Small-cell lung cancer: patients included in clinical trials are not representative of the patient population as a whole.
ABSTRACT To determine the clinical characteristics of the subpopulation of patients not included in clinical trials, their outcome, and the reasons for their ineligibility and non-participation.
We studied 57 patients (out of 178 consecutive patients with SCLC), who were not included in any of the three successive clinical trials completed at our center during the study period. We also compared 37 patients excluded from the largest clinical trial to their 73 included counterparts.
Reasons for ineligibility (n = 53) included low Karnofsky index (n = 17), advanced age (n = 12), non-feasible long-term follow-up (n = 12), previous history of cancer (n = 8), contraindication for anthracyclines (n = 5), and other medical reasons (n = 11). Only four eligible patients were not included in the trials. As compared to patients included in the studies, non-included patients had a significantly lower Karnofsky index, were older, presented more frequently with metastatic disease, and had a lower response rate to treatment and a shorter survival. However, exclusion from the trial was not an independent prognostic factor by multivariate analysis.
Selection biases were unlikely in the three trials, based on the high ratio of included/eligible patients. However, the subgroup of patients included in the trials was not representative of the patient population as a whole because of restrictive eligibility criteria. Results from published clinical trials to the overall population should be extrapolated only with caution. We suggest that the proportion and major characteristics of ineligible and non-participating patients be mentioned in any publication of a clinical trial.
- SourceAvailable from:[Show abstract] [Hide abstract]
ABSTRACT: BACKGROUND The purposes of this study were to identify prognostic factors for response to chemotherapy, overall survival, and long term survival of patients with small cell lung carcinoma and to construct a classification of patients on the basis of their expected overall survival.METHODS In the 763 patients registered in 4 consecutive clinical trials conducted by the European Lung Cancer Working Party from 1982 to 1993, the impact of 21 pretreatment variables assessable in a routine practice was analyzed for the various outcomes with a minimum follow-up of 5 years.RESULTSThe key prognostic role of disease extent was confirmed for all the outcomes. Additional independent prognostic factors for response to chemotherapy were gender, neutrophil count, and hemoglobin level; for overall survival, these factors were Karnofsky performance status, gender, and neutrophil count. Recursive partitioning and amalgamation algorithms (RECPAM) analysis classified patients into 4 groups, taking into consideration disease extent, Karnofsky performance status, age, gender, and neutrophil count. Median survival times for the 4 groups were 60, 47, 36, and 28 weeks, respectively. For long term survival, defined as a minimum survival of 2 years (9% of the patients), Karnofsky performance status was the only independent predictive factor, along with the achievement of a complete response (if this was taken into consideration). Small cell lung carcinoma remained the main cause of death among these patients. Cure was infrequent, with only 14 patients alive and disease free at 5 years (1.8%).CONCLUSIONS In this study the long term prognosis associated with small cell lung carcinoma was poor. The well-known prognostic values of disease extent and Karnofsky performance status were confirmed, but the authors also identified age and gender (which are more controversial) as independent characteristics, in addition to citing the role of complete response in the attainment of long term survival. The independent role of neutrophils observed by the authors. must be validated by further studies. Cancer 2000;89:523–33. © 2000 American Cancer Society.Cancer 08/2000; 89(3). · 5.20 Impact Factor
- [Show abstract] [Hide abstract]
ABSTRACT: BACKGROUND: Stakeholders derive many benefits from cancer clinical trials, including guidance for future oncologic treatment decisions. However, whether enrollment in cancer trials also improves patient survival independently of trial outcomes remains underinvestigated. We hypothesized that cancer trial enrollment is not associated with patient survival outcomes. STUDY DESIGN: Using the 2002 to 2008 California Cancer Registry, we identified 555,469 patients with stage I to IV solid organ tumors. Baseline characteristics were compared by trial participation status. Logistic regression determined predictors of trial enrollment. Multivariate Cox proportional hazards regression examined the impact of trial participation on overall and cancer-specific mortality with adjustment for covariates. RESULTS: Only 0.33% of our cohort was enrolled in clinical trials. Trial participants were likely to be younger than 65 (odds ratio [OR] 2.13; 95% CI 1.90 to 2.38), Hispanic rather than non-Hispanic white (OR 0.78; 95% CI 0.67 to 0.90), and have breast cancer (OR 3.14; 95% CI 2.62 to 3.77). Multivariate survival analyses demonstrated that enrollment in cancer trials predicted a lower hazard of death. However, when stratified by disease site, this survival benefit was observed only in lung, colon, and breast cancers. Sensitivity and interaction analyses confirmed these relationships. CONCLUSIONS: In this first population-based study examining trial effect in solid organ cancers, enrollment into cancer trials predicted lower overall and cancer-specific mortality among common cancer sites. Although these findings may demonstrate a survival benefit due to trial enrollment, they likely also reflect the favorable attributes of trial enrollees. Once corroborated, stakeholders must consider broader cancer trial designs representative of the cancer burden treated in the real world.Journal of the American College of Surgeons 02/2013; · 4.45 Impact Factor
