Outcomes among African-American/non-African-American patients with advanced non-small-cell lung carcinoma: report from the Cancer and Leukemia Group B.

Abstract

Among patients diagnosed with advanced non-small-cell lung carcinoma (NSCLC), African-Americans have lower survival rates than non-African-Americans. Whether this difference is due to innate characteristics of the disease in the two ethnicities or to disparities in health care is not known. We investigated whether the disparity in survival would persist when patients were treated with similar systemic therapies (i.e., in phase II and phase III Cancer and Leukemia Group B [CALGB] trials).
We assessed 504 consecutive patients (458 non-African-American and 46 African-American) receiving systemic chemotherapy in CALGB studies for advanced NSCLC during the period from 1989 through 1998. Clinical and demographic characteristics, treatment received, and survival data were obtained from the CALGB database. Cox's proportional hazards model was used to assess the effect of race/ethnicity on survival after adjustment for other known prognostic factors. All statistical tests were two-sided.
The unadjusted 1-year survival rate was 22% (95% confidence interval [CI] = 13% to 38%) for African-American patients and 30% (95% CI = 26% to 35%) for non-African-American patients, a statistically significant difference (8%; 95% CI on the difference = 5% to 12%; P =.03). Multivariable adjustment for the effect of treatment arm, histology, and metastatic site at presentation did not alter the worse outcome for African-American patients. However, the effect of race/ethnicity disappeared after adjustment for performance status and weight loss. African-American patients were more likely than non-African-Americans to present with a poor performance status (83% versus 60%) and substantial weight loss (41% versus 27%) and to be unmarried (59% versus 28%), disabled (31% versus 15%), unemployed (17% versus 7%), and Medicaid recipients (30% versus 8%).
The relationship that we observed between poor performance, weight loss, and socioeconomic status suggests that social circumstances lead to African-Americans presenting with poorer prognostic features.

