Content uploaded by Fulvio Lonardo
Author content
All content in this area was uploaded by Fulvio Lonardo
Content may be subject to copyright.
Carcinogenesis vol.29 no.9 pp.1781–1787, 2008
doi:10.1093/carcin/bgn107
Advance Access publication May 2, 2008
COX-2/EGFR expression and survival among women with adenocarcinoma of the lung
Alison L.Van Dyke
, Michele L.Cote
1,2
, Geoffrey
M.Prysak
1
, Gina B.Claeys
1
, Angie S.Wenzlaff
1
, Valerie
C.Murphy
1
, Fulvio Lonardo
3,4
and Ann G.Schwartz
1,2
Cancer Biology Program,
1
Population Studies and Prevention Program,
2
Department of Internal Medicine,
3
Department of Pathology and
4
Developmental Therapeutics Program, Karmanos Cancer Institute, Wayne
State University School of Medicine, Detroit, MI 48201, USA
To whom correspondence should be addressed. Tel: þ1 313 578 4314;
Fax: þ1 313 578 4306;
Email: avandyk@med.wayne.edu
Previous studies suggest that cyclooxygenase-2 (COX-2) expres-
sion may predict survival among patients with non-small cell lung
cancer. COX-2 may interact with epidermal growth factor recep-
tor (EGFR), suggesting that combined COX-2/EGFR expression
may provide predictive value. The extent to which their indepen-
dent or combined expression is associated with prognosis in
women with adenocarcinoma of the lung is unknown. In the pres-
ent study, we examined relationships between COX-2 expression
(n 5 238), EGFR expression (n 5 158) and dual COX-2/EGFR
expression (n 5 157) and survival among women with adenocar-
cinoma of the lung. Overall survival was estimated by construct-
ing Cox proportional hazards models adjusting for other
significant variables and stratifying by stage at diagnosis and race.
Clinical or demographic parameters were not associated with
either COX-2 or EGFR expression. Patients with COX-2-positive
tumors tended to have poorer prognosis than did patients with
COX-2-negative tumors [hazard ratio (HR) 1.67, 95% confidence
interval (CI) 1.01–2.78]. African-Americans with COX-2-positive
tumors had a statistically non-significant higher risk of death than
African-Americans with COX-2-negative tumors (HR 5.58, 95%
CI 0.64–48.37). No association between COX-2 expression and
survival was observed among Caucasians (HR 1.29, 95%
CI 0.72–2.30). EGFR expression was associated with a 44%
reduction in the risk of death (HR 0.56, 95% CI 0.32–0.98).
COX-22/EGFR1 tumor expression, but not COX-21/EGFR1
tumor expression, was associated with survival when compared
with other combined expression results. In conclusion, COX-2
and EGFR expression, but not combined COX-21/EGFR1
expression, independently predict survival of women with adeno-
carcinoma of the lung.
Introduction
Lung cancer is the second leading type of cancer in both men and
women in USA and the leading cause of cancer-related mortality. The
low overall 5 year survival rate of 15% underscores the need to de-
velop new treatment options (1). Identification of molecular pathways
involved in lung carcinogenesis may lead to novel-targeted therapies.
Two such pathways involve cyclooxygenase-2 (COX-2) and epider-
mal growth factor receptor (EGFR); both proteins are expressed in
non-small cell lung cancer (NSCLC) and may predict prognosis.
These proteins may affect lung carcinogenesis both individually
and, as recently suggested, synergistically (2). Additionally, they have
been targeted simultaneously for treatment of lung cancer (3–8).
However, the prognostic significance of the individual and combined
expression of these proteins with the patient’s clinical, epidemiolog-
ical and demographic characteristics is poorly characterized. NSCLC
in women has distinct features including a higher prevalence of ade-
nocarcinoma, a lower frequency of current smokers and distinct mu-
tation profile, making women with NSCLC an important subgroup to
study (9,10).