- [Show abstract] [Hide abstract]
ABSTRACT: Clinical trials test the efficacy of a treatment in a select patient population. We examined whether cancer clinical trial patients were similar to nontrial, "real-world" patients with respect to presenting characteristics and survival. We reviewed the SWOG national clinical trials consortium database to identify candidate trials. Demographic factors, stage, and overall survival for patients in the standard arms were compared with nontrial control subjects selected from the Surveillance, Epidemiology, and End Results program. Multivariable survival analyses using Cox regression were conducted. The survival functions from aggregate data across all studies were compared separately by prognosis (≥50% vs <50% average 2-year survival). All statistical tests were two-sided. We analyzed 21 SWOG studies (11 good prognosis and 10 poor prognosis) comprising 5190 patients enrolled from 1987 to 2007. Trial patients were younger than nontrial patients (P < .001). In multivariable analysis, trial participation was not associated with improved overall survival for all 11 good-prognosis studies but was associated with better survival for nine of 10 poor-prognosis studies (P < .001). The impact of trial participation on overall survival endured for only 1 year. Trial participation was associated with better survival in the first year after diagnosis, likely because of eligibility criteria that excluded higher comorbidity patients from trials. Similar survival patterns between trial and nontrial patients after the first year suggest that trial standard arm outcomes are generalizable over the long term and may improve confidence that trial treatment effects will translate to the real-world setting. Reducing eligibility criteria would improve access to clinical trials.CancerSpectrum Knowledge Environment 03/2014; · 14.07 Impact Factor
Annals of Oncology 10: 809-815, 1999.
1 1999 Kiuwer Academic Publishers. Printed in the Netherlands.
Small-cell lung cancer: Patients included in clinical trials are not
representative of the patient population as a whole
V. Cottin,1'2'5 D. Arpin,2 C. Lasset,3 J.-F. Cordier,2 J. Brune,2 F. Chauvin3 & V. Trillet-Lenoir1'4
lUnite d'Oncologie Medicate, Centre Hospitaller Lyon-Sud, Pierre-Benite; 2Service de Pneumologie, Hopilal Universitaire Louis Pradel;
Unite de Biostatisliques, Centre Leon Berard; AEA643, Universite Claude Bernard, Lyon, France; 5Present address: National Jewish Medical
and Research Center, Denver, CO, USA
Purpose: To determine the clinical characteristics of the
subpopulation of patients not included in clinical trials, their
outcome, and the reasons for their ineligibility and non-partic-
Patients and methods: We studied 57 patients (out of 178
consecutive patients with SCLC), who were not included in
any of the three successive clinical trials completed at our
center during the study period. We also compared 37 patients
excluded from the largest clinical trial to their 73 included
Results: Reasons for ineligibility (n = 53) included low
Karnofsky index (n — 17), advanced age (/? = 12), non-feasible
long-term follow-up (n = 12), previous history of cancer (n = 8),
contraindication for anthracyclines (n = 5), and other medical
reasons (« = 11). Only four eligible patients were not included
in the trials. As compared to patients included in the studies,
non-included patients had a significantly lower Karnofsky
index, were older, presented more frequently with metastatic
disease, and had a lower response rate to treatment and a
shorter survival. However, exclusion from the trial was not an
independent prognostic factor by multivariate analysis.
Conclusions: Selection biases were unlikely in the three
trials, based on the high ratio of included/eligible patients.
However, the subgroup of patients included in the trials was
not representative of the patient population as a whole because
of restrictive eligibility criteria. Results from published clinical
trials to the overall population should be extrapolated only with
caution. We suggest that the proportion and major character-
istics of ineligible and non-participating patients be mentioned
in any publication of a clinical trial.
Key words: clinical trials, eligibility, small-cell lung cancer
Treatment modalities in medicine must be evaluated and
compared by means of clinical trials. However, the
highly selected groups of patients included in trials
represent only a fraction of the eligible patients; in turn,
the group of eligible patients is a small subgroup of the
physician's target population, which itself is a sample of
the overall population of patients . Thus, it may be
erroneous to relate the data obtained from such poorly
representative samples to the entire affected population.
