erratum, Lancet 2002;360:1520.] ical trials experience. J Clin Oncol 2002;20: laboratories in the NCCTG N9831 trial.
8. Lewis S, Clarke M. Forest plots: trying to 1215-21. Available at http://asco.org/ac/1,1003,_12-
see the wood and the trees. BMJ 2001;322: 13. Ewer MS, Vooletich M, Valero V, Higano 002511-00_18-0034-00_19-005815,00.asp.
1479-80. CS, Benjamin RS. Trastuzumab (Herceptin) 16. Piccart-Gebhart M. A randomized three-
9. Bendell JC, Domchek SM, Burstein HJ, cardiotoxicity: clinical course and cardiac group multi-centre comparison of: 1 year
et al. Central nervous system metastases in biopsy correlations. Proc Am Soc Clin Oncol Herceptin, 2 years Herceptin, or no Her-
women who receive trastuzumab-based ther- 2002;21:123A. abstract. ceptin, in women with HER-2 positive pri-
apy for metastatic breast carcinoma. Cancer 14. Tan Chiu E, Yothers G, Romond E, et al. mary breast cancer who have completed
2003;97:2972-7. Assessment of cardiac dysfunction in NSABP adjuvant chemotherapy. Available at http://
10. Clayton AJ, Danson S, Jolly S, et al. Inci- B-31, a randomized trial comparing doxo- asco.org/ac/1,1003,_12-002511-00_18-
dence of cerebral metastases in patients rubicin and cyclophosphamide (AC) fol- 0034_19-005816,00.asp.
treated with trastuzumab for metastatic lowed by paclitaxel (2AC) followed by pac- Copyright © 2005 Massachusetts Medical Society.
breast cancer. Br J Cancer 2004;91:639-43. litaxel plus trastuzumab as adjuvant therapy

physician-journalist
The Journal is seeking a physician with substantial reporting
experience to write articles on timely topics in medicine and society
for the Perspective section. Send curriculum vitae and writing samples
to Perspective Editor, New England Journal of Medicine, 10 Shattuck St.,
Boston, MA 02115, or at [email protected].

1684 n engl j med 353;16 www.nejm.org october 20 , 2005

The New England Journal of Medicine

Downloaded from nejm.org on April 2, 2023. For personal use only. No other uses without permission.
Copyright © 2005 Massachusetts Medical Society. All rights reserved.