COX-2 is one of the two COX enzymes that catalyze the conversion
of arachidonic acid into prostanoids including prostaglandin E
2
,
which has been associated with angiogenesis, cell motility and in-
vasion, immunosuppression and inhibition of programmed cell death,
all processes that are vital to tumor progression (11–13). Between
54 and 100% of NSCLC tumors are COX-2 positive as measured
by immunohistochemistry (IHC), reverse transcription–polymerase
chain reaction and in situ hybridization (14–16). Yuan et al. (14)
found that high COX-2 messenger RNA (mRNA) expression is cor-
related with high COX-2 protein expression. Studies examining the
relationship between COX-2 tumor expression and survival among
lung cancer patients have been inconsistent, with reports of an inverse
relationship with survival (14,17–19), no association (20), or a direct
association with survival (21). COX-2 expression has been most con-
sistently associated with poorer survival among stage I and II NSCLC
patients (17–19,22). This result was replicated in a meta-analysis of
14 studies (23). Tsubochi et al. (18) found that COX-2 tumor expres-
sion predicted survival in men, patients ,65 years of age and patients
with adenocarcinomas. High COX-2 mRNA expression is associated
with a shorter relapse time and poorer 5 year survival even after taking
histology into account (17). These studies suggest a role of COX-2 in
tumor progression.
A member of the ErbB family of cell surface tyrosine kinase re-
ceptors, EGFR binds to growth factor ligands, triggering intracellular
signal transduction cascades associated with the activation of genes
involved in antiapoptotic mechanisms, cell cycle progression, angio-
genesis, gene transcription and increased cellular proliferation (24).
EGFR has been found to be overexpressed at the protein level in
NSCLC tissue at all tumor stages in comparison with uninvolved lung
tissue, and closely associated with EGFR protein expression in
NSCLC tumors are EGFR gene copy number and EGFR mRNA
transcript level (25,26). EGFR expression intensity has been shown
to increase with stage at diagnosis, and EGFR overexpression has
been found more frequently in stages II and III disease than in stage
I disease (27,28). Moreover, lymph node metastasis has been associ-
ated with higher EGFR protein expression as measured by competi-
tive ligand binding assay and IHC (28). EGFR protein expression has
also been associated with well-differentiated tumors (26,29). Conflict-
ing studies, however, report that EGFR expression is not related to
stage, grade, tumor size, sex, age, smoking status or other clinicopath-
ologic parameters (26,27,30–32). Estimates of the prevalence of
EGFR-positive lung tumors range from 13 to 80% (33). While EGFR
overexpression appears to be more intense and more prevalent among
squamous cell carcinomas (27), a meta-analaysis reported that an
average of 46.2% of adenocarcinomas were EGFR positive (33).
While this meta-analysis of the relationship between EGFR expres-
sion and survival among lung cancer patients found a 13% increased
risk of death [hazard ratio (HR) 1.13, 95% confidence interval (CI)
1.00–1.28], it did not examine this relationship by stage (33). The
question remains whether the relationship between EGFR tumor ex-
pression and survival varies by stage at diagnosis.
EGFR inhibitors have been approved to treat lung cancer since
2004. In recent years, researchers have begun to study the efficacy
of combined treatment with a COX-2 inhibitor and an EGFR inhibitor.
A phase I trial found combined therapy to be safe with a partial re-
sponse in 33% and stabilization of disease in another 24% of the
patients studied (n 5 22) (3). However, recently published clinical
trials investigating the efficacy of combined therapy with a COX-2
inhibitor and the EGFR inhibitors gefitinib or erlotinib found no ad-
ditional benefit of combined treatment in platinum therapy unrespon-
sive or chemotherapy-naive patients in comparison with results from
Abbreviations: CI, confidence interval; COX-2, cyclooxygenase-2; EGFR,
epidermal growth factor receptor; HR, hazard ratio; IHC, immunohistochem-
istry; mRNA, messenger RNA; NSAID, non-steroidal anti-inflammatory drug;
NSCLC, non-small cell lung cancer; TKI, tyrosine kinase inhibitor.
Ó The Author 2008. Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oxfordjournals.org
1781
previous studies involving treatment with gefitinib or erlotinib alone
(4–7). These studies did not test the efficacy of gefitinib or erlotinib
alone as a comparator arm. These results contrast with a reported
benefit of combined therapy in vitro in cell lines with EGFR muta-
tions, suggesting that the benefit of such treatment may only be seen
when tumors with EGFR mutations are selected (8). Only one pre-
viously published study has examined the relationship between
combined COX-2/EGFR tumor expression and survival among a small
number of patients with lung cancer (34). The authors report finding
no association between combined COX-2/EGFR tumor expression
and survival.