Firstly, the group of patients included in trials may
differ from the eligible population as a whole because of
patient and physician selection biases . For instance,
discrepancies between the results of different clinical
trials, as well as frequent failure to confirm the results
of small trials in large multicenter trials, may be due to
such biases. Previous studies have shown that selection
biases are more likely to occur when the proportion of
included patients as a percentage of eligible patients is
Secondly, the study population may differ from the
general population because of restrictive eligibility crite-
ria, even in the absence of selection bias . Restricting
the eligible population may limit the applicability of the
results and the ability to generalize the data from a given
study to the relevant patient population. Survival, re-
sponse and tolerance to treatment are commonly used
endpoints in clinical oncology protocols, and are clearly
related to the initial state of the patient. Advanced age is
one of the disqualifying criteria in many lung cancer
clinical trials, because of a presumed higher risk of
toxicity, although this may be irrelevant . Since the
prognostic impact of age on treatment results has been
shown in some studies , one may expect different
treatment results between eligible and ineligible patients.
Therefore, eligibility criteria may affect treatment results,
and are of critical importance in the design of clinical
trials. However, very few studies have been conducted to
determine the proportion of ineligible patients and the
reasons for their ineligibility [2, 4, 7], as well as the
prognosis of ineligible patients as compared to eligible
The aims of the present retrospective study were: (1)
to determine what fraction of patients referred to our
center for treatment of small-cell lung cancer (SCLC)
were not enrolled in one of the three successive clinical
trials that we completed during a five-year period [8-11];
by guest on July 13, 2011
(2) to determine the reasons for ineligibility and/or non-
participation; (3) to characterize the clinical features of
the patients not included in the trials; (4) to compare
their clinical characteristics as well as treatment issues
to those of patients included in clinical trials.
Patients and methods
Table 1. Design of the three different clinical trials performed during
the study period.
GLOT PC (86/01) AR (88/01) GLOT PC (89/01)
1/86 to 5/89
6/89 to 3/90
4/90 to 7/91
Descriptive analysis of the population of patients not included in the
From January 1986 to July 1991, 178 consecutive patients with SCLC
were referred to the Respiratory Disease Department at Louis Pradel
University Hospital, Lyon, France. Of this initial group, 121 (69%)
were included in one of the three clinical trials in which our group
participated during this period (Table 1). Most of the selection criteria
for these three trials were identical and included pathology proof of
SCLC, absence of previously diagnosed cancer (except for basocellular
carcinoma of the skin or in situ carcinoma of the uterine cervix), no
prior chemotherapy, contraindication to surgical resection, absence of
cardiac contraindication to the use of anthracyclines, and feasiblity of
long-term follow-up. Some selection criteria varied from one trial to
another and are listed in Table 2. All three trials were approved by a
local ethics committee. Based on the assessment of the above-listed
selection criteria, 57 patients were not included in the on-going trials.
They were registered, managed and treated individually, and strictly
followed. We retrospectively studied the initial clinical characteristics
of this subpopulation of excluded patients, the reasons for their
ineligibility and non-participation, the staging procedures they under-
went, the treatments they received, and their outcomes.
Table 2. Main exclusion criteria varying from one trial to another.
GLOT PC (86/01) AR (88/01)GLOT PC (89/01)
Informed consent Oral
> 50 umol/1,
Karnofsky < 50%
platelets < 100
Neutrophils < 1,
platelets < 60
> 130 >110
> 10 Not mentioned
> 35 umol/1,
> 2.5 N, alkaline
> 50 umol/1,
Comparative analysis of included and non-included patients
Because of the above-mentioned differences in selection criteria from
one trial to another, and because detailed data on patients included in
the AR 88/01 trial were not available, we chose to perform the com-
parative analysis only between patients included (// = 73) and those
not included (n = 37) in the GLOT PC 86/01 trial. In order to facilitate
the interpretation of the comparative results, the procedures of the
GLOT PC 86/01 trial are summarized below.
Pre-lreatment staging in the GLOT PC 86/01 study
Staging procedures for patients selected for the GLOT PC 86/01 phase
II study included a chest X-ray, assessment of blood cell counts, liver
function tests (aspartate and alanine amino-transferases, gamma
glutamyl-transferase, bilirubin, alkaline phosphatase, prothrombin
time), serum creatinine levels, a fiberoptic bronchoscopy with multiple
biopsies and cytology of bronchial aspirates, a unilateral iliac crest
bone marrow aspiration and biopsy, an abdominal ultrasonography,
brain and thoracic (and optionally abdominal) CT scans and a radio-
nucleide bone scan. Limited disease was defined as a disease encom-
passable in the initial large radiation field, including homolateral
pleural effusion, pericardial involvement and/or controlateral hilar
mass. Extensive disease was defined on the basis of any evidence of
extrathoracic spread and/or supraclavicular lymph nodes.