Full text

Page 1

Outcomes Among African-American/Non-African-American
Patients With Advanced Non-Small-Cell Lung Carcinoma:
Report From the Cancer and Leukemia Group B
A. William Blackstock, James E. Herndon II, Electra D. Paskett, Michael C. Perry,
Stephen L. Graziano, Joseph J. Muscato, Michael P. Kosty, Wallace L. Akerley,
Jimmie Holland, Stewart Fleishman, Mark R. Green
Background: Among patients diagnosed with advanced non-
small-cell lung carcinoma (NSCLC), African-Americans
have lower survival rates than non-African-Americans.
Whether this difference is due to innate characteristics of the
disease in the two ethnicities or to disparities in health care
is not known. We investigated whether the disparity in sur-
vival would persist when patients were treated with similar
systemic therapies (i.e., in phase II and phase III Cancer and
Leukemia Group B [CALGB] trials). Methods: We assessed
504 consecutive patients (458 non-African-American and 46
African-American) receiving systemic chemotherapy in
CALGB studies for advanced NSCLC during the period
from 1989 through 1998. Clinical and demographic charac-
teristics, treatment received, and survival data were ob-
tained from the CALGB database. Cox’s proportional haz-
ards model was used to assess the effect of race/ethnicity on
survival after adjustment for other known prognostic fac-
tors. All statistical tests were two-sided. Results: The unad-
justed 1-year survival rate was 22% (95% confidence inter-
val [CI] = 13% to 38%) for African-American patients and
30% (95% CI = 26% to 35%) for non-African-American
patients, a statistically significant difference (8%; 95% CI on
the difference = 5% to 12%; P = .03). Multivariable adjust-
ment for the effect of treatment arm, histology, and meta-
static site at presentation did not alter the worse outcome for
African-American patients. However, the effect of race/
ethnicity disappeared after adjustment for performance sta-
tus and weight loss. African-American patients were more
likely than non-African-Americans to present with a poor
performance status (83% versus 60%) and substantial
weight loss (41% versus 27%) and to be unmarried (59%
versus 28%), disabled (31% versus 15%), unemployed (17%
versus 7%), and Medicaid recipients (30% versus 8%). Con-
clusions: The relationship that we observed between poor
performance, weight loss, and socioeconomic status suggests
that social circumstances lead to African-Americans pre-
senting with poorer prognostic features. [J Natl Cancer Inst
2002;94:284–90]
Lung cancer is the leading cause of cancer-related death in
the United States and North America (1,2). Non-small-cell lung
cancer (NSCLC) is the most common histologic cell type and
often presents in an advanced stage. The prognosis for metastatic
NSCLC is extremely poor, but several trials have shown that
systemic chemotherapy can palliate cancer-related symptoms
and modestly improve survival for patients with this disease
(3–6). Although the incidence of advanced-stage disease has
increased in most race/ethnic groups, the rate of increase has
been greatest for African-American patients (7). Moreover, rela-
tive 2- and 5-year survival rates for patients with local- and/or
regional-stage disease are lower in African-Americans than in
non-African-Americans, and these differences in survival rates
have become more prominent in recent years (8). National health
surveillance statistics and other studies (9–13) examined race/
ethnicity in relation to NSCLC prognosis and found that, during
the past several decades, non-African-Americans have more fa-
vorable survival rates than African-Americans.
The prognosis for patients with advanced NSCLC is related
to several factors, including pretreatment stage, sex, weight loss,
and overall performance status (14–19). In addition, it has been
proposed that NSCLC in African-American patients is a biologi-
cally more aggressive disease because of a number of genetic
factors (20). Race/ethnicity, in and of itself, however, may not be
important in determining the prognosis for patients diagnosed
with lung cancer. It is possible that the poor outcome observed
for African-American patients may be reduced or eliminated by
adjusting for comorbid conditions and socioeconomic factors
associated with race/ethnicity.
In this study, we examined survival after treatment of ad-
vanced NSCLC among African-American and non-African-
American patients participating in phase II or phase III Cancer
and Leukemia Group B (CALGB) clinical trials. The coopera-
tive group trial design defines the extent of disease at the time of
study entry and provides for uniform therapy, thereby control-
ling for two major confounding factors affecting prognosis and
allowing investigation of other determinants of prognosis. The
mandate of the National Institutes of Health (NIH) Revitaliza-
tion Act of 1993 required that all NIH-sponsored clinical trials
include minorities and women in sufficient numbers to allow
Affiliations of authors: A. W. Blackstock, Wake Forest University School of
Medicine, Winston-Salem, NC, and University of North Carolina at Chapel Hill,
Chapel Hill, NC; J. E. Herndon II, Cancer and Leukemia Group B Statistical
Center, Duke University, Durham, NC; E. D. Paskett, Wake Forest University
School of Medicine; M. C. Perry, University of Missouri/Ellis Fischel Cancer
Center, Columbia; S. L. Graziano, SUNY Upstate Medical University, Syracuse,
NY; J. J. Muscato, Missouri Cancer Associates, Columbia; M. P. Kosty, Scripps
Clinic, La Jolla, CA; W. L. Akerley, Women and Infants Hospital of Rhode
Island, Providence; J. Holland, Memorial Sloan-Kettering Cancer Center, New
York, NY; S. Fleishman, Institute of Oncology, Long Island Jewish Medical
Center, New Hyde Park, NY; M. R. Green, Medical University of South Caro-
lina, Charleston.
Correspondence to: A. William Blackstock, M.D., Department of Radiation
Oncology, Wake Forest University School of Medicine, Medical Center Blvd.,
Winston-Salem, NC 27157 (e-mail: ablackst@wfubmc.edu).
See “Appendix” section for the institutions that participated in the study,
principal investigators, and grant numbers.
See “Notes” following “References.”
© Oxford University Press
284 ARTICLESJournal of the National Cancer Institute, Vol. 94, No. 4, February 20, 2002
by guest on July 13, 2011
jnci.oxfordjournals.org
Downloaded from