The current study was undertaken to determine the prognostic sig-
nificance of individual or combined expression of COX-2 and EGFR
in women with adenocarcinoma of the lung and to study whether
epidemiological or demographic features affect such expression.
Materials and methods
Study population
The study population and data collection methods have been described else-
where (35). Briefly, subjects were identified through the population-based
Metropolitan Detroit Cancer Surveillance System (MDCSS), a participant in
the National Cancer Institute’s Surveillance, Epidemiology, and End Results
(SEER) program. Women aged 18–74 years and diagnosed with primary
NSCLC in the tri-county (Wayne, Macomb and Oakland) area between 1
November 2001 and 31 October 2005 were eligible to participate. Ascertain-
ment was originally focused on adenocarcinoma histology but was broadened
after 1 November 2004 to include all NSCLC histologies. No proxy interviews
were conducted; therefore, women deceased at ascertainment or first contact
were ineligible. Five hundred and seventy-seven cases completed an interview
(54%), and paraffin-embedded tumor samples were obtained for 402 cases
(69.7%).
Data collection
All local institutional and review boards approved this study. Informed consent
was obtained from each subject prior to study participation. Trained inter-
viewers conducted in-person interviews to collect demographic information,
medication history (aspirin, acetaminophen and non-steroidal anti-inflammatory
drug (NSAID)), smoking history, health history and lifetime estimates of envi-
ronmental tobacco smoke exposure. Ex-smokers were individuals quitting .2
years before diagnosis. Never smokers included those who smoked ,100 cig-
arettes in their lifetime. Medical history included self-report of physician di-
agnoses of asthma, emphysema, allergies, pneumonia, bronchitis, chronic
obstructive pulmonary disease, tuberculosis and cancer. Family history of lung
cancer was coded as yes or no based on the detailed first-degree family history
information collected. Regular pill use (aspirin, acetaminophen and NSAID)
was defined as taking at least one pill, three times per week or more for at least
1 month during the participant’s lifetime. For aspirin, participants were asked
whether they took baby/senior citizen aspirin (81 mg) or adult-strength aspirin
(325 mg). Lung cancer diagnosis dates, clinical data including histological
type, grade, American Joint Committee on Cancer stage, localized/regional/distant
staging, nodal involvement and date of death or date last known alive were
obtained through the MDCSS. Staging was defined according to SEER guide-
lines as follows: localized disease is confined to the carina, hilus or main stem
bronchus or a single tumor confined to one lung; regional disease has invaded
surrounding structures, including spread to regional ipsilateral lymph nodes,
involved multiple masses in the same lobes or main stem bronchus; distant
disease has spread to distant lymph nodes or involved separate tumor nodules
in different lobe, metastasized to the contralateral lung, abdominal organs,
heart, vertebrae, skeletal muscle or was associated with malignant pericardial
or pleural effusion. The MDCSS maintains yearly follow-up of all patients with
a follow-up rate of 98%.
Immunohistochemistry
In total, 292 and 183 tumor blocks were assessed for COX-2 expression and
EGFR expression, respectively. Five micron sections were cut and stained
using standard avidin–biotin techniques with antibody COX-2 clone 229
(Zymed, South San Francisco, CA) and EGFR antibody clone 31G7 (Ventana
Medical Systems, Tuscon, AZ), respectively, manually (COX-2) or with the
Ventana automated slide stainer (EGFR). For COX-2, heat-induced epitope
retrieval was used to unveil the reactivity of chemical groups that have been
masked by formalin fixation. For EGFR, proteolytic enzyme digestion was
performed using sodium citrate. The COX-2 antibody, a mouse monoclonal
antibody, was diluted 1:50 and incubated for 2 h at room temperature. The
monoclonal EGFR antibody was pre-optimized; primary EGFR antibody in-
cubation was for 32 min. Positive and negative controls were included with
each run. Positive controls included colon cancer (COX-2) and cytoplasmic
and invasive breast ductal carcinoma tissue (EGFR). Negative controls were
obtained by omitting the primary antibody. The pathologist was blinded to all
subject characteristics. Sections were scored for percent positivity of tumor
cells (0, ,10, 11–50 and 51–100%) and for staining intensity (none, faint,
moderate and intense). Sections were considered positive if percent positivity
was .10% and staining intensity was faint or higher.