Treatment modalities in the GLOT PC 86/01 study
Patients included in the phase II study were treated as previously
reported [8, 9]. Briefly, induction chemotherapy (DEI) consisted of a
combination of doxorubicin 50 mg/m2 on day 1, etoposide 150 mg/m2
on days 1 and 2, ifosfamide 2 g/m2 using a short two-hour infusion on
days 1 and 2, and intravenous mesna 30 min before (2 g) and four and
eight hours after (1 g) ifosfamide infusion, repeated every 21 days. For
patients with limited disease, the treatment consisted of four courses of
induction chemotherapy, followed by three courses of modified chemo-
therapy, alternated with radiotherapy given twice a day at a total dose
of 51 Gy, and completed by two cycles of the initial chemotherapy.
Patients with extensive disease received six additional cycles of DEI for
a total of 10 courses.
Assessment of treatment efficacy and toxicity in the GLOT PC 86/01
Assessment of the initial response of all patients included in the trial
was performed two weeks after the fourth chemotherapy course. The
restaging procedures included a chest radiograph, a CT scan of the
chest and a fiberoptic bronchoscopy with biopsies and cytology of
bronchial aspirates. In addition, any test with initially positive results
had to be repeated at this time. Complete response was defined as a
complete regression of the primary tumor, including normalization of
bronchoscopic abnormalities. Partial response was defined as a > 50%
overall regression of the tumor, or a complete clinical regression with
persistence of positive bronchial cytology or biopsy results. Objective
response was defined by either complete or partial response. All tumor
response evaluations were confirmed by an independant review com-
mittee. Early death was defined by less than two months' survival.
Toxicity was evaluated using the World Health Organisation's criteria.
All registered patients were included in the statistical analysis. Com-
parisons were calculated using the chi-squared test for qualitative
values, and the Student's /-test for quantitative values. Overall survival
curves were calculated using the Kaplan-Meier method . The
endpoints were the date of the first course of chemotherapy, and the
date of death or relapse or the last date of follow-up in surviving
patients. Prognostic factors were determined using the log rank test
for univariate analysis and the Cox model for multivariate analysis
. Results were considered significant for /"-values less than 0.05.
by guest on July 13, 2011
Characteristics of the population of non-included patients
Of 178 consecutive patients with SCLC who were re-
ferred to our institution during the study period, 57
(31%) were not included in the on-going trials. We
studied the clinical characteristics of this subpopulation
of excluded patients and the reasons for their non-
participation in the protocols. This group consisted of
48 males and 9 females, with a mean age of 67 ± 8 years
(range 41-88). The age was significantly higher in the
female population (73 vs. 65, P = 0.02). Twenty-eight
percent of the patients had histories of severe coronary
or peripheral artery disease.
Reasons for non-participation
Patients who were not included in clinical trials either (i)
met ineligibility criteria, or (ii) were eligible but refused
to participate or were not asked to enroll. As shown in
Table 3, 53 patients (93% of non-participants) were con-
sidered ineligible. Four patients were eligible but refused
to enter the study (n = 2), or were not asked to participate
(« = 2, mainly because they were members of the family
of a physician). Given this low fraction of eligible pa-
tients among the non-participants, the likelihood of
significant selection biases in the three clinical trials was
low. Of the 53 ineligible patients, 18 (31.5 %) had multi-
ple reasons for non-enrollment in the study. Long-term
medical follow-up was considered non-feasible for 12
patients: 7 lived too far from our institution and 5 were
unable to understand the treatment protocol and to
follow the precise schedules. The sites of the previous
tumor of the 8 patients who could not be included be-
cause of a history of cancer were head and neck (n - 2),
bladder (n - 2), non-SCLC (n = 1), prostate (« = 1), cen-
tral nervous system (n = 1) and breast (« = 1).
For the 57 non-included patients, the staging procedures
were left to the discretion of the physician whereas they
were mandatory in the clinical trials. Patients not in-
cluded in the protocols underwent only a fraction of the
procedures for initial staging and response assessment
(data not shown). Overall, only 18 (31%) patients under-
went all listed staging procedures. According to the
staging definition used by our group (see above), 43
(75%) non-participating patients were considered to
have extensive disease.
Similarly, the treatment modalities for non-included
patients were left to the discretion of the physician.
Seven patients received only symptomatic treatment.
One patient with limited disease received only chest
radiotherapy. Chemotherapy was given to 49 (86%) of
the patients as an initial treatment, and was followed by
brain or chest radiotherapy in 32% of these patients. The
Table 3. Reasons for non-inclusion.