Page 2

valid subset analysis to ascertain differences in the effect of
treatments among women and minority participants (21). Our
analysis in this study is an attempt to fulfill the intent of the
mandate in a rational and ethically sound manner.
PATIENTS AND METHODS
Patient Selection
During the period from 1989 through 1998, the CALGB con-
ducted five consecutive phase II/III clinical trials evaluating sys-
temic chemotherapy as an initial treatment for 529 patients with
advanced NSCLC. In this study, we retrospectively evaluated
the database from these five trials. A majority of patients out-
lined in these CALGB trials completed an extensive demo-
graphic questionnaire, which provided information on level of
education, total yearly income, medical coverage, and marital/
employment status.
The five CALGB protocols were protocols 8931, 8932, 9132,
9532, and 9731. During the period from July 1989 through Feb-
ruary 1991, 291 patients with American Joint Committee on
Cancer stage IIIB or IV NSCLC were accrued to protocol 8931,
a randomized placebo-controlled phase III trial that assessed the
impact of adding hydrazine sulfate to the standard chemotherapy
regimen of cisplatin and vinblastine (22). During the period from
February 1989 through June 1990, 40 patients were accrued to
protocol 8932, a phase II study that evaluated the efficacy of
cisplatin and intravenous etoposide for the treatment of meta-
static or recurrent NSCLC (23). During the period from July
1991 through March 1992, 101 patients with stage IIIB or IV
NSCLC were accrued to protocol 9132, a randomized phase II
study that determined the efficacy of ifosfamide, mesna, cisplat-
in, and granulocyte colony-stimulating factor (G-CSF) or eto-
poside, cisplatin, and G-CSF (24). During the period from July
1995 through July 1996, 83 patients were accrued to protocol
9532, a randomized phase II study that evaluated the efficacy of
paclitaxel and ifosfamide versus vinorelbine and ifosfamide for
patients with stage IV or recurrent NSCLC (25). During the
period from July 1997 through April 1998, 39 patients were
accrued to CALGB protocol 9731, a phase II study that evalu-
ated the efficacy and toxicity of weekly paclitaxel for patients
with chemotherapy-naive, stage IIIB or stage IV NSCLC (26).
All patients had an Eastern Cooperative Oncology Group
performance status of 0 or 1 and adequate renal and hepatic
functions. No patient received systemic chemotherapy before
entry into the study. Additional details of the design of the
studies and primary findings can be found elsewhere (22–26).
For inclusion in this analysis, we selected all protocol-eligible
patients with follow-up information whose race was reported
(442 Caucasian, five Hispanic, three Asian, one Native Ameri-
can, one Asian Indian, two Filipino, 46 African-American, and
four not otherwise specified). Patients who were not considered
eligible in the original analyses (22–26) and registered patients
who withdrew from the study before receiving the protocol treat-
ment are not included in this study. The current analysis did not
differentiate between the two treatment arms in CALGB proto-
col 8931 because the primary analyses of that study showed
almost superimposable survival curves (22). Although 529 eli-
gible patients are included in this article, data were missing for
25 patients. Thus, the multivariate survival analyses use data
from 504 patients, of which 458 were non-African-American
and 46 were African-American patients.
To address the impact of potential socioeconomic factors on
treatment outcome, an additional analysis was performed on
patients from CALGB protocols 8931, 9532, and 9731. Infor-
mation was available from 300 non-African-Americans and 30
African-American patients regarding marital status, employ-
ment, educational level, insurance, and income for patients
treated on these three trials. Information was not available for
patients treated on CALGB protocols 8932 and 9132.
Study Endpoints and Statistical Methods
Frequency distributions of selected clinical variables (Table
1) were compared between African-Americans and non-African-
Americans by chi-square and Fisher’s exact tests. The Cox pro-
portional hazards model was used to compute relative risk ratios
for African-Americans versus non-African-Americans and to si-
multaneously assess and control for the confounding influence
of other prognostic covariables (27,28). P values from these
analyses are two-sided. After adjustment for known prognostic
factors and other potential confounders, a variable describing race/
ethnicity (African-American versus non-African-American) was
added to the model to determine whether its inclusion strength-
ened the model. For a subset of patients with socioeconomic
data, chi-square tests were used to determine the relationship
between socioeconomic measurements and race/ethnicity. Sur-
vival times were estimated with the use of Kaplan–Meier curves
and were compared by a two-sided log-rank test using the SAS
6.09 and S-plus statistical package (SAS Institute, Inc., Cary,
NC) (29,30).
RESULTS
Among the 504 patients included in this analysis, 46 were
African-American. Table 1 reflects the distribution of character-
istics by race/ethnicity for all patients. The proportion of non-
African-Americans who were 60 years of age or older was
statistically significantly greater than that of African-Americans
(P ? .04). African-Americans were more often male (76%) than
non-African-Americans (69%).
We determined the survival estimates for all patients across
all protocols by the Kaplan–Meier method (29,30). Although the
relative proportion of non-African-Americans and African-
Americans differed by trial, the relative prognosis for the two
race/ethnic groups did not differ. Without adjustment for other
prognostic covariables, the survival pattern for non-African-
Americans and African-Americans was statistically significantly
different (P ? .03). The 1-year survival rate was 30% (95% CI
? 26% to 35%) for non-African-Americans and 22% (95% CI
? 13% to 38%) for African-Americans (Fig. 1). The median
survival for non-African-Americans and African-Americans was
7.9 and 7.7 months, respectively.
We next investigated factors in addition to race/ethnicity that
may be associated with survival (Table 2). Patients with a per-
formance status of 1 (restricted in physically strenuous activity
but ambulatory and able to carry out work of a light or sedentary
nature, e.g., light housework or office work) were at a 40%
higher risk of dying (risk ratio [RR] ? 1.40; 95% CI ? 1.15 to
1.71) than patients with a performance status of 0 (P<.001).
Patients with squamous cell lung cancer had a statistically sig-
nificantly shorter life span than patients with other histologies,
and patients with a pretreatment weight loss of greater than or
equal to 5% had a statistically significantly higher risk of dying
Journal of the National Cancer Institute, Vol. 94, No. 4, February 20, 2002ARTICLES 285
by guest on July 13, 2011
jnci.oxfordjournals.org
Downloaded from