Statistical analysis
To test for differences between positive and negative tumor expression, Stu-
dent’s t-tests were performed for continuous variables; v
2
analysis was con-
ducted for categorical variables. Stepwise logistic regression was performed to
test for associations between clinical and risk factor variables and COX-2,
EGFR and combined COX-2/EGFR expression.
Cox proportional hazards models were used to estimate the relationship
between COX-2 and EGFR tumor expression and prognosis among women
with NSCLC. Age, race, stage and other variables significant in univariate
analyses were included in these models. Stepwise regression was performed
to select variables for inclusion; criteria to stay in the model was P 5 0.10.
Cox regression plots were constructed for COX-2þ versus COX-2 and
EGFRþ versus EGFR. Analysis was stratified by stage (localized and
regional/distant) and race (African-American and Caucasian). The following
additional comparisons were made: COX-2þ/EGFRþ versus COX-2/
EGFR, COX-2þ/EGFRþ versus other (COX-2/EGFR, COX-2þ/
EGFR and COX-2/EGFRþ) and COX-2þ/EGFRþ versus
COX-2/EGFRþ. Additionally, COX-2/EGFRþ was compared with other
(COX-2/EGFR, COX-2þ/EGFR and COX-2þ/EGFRþ) to analyze the
association between combined tumor expression relative to the lowest risk
group (COX-2/EGFRþ). For each of these comparisons, the proportional
hazards assumption was tested by plotting the ln(ln(S(t))) versus ln(t). Anal-
ysis of the relationships between COX-2 and EGFR staining intensity and
percent positive staining and survival were also conducted.
Results
IHC results
Of the 402 tumor blocks obtained, the distribution of histological type
is as follows: 33 (8.2%) squamous cell carcinoma, 304 (75.4%) ade-
nocarcinoma, 10 (2.5%) large cell carcinoma and 55 (13.7%) NSCLC
unspecified. As the majority of cases were adenocarcinoma and to
control for variability in underlying disease processes, subsequent
analyses were restricted to cases with this histological type. As nine
of the adenocarcinoma tumor blocks were not lung tissue, these cases
were excluded from analyses. In 53 and 60 blocks, there was insuffi-
cient material for COX-2 or EGFR analysis, respectively. The remain-
der of the blocks were not available at the time of staining for COX-2
or EGFR. One hundred and fifty-nine (66.8%) of the adenocarcino-
mas were COX-2þ and 108 (68.4%) were EGFRþ. Dual COX-2/
EGFR expression was available for 157 adenocarcinoma cases with
the following results: 18 (11.5%) COX-2/EGFR, 32 (20.4%)
COX-2þ/EGFR, 30 (19.1%) COX-2/EGFRþ and 77 (49.0%)
COX-2þ/EGFRþ.
Demographic characteristics
None of the following demographic characteristics were significantly
different between COX-2þ and COX-2 or between EGFRþ and
EGFR lung cancer patients: race, age at diagnosis, education level,
body mass index, passive cigarette smoke exposure as a child or as an
adult at work or home, smoking pack-year history, smoking status,
history of chronic obstructive lung disease or family history of lung
cancer (Table I).
Clinical characteristics
Neither COX-2 tumor expression nor EGFR tumor expression was
associated with the following clinical parameters: stage, nodal involve-
ment, grade, radiation treatment or surgery (Table II). The percentage
of cells staining positive for COX-2 was greater in African-Americans
than in Caucasians (P 5 0.0089): 0% (19.6 versus 34.8%), ,10%
(2.2 versus 13.0%), 11–50% (28.3 versus 19.0%) and 51–100%
A.L.Van Dyke et al.