Reasons Patients not
included in the included in the
(n = 57)
GLOT PC 86/
01 only (n = 37)
Low performance status
Age over the limit
Non-feasible long-term follow-up
Previous history of cancer
Contraindication to anthracyclines
Serum creatinine over the limit
Prothrombin time over the limit
Lack of histological proof
Previous chemotherapy for SCLC
Other reasons of non-participation
The numbers in parentheses refer to the numbers of patients presenting
each criteria as the only reason for non-inclusion.
chemotherapy regimens included (1) the above-described
DEI combination; (2) a multidrug regimen consisting of
cyclophosphamide 500 mg/m2 and vincristin 1 mg/m2
on days 1 and 8, CCNU 60 mg/m2 orally on day 1 and
procarbazin 50 mg/m2 orally on days 1 to 14; and (3) a
combination of cisplatin 100 mg/m2 on day 1 and etopo-
side 100 mg/m2 on days 1 to 3.
Response to therapy was assessed for 42 of the 50 treated
patients. Objective tumor regression was described in 18
patients (36%). Forty-five patients died as a result of
cancer progression, five of drug-related toxicity, and five
of various complications apparently unrelated to treat-
ment (three of cerebral stroke, one of gastro-intestinal
hemorrhage and one of peritonitis). None of the nine
patients excluded because of previous history of cancer
showed evidence of progression of any neoplastic disease
other than SCLC. The average survival time in this
population was six months.
The prognostic factors that significantly affected sur-
vival were a Karnofsky index lower than 50% (P - 0.002),
age greater than or equal to 70 years (P - 0.02), and the
presence of metastasis in the central nervous system
(P - 0.01). Objective response to therapy (P - 0.016) and
exclusion from the study only for geographical reasons
(P - 0.034) were associated with a better outcome.
Neither weight loss, disease stage, nor any other cate-
gory of the exclusion criteria had a statistically signifi-
cant impact on survival.
Comparative analysis of included and non-included patients
To further characterize the subpopulation of non-partic-
ipants, we compared the pre-treatment characteristics of
by guest on July 13, 2011
Table 4. Comparison of the characteristics of the patients.
Table 5. Comparative results of pre-treatment staging.
(n = 73, %)
(n = 37, %)
Age (mean ± SD)
Karnofsky index < 50%
Weight loss > 10%
Compression of the vena
57 + 6.9
4(5) 3(8) NS
Abbreviation: NS - non-significant.
the 73 patients included in the GLOT PC 86/01 trial to
those of the 37 patients considered ineligible for this
trial during the same period. The causes of exclusion of
these 37 patients were representative of those of the
overall population of excluded patients. As shown in
Table 4, patients not included in the phase II study had
a significantly lower Karnofsky index and were older
than included patients. Hemoptysis was more frequent
in the included than in the excluded patients, whereas
other symptoms did not differ significantly between the
Since disease stage is a major prognostic factor in pa-
tients with SCLC, we compared the results of the pre-
treatment staging procedures performed in included and
excluded patients. Extensive disease was significantly
more frequent in excluded patients than in patients
included in the GLOT PC 86/01 trial (Table 5). Cerebral
metastases were more common in non-enrolled patients,
while no difference was found for the other sites of
Treatment modalities and results
Of the 37 excluded patients, 16 (43%) were offered DEI
chemotherapy, the treatment given in the on-going
GLOT PC 86/01 trial. The toxic effects of the treatment
were not worse in these 16 patients than in patients
included in the GLOT PC 86/01 trial (i.e., 12% vs. 14%
of grade IV hematologic toxicity, P > 0.05). As shown in
Table 6, the objective response rate was significantly
lower in the excluded patients (40%), than in patients
included in the phase II study (78%). Non-enrolled
patients had worse prognoses, as shown by a lower
survival rate (Table 6 and Figure 1). The causes of death
were similar in the two groups, with the majority of the
patients dying of cancer progression. However, early
Extent of diseasePatients
the GLOT PC
(n = 73, %)
the GLOT PC
(n = 37, %)
Sites of metastases
Central nervous system
Abbreviation: NS - non-significant.
Table 6. Comparative results of response and survival.
eluded in the
GLOT PC 86/01
study (n = 73, %)
included in the
GLOT PC 86/01
study (n = 37, %)
Alive at one year
Alive at two years
Causes of death
Abbreviation: NS — non-significant.
16 24 32
Months from diagnosis
Figure 1. Survival of patients with small-cell lung cancer who were
included or not included in the protocols.
death from progressive disease was observed in 11 non-
enrolled patients (29%) but none of the included pa-
tients (P < 0.001).
Analysis of prognostic factors
Based on the worse prognosis for non-participating
patients, we next questioned whether predictive factors
for survival could be identified in excluded patients by
univariate and multivariate analysis. Based on the uni-
variate analysis performed on the overall population of
by guest on July 13, 2011
110 patients, the statistically significant predictive factors
for survival were a Karnofsky index of less than 50%
(P = 0.0007), age greater than 70 years (P = 0.013), ex-
tensive disease (P = 0.00012), the presence of bone
marrow metastasis (P = 0.0014) and exclusion from the
study (P - 0.01). However, by multivariate analysis, only
disease stage and bone marrow metastasis were signifi-
cant. Since data on the latter parameter were missing for
many of the non-included patients (data not shown), we
also performed the multivariate analysis without taking
it into account. Disease stage, Karnofsky index and age
remained statistically significant independent factors.
