n engl j med
353;2
www.nejm.org july
14, 2005
123
The
new england
journal
of
medicine
established in 1812
july
14
,
2005
vol. 353 no. 2
Erlotinib in Previously Treated Non–Small-Cell Lung Cancer
Frances A. Shepherd, M.D., José Rodrigues Pereira, M.D., Tudor Ciuleanu, M.D., Eng Huat Tan, M.D.,
Vera Hirsh, M.D., Sumitra Thongprasert, M.D., Daniel Campos, M.D., Savitree Maoleekoonpiroj, M.D.,
Michael Smylie, M.B., Ch.B., Renato Martins, M.D., Maximiliano van Kooten, M.D., Mircea Dediu, M.D.,
Brian Findlay, M.D., Dongsheng Tu, Ph.D., Dianne Johnston, Andrea Bezjak, M.D., Gary Clark, Ph.D.,
Pedro Santabárbara, M.D., Ph.D., and Lesley Seymour, M.D., Ph.D.,
for the National Cancer Institute of Canada Clinical Trials Group*
abstract
From the Departments of Medical Oncol-
ogy and Hematology (F.A.S.) and Radiation
Oncology (A.B.), the University Health Net-
work, Princess Margaret Hospital Site,
and the University of Toronto (F.A.S., A.B.)
— both in Toronto; the Instituto de Can-
cer Arnaldo Vieira de Carvalho, São Paulo
(J.R.P.); the Oncological Institute Ion Chiri-
cuta, Cluj-Napoca, Romania
(T.C.); the De-
partment of Medical Oncology, National
Cancer Centre, Singapore (E.H.T.); the De-
partment of Oncology, McGill University,
Montreal (V.H.); the Faculty of Medicine,
Chiangmai University, Chiangmai, Thailand
(S.T.); the Confidence Medical Center, San
Isidro, Argentina (D.C.); Pramongkutklao
Hospital, Bangkok, Thailand (S.M.); Cross
Cancer Institute, Edmonton, Alta., Canada
(M.S.); the Instituto Nacional de Cancer,
Praça da Cruz Vermelha, Rio de Janeiro,
Brazil (R.M.); the Instituto Medico Alex-
ander Fleming, Buenos Aires (M.K.); the
Oncology Institute, Bucharest, Romania
(M.D.); Hôtel Dieu Health Sciences Hos-
pital, St. Catharines, Ont., Canada (B.F.);
the National Cancer Institute of Canada
Clinical Trials Group, Kingston, Ont., Can-
ada (D.T., D.J., L.S.); and OSI Pharmaceu-
ticals, Boulder, Colo. (G.C., P.S.).
*The investigators and centers participat-
ing in this National Cancer Institute of
Canada Clinical Trials Group study are list-
ed in the Appendix.
N Engl J Med 2005;353:123-32.
Copyright © 2005 Massachusetts Medical Society.
background
We conducted a randomized, placebo-controlled, double-blind trial to determine
whether the epidermal growth factor receptor inhibitor erlotinib prolongs survival in
non–small-cell lung cancer after the failure of first-line or second-line chemotherapy.
methods
Patients with stage IIIB or IV non–small-cell lung cancer, with performance status
from 0 to 3, were eligible if they had received one or two prior chemotherapy regimens.
The patients were stratified according to center, performance status, response to prior
chemotherapy, number of prior regimens, and prior platinum-based therapy and were
randomly assigned in a 2:1 ratio to receive oral erlotinib, at a dose of 150 mg daily,
or placebo.
results
The median age of the 731 patients who underwent randomization was 61.4 years;
49 percent had received two prior chemotherapy regimens, and 93 percent had re-
ceived platinum-based chemotherapy. The response rate was 8.9 percent in the erlo-
tinib group and less than 1 percent in the placebo group (P<0.001); the median duration
of the response was 7.9 months and 3.7 months, respectively. Progression-free sur-
vival was 2.2 months and 1.8 months, respectively (hazard ratio, 0.61, adjusted for
stratification categories; P<0.001). Overall survival was 6.7 months and 4.7
months, respectively (hazard ratio, 0.70; P<0.001), in favor of erlotinib. Five percent
of patients discontinued erlotinib because of toxic effects.
conclusions
Erlotinib can prolong survival in patients with non–small-cell lung cancer after first-
line or second-line chemotherapy.