Page 3

Fig. 1. Survival estimates
among patients from all ad-
vanced non-small-cell lung
cancer (NSCLC) trials com-
bined (non-African-Americans
? 458; African-Americans
? 46). The number of deaths
is noted through the entire pe-
riod of follow-up in all trials.
Among non-African-Ameri-
cans, at 6 months, there were
281 patients at risk (95%
confidence interval [CI] ?
0.57 to 0.66); at 18 months,
there were 75 patients at risk
(95% CI ? 0.14 to 0.21).
Among African-Americans,
at 6 months, there were 27 pa-
tients at risk (95% CI ? 0.46
to 0.75); at 18 months, there
were two patients at risk
(95% CI ? 0.01 to 0.17).
Table 1. Clinical variables for all patients enrolled in Cancer and Leukemia Group B trials for treatment of advanced non-small-cell lung cancer
Treatment armNo. of patients
No. of non-African-Americans,
n ? 458 (%)
No. of African-Americans,
n ? 46 (%)
P
Protocol
8931
8932
9132
9532
9731
Performance status*
ECOG 0
ECOG 1
Histology
Adenocarcinoma
Squamous
Other
AJCC stage of disease†
IIIB
IV
Recurrent
Age, y
<60
?60
Sex
Male
Female
Weight loss, total
None
?5%
254
36
93
86
35
223 (49)
32 (7)
91 (20)
80 (17)
32 (7)
31 (67)
4 (9)
2 (4)
6 (13)
3 (7)
.032
191
313
183 (40)
275 (60)
8 (17)
38 (83)
.002
250
120
134
231 (50)
106 (23)
121 (27)
19 (41)
14 (31)
13 (28)
.439
73
311
126
62 (13)
282 (62)
114 (25)
5 (11)
29 (63)
12 (26)
.910
208
296
182 (40)
276 (60)
26 (57)
20 (43)
.040
349
155
314 (69)
144 (31)
35 (76)
11 (24)
.320
360
144
333 (73)
125 (27)
27 (59)
19 (41)
.059
*ECOG ? Eastern Cooperative Oncology Group. ECOG 0 ? fully active, able to carry on all predisease performance without restriction. ECOG 1 ? restricted
in physically strenuous activity but ambulatory and able to carry out work of a light or sedentary nature, e.g., light housework or office work.
†AJCC ? American Joint Committee on Cancer.
286 ARTICLESJournal of the National Cancer Institute, Vol. 94, No. 4, February 20, 2002
by guest on July 13, 2011
jnci.oxfordjournals.org
Downloaded from