Interestingly, exclusion from the protocol was not an
independent prognostic factor by multivariate analysis.
Only a small fraction of patients with SCLC are in-
cluded in clinical trials, raising the concern that in-
cluded patients may not be representative of the patient
population as a whole. Moreover, little is known about
the subpopulation of patients who are not included in
clinical trials. In this study we report the clinical char-
acteristics, treatment modalities, and evolution of a
cohort of patients referred to our institution for SCLC,
but who were not included in three different consecutive
clinical trials [8-11].
All but two patients who were eligible were given the
opportunity to enter a clinical trial, and the majority
of patients offered participation enrolled in one of the
studies. Thus, the percentage of included patients among
eligible patients was very high, and the risk of selection
biases in these three trials was low . The accrual level
of patients in clinical trials is usually determined by
treatment-related factors, and by the study design .
It may also be influenced, to a lesser extent, by other
factors such as age, gender, and occupational status [3,
5, 14, 15]. The high rate of participation may reflect the
non-randomized design of all three trials. Moreover,
French law required only oral consent for participation
at the time of the first study. Full informed and written
consent has been shown to decrease patient participation,
especially when the treatment is randomly allocated [14,
16-19]. Two patients were not asked to participate be-
cause 'they belonged to the family of a physician'; this
clearly illustrates the reluctance of physicians (or their
relatives) to take part in clinical trials, as well as physi-
cians' concern of a possible negative impact of enroll-
ment in a trial on doctor-patient relationships .
However, of 178 patients with SCLC who were re-
ferred to our center during the study period, almost one-
third were ineligible for the ongoing clinical trials. Since
various exclusion criteria were noted, this population of
ineligible patients may be expected to be heterogeneous.
A majority of patients were excluded for medical reasons
such as low Karnofsky index, advanced age, or contra-
indication for the use of anthracyclines. Some biological
criteria (such as abnormal prothrombin time) were al-
ways associated with at least one other major exclusion
criteria, and were therefore irrelevant to the selection of
high-risk patients. Previous history of cancer is usually
considered a pertinent exclusion criterion because of a
theoretical risk of death due to a different neoplastic
disease than SCLC. However, since none of these can-
cers progressed, they also did not interfere with the
assessment of the SCLC treatment efficacy. This implies
that patients who are disease-free for at least five years
after treatment could be included in future studies. On
the other hand, 28% of the non-included patients had
histories of atherosclerosis, and three died of cerebral
stroke, confirming that a careful examination of cardio-
vascular tobacco-related diseases should be performed,
although this is not a recognized prognostic factor in
patients with SCLC .
Of particular interest is the high proportion of pa-
tients whose only ineligibility criterion was advanced
age. The underrepresentation of elderly patients in can-
cer treatment protocols has previously been reported
, so that the available information on the efficacy of
cancer treatments may be irrelevant for the majority of
elderly patients. The age range considered acceptable for
participation in cancer clinical trials should be extended
whenever possible, or specific protocols should be de-
signed to better assess treatment results and tolerance in
elderly patients [15, 21].
The 'non-feasibility of a long-term follow-up' was
the most common of the non-medical reasons for ineli-
gibility. Determining the ability of a given patient to
understand the objectives and schedules of the treat-
ment is difficult and very subjective. Negrier et al. 
showed in a study of 24 patient-candidates for an anti-
cancer immunotherapy trial, that although most patients
correctly understood the information given to them
prior to enrollment, two-thirds of them later requested
additional information. However, these patients were
referred by their physicians as potential participants in
clinical trials and may not reflect the affected population
as a whole. Moreover, the amount of information that
patients require and the degree to which they want to
participate in their own medical management is highly
variable from one patient to another. For instance, older
patients prefer a non-participatory patient role .
Staging procedures were not extensively performed in
patients who were not included in clinical trials. A series
of procedures is usually recommended in SCLC treat-
ment protocols; however, in the non-protocol setting,
once a site of metastatic disease has been identified,
extensive radiologic investigations may not be consid-
ered necessary [24, 25]. Moreover, we previously dem-
onstrated in the group of patients included in the GLOT
PC 86/01 protocol, that a step-by-step use of simple
clinical and biological tests can approximate the proba-
bility of metastatic SCLC, so that it may be possible to
avoid some aggressive staging procedures outside clin-
ical trials .