The New England Journal of Medicine
Downloaded from nejm.org at TRIAL HUAZHONG UNIVERSITY OF S&T on March 7, 2012. For personal use only. No other uses without permission.
Copyright © 2005 Massachusetts Medical Society. All rights reserved.
n engl j med
353;2
www.nejm.org july
14
,
2005
The
new england journal
of
medicine
124
ung cancer is the leading cause of
cancer death among men and women in
North America.
1
In advanced non–small-cell
lung cancer, chemotherapy offers symptomatic re-
lief and modest improvement in survival
2
; respons-
es are brief, with a median time to progression of
three to five months. Second-line chemotherapy
with docetaxel can prolong survival after platinum-
based therapy for non–small-cell lung cancer.
3,4
However, there is at present no defined role for
third-line chemotherapy. The futility of offering
third-line chemotherapy was demonstrated by Mas-
sarelli et al.,
5
who reported a response rate of only
2 percent and a median survival of four months.
Shepherd et al.
6
showed that among patients treat-
ed with docetaxel after the failure of two or more
chemotherapy regimens, survival was identical to
that among patients treated with supportive care.
The epidermal growth factor receptor (EGFR)
family is part of a complex signal-transduction net-
work that is central to several critical cellular pro-
cesses. Since EGFR is often found in non–small-
cell lung cancer cells,
7,8
it has been the focus of
efforts to develop new agents that target the EGFR
pathway. Erlotinib (Tarceva, OSI Pharmaceuticals)
and gefitinib (Iressa, AstraZeneca) inhibit the ty-
rosine kinase activity of EGFR and have been stud-
ied extensively.
9-12
In randomized phase 2 trials of
gefitinib (Iressa Dose Evaluation in Advanced Lung
Cancer [IDEAL] 1 and 2),
10,11
the tumors of 10 to
20 percent of patients who were previously treat-
ed with platinum-based regimens responded, and
in a phase 2 trial of erlotinib among previously
treated patients with non–small-cell lung cancer in
which 10 percent or more of the cells expressed
EGFR, the response rate was 12.3 percent.
12
These
promising rates are perhaps higher than those pos-
sible with other forms of chemotherapy,
3-6
but it
is unknown whether treatment with an EGFR inhib-
itor prolongs survival. For this reason, the National
Cancer Institute of Canada Clinical Trials Group
(NCIC CTG) conducted a trial (BR.21) to compare
erlotinib with placebo after the failure of standard
chemotherapy for non–small-cell lung cancer. The
inclusion of a control group receiving placebo was
considered ethical in view of the lack of benefit from
further chemotherapy after the failure of standard
treatment.
5,6
study design
This international, phase 3, randomized, double-
blind, placebo-controlled trial of erlotinib after the
failure of first-line or second-line chemotherapy
for non–small-cell lung cancer was designed by the
NCIC CTG. Patients were randomly assigned in a
2:1 ratio to receive oral erlotinib at a dose of 150 mg
daily or placebo. Randomization was performed
centrally by Applied Logic Associates (Houston),
with the use of the minimization method.
13
Pa-
tients were stratified according to center, Eastern
Cooperative Oncology Group performance status
(0 or 1 vs. 2 or 3, with higher scores indicating
greater impairment), best response to prior thera-
py (complete or partial response vs. stable disease
vs. progressive disease), number of prior regimens
received (one vs. two), and exposure to prior plati-
num therapy (yes vs. no).
The primary end point was overall survival.