Page 4

than patients without weight loss (P ? .024). Race was incor-
porated into the clinical multivariate model presented in Table 2.
After adjustment for known clinical factors (Table 2), the effect
of race on survival was not statistically significant. That is, race
did not have an independent prognostic effect on survival. Only
in a multivariable model that did not adjust for weight loss and
poor performance status did the effect of race become statisti-
cally significant, with a relative risk of 1.45 (95% CI ? 1.05
to 1.95).
Race/ethnicity was then incorporated into the clinical multi-
variate model. After adjustment for known clinical factors, race/
ethnicity did not statistically significantly affect the survival of
patients with NSCLC (Table 2). In fact, race/ethnicity did not
have an independent prognostic effect on survival (RR ? 1.14;
95% CI ? 0.82 to 1.5). Only in a multivariable model that did
not adjust for weight loss and poor performance status did the
effect of race/ethnicity become statistically significant (RR ?
1.45; 95% CI ? 1.05 to 1.95).
A statistically significant relationship was observed between
race/ethnicity and several clinical factors included in the multi-
variate model. Although not statistically significant (P ? .06),
the percentage of patients presenting with weight loss of greater
than or equal to 5% was greater among African-American pa-
tients than among non-African-American patients (RR ? 1.28;
95% CI ? 1.03 to 1.59). The percentage of patients with a
performance status of 1 was also statistically significantly
greater among the African-American patients (83%) than among
the non-African-American patients (60%) (P ? .002).
Table 3 summarizes the relationship between race/ethnicity
and socioeconomic measurements within the subgroup of pa-
tients for whom the information was available. A comparison of
marital status found that 58% of African-Americans were un-
married versus 27% of non-African-American patients. More
African-American patients were disabled or unemployed (31%
and 17%, respectively) than non-African-American patients
(15% and 7%, respectively). African-American patients were
more likely than non-African-American patients to receive Med-
icaid (P ? .001) and less likely to have private group health
insurance (P ? .037) and had a substantially lower median
income when compared with non-African-American patients
(Table 3).
DISCUSSION
NSCLC remains the leading cause of cancer-related death in
the United States. Across all stages of the disease, survival for
African-American patients continues to lag behind that reported
for non-African-American patients (31–33). A number of poten-
tial explanations for this observation have been proposed, in-
cluding later stage at diagnosis, differences in treatment, and
Table 2. Clinical predictors of survival among patients enrolled in Cancer and
Leukemia Group B trials for treatment of advanced non-small-cell
lung cancer*
Variable
Risk ratio
(95% CI)†