The prognostic factors for survival in the non-included
population were consistent with published data [6, 20,
by guest on July 13, 2011
27-32], apart from the non-significance of female gender.
Not surprisingly, patients excluded only for non-medical
reasons such as geographical contraints had better out-
comes than the other excluded patients.
Comparative analysis between patients included and
not included in clinical trials has rarely been reported
. We chose to perform this comparison only in the
patients who were ineligible for the largest of our clinical
trials (GLOT PC 86/01) [8, 9], because ineligibility
criteria differed from one trial to the other, and because
sufficient detailed data for the AR 88/01 trial were not
available. Non-enrolled patients had a significantly low-
er Karnofsky index and were older than their counter-
parts included in the trial, a finding directly related to
the definition of the ineligibility criteria. The higher
incidence of hemoptysis in included patients may be a
consequence of early detection of the cancer in patients
with hemoptysis, who thus have a better general status
and a higher likelihood of inclusion in a clinical trial.
Ineligible patients presented more frequently with meta-
static disease, although fewer extensive staging proce-
dures were performed in these patients. Consistent with
the higher incidence of metastasis, response to treatment
was poor, and the response rate of 40% may be over-
estimated because of inadequate staging reassessment.
Thus, this study demonstrates that non-participating
patients differed notably from patients included in the
The different results seen between included and ex-
cluded patients may have several explanations, such as
differences in (a) initial characteristics influencing the
prognosis, (b) diagnostic procedures performed, (c)
treatment regimens, or (d) the effect of being enrolled in
a treatment protocol. Although we cannot rule out the
possibility that the difference in outcome of non-enrolled
patients may be partially due to less optimal treatment
schedules or dosages, based on our findings it is more
likely related to their initial characteristics. Also, the
multivariate analysis clearly showed that the ineligibility
for clinical trials was not an independant predictive
factor for survival. Although our study was not designed
to assess this particular point, our results are in contrast
to those of Karjalainen and Palva , who showed in a
large non-randomized comparative study, that being
enrolled in a treatment protocol improved the survival
of patients with multiple myeloma.
Taken together, these findings demonstrate that the
clinical characteristics of the patients included in our
trials differed from those of the overall population of
patients treated for SCLC. Since the percentage of
included patients among eligible patients was high, this
difference was more likely due to restrictive eligibility
criteria than to patient or physician biases. Based on the
lack of representativeness of the subgroup of included
patients, caution should be used in applying data to the
overall population and in defining treatment guidelines.
As a simple measure to insure that the published results
are fully interpretable, we suggest that the rate and
major characteristics of ineligible and non-participating
patients be mentioned in any publication of a clinical
We are endebted to the members of the Groupe Lyon-
nais d'Oncologie Thoracique (GLOT) who participated
in the clinical studies, and to the physicians who referred
the patients. We thank Professor J. P. Boissel and A. van
Linden who reviewed the manuscript.
1. Collet JP, Boissel JP. Sick population-treated population: The
need for a better definition. The VALIDATA Group. Eur J Clin
Pharmacol 1991; 41: 267-71.
2. Amman K, Amato D, Wood W et al. Selection bias in clinical
trials. J Clin Oncol 1985; 3: 1142-7.
3. McCuster J, Wax A, Bennett JM. Cancer patient accessions into
clinical trials: A pilot investigation into some patient and physi-
cian determinants of entry. Am J Clin Oncol 1982; 5: 227-36.
4. Lee JY, Breaux SR. Accrual of radiotherapy patients to clinical
trials. Cancer 1983; 52: 1014-6.
5. Lee JY, Marks JE, Simpson JR. Age as a criterion for eligibility in
a lung cancer clinical trial. Cancer Clin Trials 1982; 5: 449-52.
6. Allan SG, Stewart ME, Love S et al. Prognosis at presentation of
small-cell carcinoma of the lung. Eur J Cancer 1990; 26: 703-5.
7. Quoix E, Finkelstein H, Wolkove N, Kreisman H. Treatment of
small-cell lung cancer on protocol: Potential bias of results. J Clin
Oncol 1986; 4: 1314-20.
8. Trillet-Lenoir V, Mornex F, Chauvin F et al. Limited disease
small-cell lung cancer: Alternating combination of doxorubicin,
etoposide, ifosfamide and hyperfractionated radiotherapy. Lung
Cancer 1993; 10: 35-45.
9. Mornex F, Trillet V, Chauvin F, et al. Hyperfractionated radio-
therapy alternating with multidrug chemotherapy in the treat-
ment of limited small-cell lung cancer (SCLC). Int J Radiat Oncol
Biol Phys 1990; 19: 23-30.
10. Trillet-Lenoir V, Green J, Manegold C et al. Recombinant gran-
ulocyte colony stimulating factor reduces the infectious compli-
cations of cytotoxic chemotherapy. Eur J Cancer 1993; 3: 319-24.