Secondary end points included progression-free
survival, overall response rate (complete and par-
tial), duration of response, toxic effects, and qual-
ity of life. Responses were assessed with the use of
the Response Evaluation Criteria in Solid Tumors
(RECIST),
14
and toxic effects were assessed accord-
ing to the Common Toxicity Criteria of the Nation-
al Cancer Institute (version 2.0). The European Or-
ganization for Research and Treatment of Cancer
(EORTC) quality-of-life questionnaire (QLQ-C30)
and the quality-of-life questionnaire for patients
with lung cancer (QLQ-LC13) were used to evalu-
ate patients’ quality of life.
The protocol was approved by the ethics review
boards at all participating institutions, and all pa-
tients provided written informed consent. Support
was provided by the NCIC and OSI Pharmaceuti-
cals. Data were collected, managed, and analyzed
by the NCIC CTG, and the manuscript was written
by members of the NCIC CTG. OSI Pharmaceuti-
cals reviewed the final manuscript and provided
comments on it. Confidentiality was maintained by
both the NCIC CTG and OSI Pharmaceuticals. The
study chair, Dr. Shepherd, and the physician coordi-
nator, Dr. Seymour, reviewed all the data and con-
firmed their completeness and accuracy.
l methods
The New England Journal of Medicine
Downloaded from nejm.org at TRIAL HUAZHONG UNIVERSITY OF S&T on March 7, 2012. For personal use only. No other uses without permission.
Copyright © 2005 Massachusetts Medical Society. All rights reserved.
n engl j med
353;2
www.nejm.org july
14, 2005
erlotinib in previously treated non–small-cell lung cancer
125
eligibility criteria
Patients 18 years of age or older with an Eastern
Cooperative Oncology Group (ECOG) performance
status between 0 and 3 were eligible in the presence
of documented pathological evidence of non–small-
cell lung cancer. The patients had to have received
one or two regimens of combination chemother-
apy and not be eligible for further chemotherapy.
Patients 70 years of age or older may have received
therapy with one or two single agents. Patients had
to have recovered from any toxic effects of therapy
and were randomly assigned to the study treatment
at least 21 days after chemotherapy (14 days after
treatment with vinca alkaloids or gemcitabine) and
7 days after radiation. Adequate hematologic and
biochemical values were required.
Patients with prior breast cancer, melanoma, or
hypernephroma were ineligible, as were those with
other malignant diseases (except basal-cell skin
cancers) within the preceding five years. Other ex-
clusion criteria were symptomatic brain metasta-
ses, clinically significant cardiac disease within one
year, ventricular arrhythmias requiring medication,
and clinically significant ophthalmologic or gastro-
intestinal abnormalities.
study procedures
Within seven days before randomization, a history
and physical examination were obtained and hema-
tologic and biochemical testing, chest radiography,
and assessments of toxic effects and quality of life
were obtained. Computed tomographic scans of
the chest and abdomen were obtained within 28
days before randomization. For a patient to be eval-
uated for a response, at least one measurable lesion
was required, but measurable disease was not man-
datory for eligibility. Only patients with measur-
able disease were included in the analyses of com-
plete or partial response.
Administration of the study medication was to
start within two days after randomization. For
grade 2 diarrhea, loperamide was recommended
without reduction of the dose of erlotinib. For
grade 3 diarrhea, the study treatment was with-
held until the diarrhea was grade 1 or less, and then
erlotinib at a dose of 100 mg daily was started. For
grade 1 or 2 rash, treatment modification was not
recommended. For grade 3 rash, treatment was
withheld, the rash was treated symptomatically, and
erlotinib at a dose of 100 mg daily was restarted
when the rash was grade 1 or less.