1782
(50.0 versus 33.2%). There was a trend toward higher COX-2 staining
intensity among African-Americans in comparison with Caucasians
(P
trend
5 0.0270). Percentage of cells staining positive for EGFR and
EGFR staining intensity did not differ between African-Americans and
Caucasians (P 5 0.46 and P 5 0.74, respectively).
Aspirin and NSAID use
In multivariate analysis, no relationship was observed between a his-
tory of regular (ever/never) aspirin (adult or baby strength), acetamin-
ophen or other NSAID use or duration of use and COX-2 tumor
expression (data not shown). For every 1 year increase in time be-
tween first regular use of adult-strength aspirin and diagnosis, there
was a 6.2% decrease in the probability of having a COX-2þ tumor
after adjusting for race, age at diagnosis, smoking pack-year history
and grade (odds ratio 0.94, 95% CI 0.89–0.98). Time between first
regular pill use and diagnosis was not significantly associated with
probability of having a COX-2-positive tumor for baby aspirin, acet-
aminophen or NSAIDs (data not shown).
Tumor expression and survival
Of the 239 adenocarcinoma cases followed for whom COX-2 and/or
EGFR results were available, only one participant was lost to follow-
up. Median survival times were 37.5 months for cases with COX-2
tumors and 33.0 months for cases with COX-2þ tumors. Median
survival times for cases with EGFR tumors were 35 months and
40 months for cases with EGFRþ tumors. Patients with COX-2þ
tumors had a statistically significant poorer prognosis in comparison
with patients with COX-2 tumors (HR 1.67, 95% CI 1.01–2.78)
(Table III; Figure 1). When analysis was stratified by stage at diag-
nosis, COX-2 expression was no longer significantly associated with
survival time for patients diagnosed at a localized stage or for those
diagnosed at a regional/distant stage (data not shown). A statistically
significant reduction in the risk of death was observed for cases with
EGFRþ tumor status (HR 0.56, 95% CI 0.32–0.98) (Table III; Figure 2).
When analysis was stratified by stage (localized versus regional/distant),
EGFR was associated with a statistically non-significant reduction in
risk of death among patients diagnosed at a regional or distant stage
Table II. Clinical characteristics by COX-2 and EGFR tumor expression in women with adenocarcinoma of the lung
COX-2 EGFR
Positive, n (%) Negative, n (%) P value Positive, n (%) Negative, n (%) P value
Stage
Localized 72 (45.3) 40 (51.3) 49 (45.4) 21 (42.0)
Regional 64 (40.2) 25 (32.0) 39 (36.1) 23 (46.0)
Distant 23 (14.5) 13 (16.7) 0.70 20 (18.5) 6 (12.0) 0.80
Nodal involvement
No 112 (71.3) 56 (72.7) 76 (71.0) 31 (64.6)
Yes 45 (28.7) 21 (27.3) 0.82 31 (29.0) 17 (35.4) 0.42
Grade
Well differentiated 26 (16.4) 23 (29.1) 18 (16.7) 13 (26.0)
Moderately differentiated 57 (35.8) 29 (36.7) 38 (35.2) 22 (44.0)
Poorly differentiated 57 (35.8) 17 (21.5) 39 (36.1) 11 (22.0)
Unknown 19 (12.0) 10 (12.7) 0.50 13 (12.0) 4 (8.0) 0.18
Radiation
No 117 (73.6) 64 (81.0) 76 (70.4) 38 (76.0)
Yes 42 (26.4) 15 (19.0) 0.21 32 (29.6) 12 (24.0) 0.46
Surgery
No 22 (13.8) 8 (10.1) 15 (13.9) 2 (4.0)
Yes 137 (86.2) 71 (89.9) 0.42 93 (86.1) 48 (96.0) 0.10
Table I. Participant characteristics by COX-2 and EGFR tumor expression in women with adenocarcinoma of the lung
COX-2 EGFR
Positive, n (%) Negative, n (%) P value Positive, n (%) Negative, n (%) P value
n (%) 159 (66.8) 79 (33.2) 108 (68.4) 50 (31.6)
Race