11. Jarry O, Fournel P, for the Groupe Lyonnais d'Oncologie
Thoracique. A randomized trial of four versus eight courses of
chemotherapy with ifosfamide, epirubicin and etoposide in exten-
sive small-cell lung cancer. Proceedings of the Seventh World
Conference on Lung Cancer 1996; 14 (Abstr 387).
12. Kaplan EL, Meier P. Non-parametric estimation from incom-
plete observations. J Am Stat Assoc 1958; 53: 457-81.
13. Cox DR. Regression models and life tables (with discussion). J R
Stat Soc 1972; 34: 187-220.
14. Barofsky I, Sugarbaker PH. Determinants of patient non-partic-
ipation in randomized clinical trials for the treatment of sarco-
mas. Cancer Clin Trials 1979; 2: 237-46.
15. Goodwin JS, Hunt WC, Humble CG et al. Cancer treatment
protocols. Who gets chosen? Arch Intern Med 1988; 148: 2258 60.
16. Zelen M. A new design for randomized clinical trials. N Engl J
Med 1979; 300: 1242-5.
17. Taylor KM, Margolese RG, Soskolne CL. Physicians' reasons for
not entering eligible patients in a randomized clinical trial of
surgery for breast cancer. N Engl J Med 1984; 310: 1363-7.
18. Simes RJ, Tattersall MH, Coates AS et al. Randomised compar-
ison of procedures for obtaining informed consent in clinical
trials of treatment for cancer. Br Med J 1986; 293: 1065-8.
19. Fost N. Consent as a barrier to research. N Engl J Med 1979; 300:
by guest on July 13, 2011
20. Rawson NS, Peto J. An overview of prognostic factors in small-
cell lung cancer. A report from the Subcommittee for the Man-
agement of Lung Cancer of the United Kingdom Coordinating
Committee on Cancer Research. Br J Cancer 1990; 61: 597-604.
21. Fentiman IS, Tirelli U, Monfardini S, et al. Cancer in the elderly:
Why so badly treated? Lancet 1990: 335: 1020-2.
22. Negrier S, Lanier-Demma F, Lacroix-Kante Vet al. Evaluation of
the informed consent procedure in cancer patients candidate to
immunotherapy. Eur J Cancer 1995; 31A: 1650-2.
23. Cassileth BR, Zupkis RV, Sutton-Smith K, March V. Information
and participation preferences among cancer patients. Ann Intern
Med 1980; 92: 832-6.
24. Johnson BE. Management of small-cell lung cancer. Clin Chest
Med 1993; 14: 173-87.
25. Darling GE. Staging of the patient with small-cell lung cancer.
Chest Surg Clin N Am 1997; 7: 81-94.
26. Chauvin F, Trillet V, Court-Fortune I et al. Pre-treatment staging
evaluation in small-cell lung carcinoma. A new approach to
medical decision making. Chest 1992; 102: 497-502.
27. Osterlind K, Andersen PK. Prognostic factors in small cell lung
cancer: Multivariate model based on 778 patients treated with
chemotherapy with or without irradiation. Cancer Res 1986; 46:
28. Albain KS, Crowley JJ, LeBlanc M, Livingston RB. Determi-
nants of improved outcome in small-cell lung cancer: An analysis
of the 2,580-patient Southwest Oncology Group data base. J Clin
Oncol 1990; 8: 1563-74.
29. Ihde DC, Makuch RW, Carney DN et al. Prognostic implications
of stage of disease and sites of metastases in patients with small-
cell carcinoma of the lung treated with intensive combination
chemotherapy. Am Rev Respir Dis 1981; 123: 500-7.
30. Gronowitz JS, Bergstrom R, Nou E et al. Clinical and serologic
markers of stage and prognosis in small cell lung cancer. A
multivariate analysis. Cancer 1990; 66: 722-32.
31. Spiegelman D, Maurer LH, Ware JH et al. Prognostic factors in
small-cell carcinoma of the lung: An analysis of 1,521 patients. J
Clin Oncol 1989; 7: 344-54.
32. Sagman U, Maki E, Evans WK, et al. Small-cell carcinoma of the
lung: Derivation of a prognostic staging system. J Clin Oncol
1991; 9: 1639^9.
33. Karjalainen S, Palva I. Do treatment protocols improve end
results? A study of survival of patients with multiple myeloma in
Finland. BMJ 1989; 299: 1069-72.
Received 8 March 1999; accepted 14 May 1999.
Prof. V. Trillet-Lenoir, MD
Unite d'Oncologie Medicale
Centre Hospitalier Lyon-Sud
by guest on July 13, 2011
by guest on July 13, 2011