History taking, physical examination, and he-
matologic and biochemical testing were performed
every four weeks, and radiologic investigations ev-
ery eight weeks. Patients’ quality of life was evaluat-
ed every four weeks in countries with validated ver-
sions of the questionnaires.
egfr expression
Separate written consent for optional tissue bank-
ing and correlative studies was obtained. EGFR ex-
pression was determined with the use of immuno-
histochemistry in a central laboratory that used
Dako kits (DakoCytomation). Positivity was defined
as more than 10 percent of cells staining at any in-
tensity for EGFR.
statistical analysis
The trial was designed to detect, with 90 percent
power and a two-sided type I error of 5 percent, a
33 percent improvement in median survival from
four months as estimated in the placebo group. For
the final analysis, 582 deaths were required and
were projected to occur with a sample size of 700
patients enrolled over a period of 14 months with
6 months of follow-up. The required number of
deaths had occurred by January 2004, and the data-
base was locked as of April 23, 2004. There was no
interim analysis. Tumor responses were validated
centrally for the first 333 patients in the trial.
The stratified log-rank test, accounting for strat-
ification factors at randomization (except center)
and EGFR protein expression (positive vs. negative
vs. unknown), was used to compare progression-
free survival and overall survival between treatment
groups. Exploratory forward stepwise regression
analyses with the use of the Cox model were per-
formed to adjust for treatment effect and to iden-
tify prognostic factors for progression-free survival
and overall survival. Candidate covariates includ-
ed EGFR expression, stratification factors (except
center), sex, age (60 years or less vs. more than 60
years), race or ethnic group (Asian vs. others), prior
radiotherapy (yes vs. no), histologic subtype of can-
cer (adenocarcinoma vs. others), and smoking sta-
tus (smoker vs. nonsmoker vs. unknown). Race was
self-reported or determined by study personnel and
was not based on country of domicile. Fisher’s exact
test was used to compare response rates between
levels of potential predictors and rates of toxic ef-
fects between treatments. Times to deterioration
(a 10-point increase from the baseline score) for
The New England Journal of Medicine
Downloaded from nejm.org at TRIAL HUAZHONG UNIVERSITY OF S&T on March 7, 2012. For personal use only. No other uses without permission.
Copyright © 2005 Massachusetts Medical Society. All rights reserved.
n engl j med
353;2
www.nejm.org july
14
,
2005
The
new england journal
of
medicine
126
cough, dyspnea, and pain were identified prospec-
tively as the primary end points for the analysis of
quality of life
15
and were analyzed with the use of
the log-rank test, with adjustment according to
the Hochberg method
16
for the comparison of
multiple end points. All P values were two-sided.
patient characteristics
Between August 2001 and January 2003, 731 pa-
tients were randomly assigned to erlotinib (488) or
placebo (243). Twenty-two patients (12 assigned to
erlotinib and 10 assigned to placebo) were ineligi-
ble for the following reasons: three prior chemo-
therapy regimens (9); single-agent chemotherapy
for patients less than 70 years of age (2); inadequate
time since the last treatment (5); abnormal bio-
chemistry results (4); and symptomatic brain me-
tastases (2). All 731 patients were included in the
efficacy analyses, and 727 treated patients (485 as-
signed to erlotinib and 242 assigned to placebo)
were included in the safety analyses. Eight patients
assigned to erlotinib (1.6 percent) and 18 assigned
to placebo (7.4 percent) received other EGFR in-
hibitors after study medication was discontinued.
The groups were balanced with respect to base-
line characteristics and important prognostic var-
iables (Table 1).
response and survival
In patients with at least one target lesion, the le-
sions were evaluated according to RECIST (427
patients assigned to erlotinib and 211 assigned to
placebo). In the erlotinib group, the rates of com-
plete response and partial response were 0.7 per-
cent and 8.2 percent, respectively (median dura-
tion, 7.9 months); in the placebo group, the rate of
partial response was less than 1 percent (P<0.001),
but these responses were not externally validated.