Risk ratio
(95% CI)‡
Protocol
8932 vs. 8931
9132/arm No. 1 vs. 8931
9132/arm No. 2 vs. 8931
9532/arm No. 1 vs. 8931
9532/arm No. 2 vs. 8931
9731 vs. 8931
Performance status (PS)§: ECOG PS
? 1 vs. ECOG PS ? 0
Histology
Squamous cell carcinoma vs.
adenocarcinoma
Other vs. adenocarcinoma
Metastatic site
Adrenal
Liver
Bone
Chest pain
Weight loss, ?5%
African-American vs. non-
African-American
0.97 (0.69 to 1.38)
0.92 (0.66 to 1.27)
1.04 (0.75 to 1.44)
0.63 (0.45 to 0.88)
0.59 (0.41 to 0.85)
0.66 (0.45 to 0.99)
1.40 (1.15 to 1.71)
0.98 (0.69 to 1.39)
0.93 (0.67 to 1.29)
1.06 (0.76 to 1.48)
0.64 (0.45 to 0.90)
0.60 (0.41 to 0.86)
1.39 (1.14 to 1.69)
1.39 (1.14 to 1.69)
1.30 (1.03 to 1.64)1.31 (1.04 to 1.65)
1.04 (0.83 to 1.30)1.04 (0.83 to 1.30)
1.45 (1.12 to 1.87)
1.40 (1.07 to 1.82)
1.34 (1.10 to 1.64)
1.20 (0.98 to 1.46)
1.30 (1.05 to 1.60)
1.45 (1.05 to 1.95)
1.44 (1.11 to 1.86)
1.41 (1.09 to 1.84)
1.35 (1.11 to 1.65)
1.18 (0.96 to 1.44)
1.28 (1.03 to 1.59)
1.14 (0.82 to 1.58)
*CI ? confidence interval.
†Multivariate analysis without adjustment for race/ethnicity.
‡Multivariate analysis with adjustment for race/ethnicity.
§ECOG ? Eastern Cooperative Oncology Group. ECOG 0 ? fully active,
able to carry on all predisease performance without restriction. ECOG 1 ?
restricted in physically strenuous activity but ambulatory and able to carry out
work of a light or sedentary nature, e.g., light housework or office work.
Table 3. Relationship between socioeconomic characteristics and
race/ethnicity of patients enrolled in Cancer and Leukemia Group B trials for
treatment of advanced non-small-cell lung cancer
Characteristic
Non-African-
American
African-
American
P*
Total No. of patients
Marital status
Single, never married
Married
Separated
Divorced
Widowed
Missing
Employment status
Full-time
Part-time
Homemaker
Disabled
Unemployed
Retired
Other
Missing
Insurance‡
Medicaid
Medicare
Disability insurance
Health maintenance organization
Individual health
Group health
National insurance
Department of Veterans Affairs/Military
No insurance
Income, U.S. dollars
<9999
10000–19999
20000–29999
30000–49999
50000–79000
?80000
Refused to answer
Missing
300 (%)†30 (%)†
.001
20 (7)
220 (73)
5 (2)
33 (11)
21 (7)
1 (<1)
5 (17)
12 (41)
5 (17)
1 (3)
6 (21)
1 (1)
.027
58 (19)
19 (6)
17 (6)
44 (15)
21 (7)
134 (45)
4 (1)
3 (<1)
5 (17)
0
0
9 (31)
5 (17)
10 (34)
0
1 (1)
25 (8)
110 (37)
193 (65)
21 (7)
56 (19)
97 (32)
6 (2)
12 (4)
20 (7)
9 (30)
11 (37)
22 (76)
2 (7)
5 (17)
4 (13)
0
0
3 (10)
.001
>.999
.348
.945
.788
.037
>.999
.611
.452
20 (11)
43 (23)
31 (16)
32 (17)
11 (6)
6 (3)
46 (24)
4 (13)
2 (7)
4 (13)
0
0
0
4 (13)
16 (53)
.106
*P values were derived from a Chi-square test.
†Percentages are calculated on the basis of available data.
‡Some patients were covered by multiple insurance policies.
.........................
Journal of the National Cancer Institute, Vol. 94, No. 4, February 20, 2002ARTICLES 287
by guest on July 13, 2011
jnci.oxfordjournals.org
Downloaded from