In an intention-to-treat analysis of all patients ran-
domly assigned to treatment, the disease-control
rate (i.e., the rate of complete or partial responses
and stable disease) in the erlotinib group was 45
percent; 38 percent of the patients had progressive
disease, and among the remaining 17 percent pro-
gression was not confirmed. The likelihood of a re-
sponse to erlotinib (Table 2) among patients with
non–small-cell lung cancer was not significantly
altered by performance status, prior treatments,
prior response, or age, but it was higher among
women (P=0.006), nonsmokers (P<0.001), Asians
results
* Because of rounding, not all percentages sum to 100.
† Race or ethnic group was self-reported or determined by study personnel and
was not based on country of domicile.
‡ A higher score indicates greater impairment.
§ Epidermal growth factor receptor (EGFR) expression was assessed by immu-
nohistochemistry.
Table 1. Baseline Characteristics of the Patients.*
Characteristic
Erlotinib
(N=488)
Placebo
(N=243)
Age (yr)
Median 62 59
Range 34–87 32–89
<60 (% of patients) 42.6 51.0
≥60 (% of patients) 57.4 49.0
Sex (% of patients)
Male 64.5 65.8
Female 35.5 34.2
Race or ethnic group (% of patients)†
Asian 12.9 12.2
Other 87.1 87.8
Performance status (% of patients)‡
0 13.1 14.0
1 52.5 54.3
2 25.8 23.0
3 8.6 8.6
Weight loss >10% (% of patients) 11.0 12.0
Pathological subtype (% of patients)
Adenocarcinoma 50.4 49.0
Squamous-cell carcinoma 29.5 32.1
Other 20.1 18.9
Prior chemotherapy (% of patients)
One regimen 50.6 50.2
Two or more regimens 49.4 49.8
Platinum-based therapy 92.0 91.8
Response to prior chemotherapy
(% of patients)
Complete or partial response 38.1 37.9
Stable disease 34.0 34.2
Progressive disease 27.9 28.0
Smoking status (% of patients)
Current smoker or ever smoked 73.4 77.0
Never smoked 21.3 17.3
Unknown 5.3 5.8
EGFR protein expression (% of patients)§
Positive 24.0 27.6
Negative 19.1 19.8
Unknown 56.9 52.6
The New England Journal of Medicine
Downloaded from nejm.org at TRIAL HUAZHONG UNIVERSITY OF S&T on March 7, 2012. For personal use only. No other uses without permission.
Copyright © 2005 Massachusetts Medical Society. All rights reserved.
n engl j med
353;2
www.nejm.org july
14, 2005
erlotinib in previously treated non–small-cell lung cancer
127
(P=0.02), patients with adenocarcinoma (P<0.001),
and patients in whom 10 percent or more of the tu-
mor cells expressed EGFR (P=0.10). In multiple
logistic-regression analyses, never having smoked
(P<0.001), the presence of adenocarcinoma (P=
0.01), and EGFR expression (P=0.03) were associ-
ated with responsiveness to erlotinib.
At the time of analysis, 587 deaths had occurred
(378 in the erlotinib group and 209 in the placebo
group). Figure 1 shows Kaplan–Meier curves for
overall survival and progression-free survival. Me-
dian overall survival in the erlotinib group was 6.7
months, and in the placebo group it was 4.7
months (adjusted hazard ratio, 0.70; 95 percent
confidence interval, 0.58 to 0.85; P<0.001). In the
Cox regression analysis, erlotinib remained asso-
ciated with longer survival (P=0.002), as did Asian
origin (P=0.01), adenocarcinoma on histologic
examination (P=0.004), and never having smoked
(P=0.048 vs. current or past smoking). Table 3
shows the exploratory subgroup analyses. Although
the sample sizes may be inadequate to detect small
or moderate differences, a benefit from erlotinib
was apparent in most of the subgroups. The inter-
action between treatment and the covariate defin-
ing the subgroup was statistically significant only
for smoking status. At the time of analysis, 682 pa-
tients had had progression of disease (450 in the
erlotinib gr