Page 5

differences in the biologic aggressiveness of the disease. Al-
though these factors may be important, perhaps the most rel-
evant factor for African-American patients may be access to
routine cancer care. Indeed, on the basis of data from the Black/
White Cancer Survival Study Group (34), it has been hypoth-
esized that less access to routine cancer care results in a worse
patient condition (i.e., greater weight loss and poorer perfor-
mance status) at diagnosis. Using the CALGB clinical trial da-
tabase, we investigated this hypothesis. The CALGB lung can-
cer database offers several advantages over other data sources.
First, stage of disease is comparable because of the constraints of
the protocol entry criteria. Second, treatment and follow-up care
are conducted according to a uniform standard. Third, detailed
information on prognostic covariates, such as performance status
and weight loss, is collected. However, one ongoing limitation
of the CALGB data and data obtained by other national coop-
erative groups is the small number of participants from many
racial groups, including African-Americans.
In this cohort of 504 patients treated with chemotherapy for
advanced NSCLC, the 46 African-Americans and 458 non-
African-Americans had similar clinical stages of disease at the
time of protocol entry. However, the percentage of patients with
a performance status of 1 and/or with greater than or equal to 5%
weight loss was higher among African-Americans than among
non-African-Americans (83% African-Americans versus 60%
non-African-Americans with a performance status of 1; 41%
African-Americans versus 27% non-African-Americans with
?5% weight loss). These parameters, which may be perhaps
indicators of delayed access to cancer care, are known predictors
for poor outcome in patients with NSCLC. The 1-year survival
rate observed in the African-American patient cohort was 22%,
which was marginally better than that achieved for patients re-
ceiving the best supportive care in other trials in advanced
NSCLC (35). In our analysis, the 1-year survival rate for non-
African-American patients was 30%, a statistically significantly
higher survival rate than that for African-Americans. Being
African-American did not, however, predict for worse survival
above and beyond the clinical factors of diminished performance
status or pretreatment weight loss. The 1-year survival rate for
patients with a performance status of 1 across both African-
American and non-African-American cohorts was comparable,
supporting our postulate that equal treatment of racially diverse
patients with equivalent pretreatment prognostic features results
in equal outcome.
Perhaps the most important observation in our analysis is the
relationship among poor performance status, low socioeconomic
status, and the overall outcome observed in African-American
patients. Table 3 shows that African-American patients were
more likely to be unmarried, to be uninsured or rely upon Med-
icaid, and, in general, to have a lower income. All of these
factors have been shown to predict a poor clinical outcome
irrespective of race/ethnicity in a variety of diseases (36–39). In
our study population, African-American patients were also more
likely at study entry to have a lower performance status, as a
result of either their malignancy or other pre-existing comor-
bidities. Cancer patients with even moderately impaired ambu-
lation or pain at diagnosis have reduced survival and greatly
increased toxicity from chemotherapy when compared with pa-
tients who have a better performance status (40–43). It is then
reasonable to conclude that the survival of African-American
patients with advanced NSCLC is not related to race/ethnicity or
to inherent differences in the biology of their cancers; it is
related to the clinical features at presentation that reflect an
increase in the overall tumor burden (i.e., performance status and
weight loss) at the time treatment is initiated.
From our analysis, we suggest that for low-income patients,
social circumstances lead to a lack of resources and knowledge,
poor living conditions, and a lack of access to care services that
influence treatment efficacy or toxicity. Because equal treat-
ment yields equal outcome among patients with the same
stage of disease regardless of race/ethnicity, continued efforts
are needed to encourage disease awareness, promote early de-
tection, and increase minority participation in clinical trials. Ul-
timately, potential interventions need to be introduced that will
decrease the mortality for all patients diagnosed with advanced
NSCLC.
APPENDIX
Appendix table 1. Institutions that participated in the study, principal investigator, and grant number*
InstitutionPrincipal investigatorSupport grant No.†
CALGB Statistical Office, Durham, NC
Baptist Cancer Institute CCOP, Memphis, TN
Christiana Care Health Services, Inc. CCOP, Wilmington, DE
Dana-Farber Cancer Institute, Boston, MA
Dartmouth Medical School–Norris Cotton Cancer Center, Lebanon, NH
Duke University Medical Center, Durham, NC
Eastern Maine Medical Center, Bangor, ME
Long Island Jewish Medical Center, Lake Success, NY
Massey Cancer Center, Richmond, VA
Medical University of South Carolina, Charleston, SC
Memorial Sloan-Kettering Cancer Center, New York, NY
Mount Sinai Medical Center, Miami, FL
Mount Sinai School of Medicine, New York, NY
North Shore University Hospital, Manhasset, NY
Rhode Island Hospital, Providence, RI
Roswell Park Cancer Institute, Buffalo, NY
Southeast Cancer Control Consortium, Inc., Goldsboro, NC
Southern Nevada Cancer Research Foundation, Las Vegas, NV
SUNY Upstate Medical University, Syracuse, NY
Syracuse Hematology–Oncology CCOP, Syracuse, NY
Stephen George, Ph.D.
Lee S. Schwartzberg, M.D.
Irving M. Berkowitz, D.O.
George P. Canellos, M.D.
Marc Ernstoff, M.D.
Jeffrey Crawford, M.D.
Philip L. Brooks, M.D.
Marc Citron, M.D.
John D. Roberts, M.D.
Mark Green, M.D.
George Bosl, M.D.
Enrique Davila, M.D.
Lewis Silverman, M.D.
Daniel R. Budman, M.D.
Louis A. Leone, M.D.
Ellis Levine, M.D.
James N. Atkins, M.D.
John Ellerton, M.D.
Stephen L. Graziano, M.D.
Jeffrey Kirshner, M.D.
CA33601
CA71323
CA45418
CA32291
CA04326
CA47577
CA35406
CA11028
CA52784
CA03927
CA77651
CA45564
CA04457
CA35279
CA08025
CA02599
CA45808
CA35421
CA21060
CA45389
(Appendix continues)
288 ARTICLESJournal of the National Cancer Institute, Vol. 94, No. 4, February 20, 2002
by guest on July 13, 2011
jnci.oxfordjournals.org
Downloaded from