Caucasian 117 (73.6) 67 (84.8) 79 (73.2) 38 (76.0)
African-American 36 (22.6) 10 (12.7) 25 (23.2) 9 (18.0)
Other 6 (3.8) 2 (2.5) 0.15 4 (3.7) 3 (6.0) 0.62
Age
Mean (SD) 59.7 (9.5) 60.7 (8.9) 0.46 59.6 (9.4) 57.7 (9.7) 0.24
Smoking status
Never 17 (10.7) 12 (15.2) 13 (12.0) 5 (10.0)
Former 50 (31.4) 24 (30.4) 28 (25.9) 13 (26.0)
Current 92 (57.9) 43 (54.4) 0.41 67 (62.0) 32 (64.0) 0.74
Pack-years among smokers
Mean (SD) 44.4 (29.1) 41.6 (23.0) 0.45 45.7 (29.3) 42.9 (22.4) 0.54
History of COPD
a
No 109 (68.6) 57 (72.2) 73 (67.6) 32 (64.0)
Yes 50 (31.4) 22 (27.8) 0.57 35 (32.4) 18 (36.0) 0.66
Family history of lung cancer
No 113 (71.1) 60 (76.0) 76 (70.4) 37 (74.0)
Yes 46 (28.9) 19 (24.0) 0.43 32 (29.6) 13 (26.0) 0.64
a
COPD, chronic obstructive pulmonary disease, including chronic bronchitis and emphysema.
COX-2, EGFR and survival among women with lung cancer
1783
(HR 0.54, 95% CI 0.28–1.03, P 5 0.06) but not among patients
diagnosed at a localized stage (HR 0.95, 95% CI 0.26–3.47). The data
were reanalyzed excluding four participants who died of causes un-
related to lung cancer; results did not change.
Median survival times in months among cases with adenocarci-
noma of the lung by combined COX-2/EGFR expression were as
follows: COX-2/EGFR, 39.0 months; COX-2þ/EGFR, 33.0
months; COX-2/EGFRþ, 40.0 months and COX-2þ/EGFRþ,
40.0 months. Probability of survival was no different in patients with
COX-2þ/EGFRþ tumors than it was for patients with all other tumor
results combined (COX-2þ/EGFR, COX-2/EGFRþ or COX-2/
EGFR) (HR 1.10, 95% CI 0.65–1.85) or for patients with COX-2/
EGFR tumors (HR 0.72, 95% CI 0.29–1.77) (Table III). When
patients with COX-2þ/EGFRþ tumors were compared with those
who had COX-2þ/EGFR tumors, addition of EGFR expression
was associated with a statistically non-significant reduction in risk
of death (HR 0.64, 95% CI 0.32–1.31). Risk of death was significantly
reduced in women with COX-2/EGFRþ compared with women
with other tumor results (COX-2/EGFR, COX-2þ/EGFR and
COX-2þ/EGFRþ) (HR 0.43, 95% CI 0.20–0.92, P 5 0.03).
When analyses were stratified by race, COX-2þ tumor expression
was associated with a greater reduction in median survival time
among African-American women (41.5 versus 28 months) than
among Caucasian women (37 versus 33 months). However, the risk
of death associated with COX-2 tumor expression was not statistically
significant among African-American women (HR 5.58, 95% CI 0.64–
48.37) or Caucasian women (HR 1.29, 95% CI 0.72–2.30) after ad-
justment for age at diagnosis, surgery, stage, family history of lung
cancer, smoking pack-year history, history of chronic obstructive pul-
monary disease and radiation (Table III). Conversely, EGFR expres-
sion was associated with an improvement in median survival time
among Caucasian women (33 versus 40 months) but not among
African-American women (36 versus 33 months). EGFR tumor ex-
pression was associated with a statistically non-significant decrease in
survival for Caucasian women (HR 0.53, 95% CI 0.27–1.04). When
cases with COX-2þ/EGFRþ tumors were compared with cases with
other expression results (COX-2þ/EGFR, COX-2/EGFRþ and
COX-2/EGFR), no association was found with risk of death in
African-Americans or Caucasians.
There was no trend in the association between COX-2 staining
intensity and survival (P
trend
5 0.40) (data not shown). Moderate
EGFR staining intensity was associated with a 55% reduction in
the risk of death (HR 0.45, 95% CI 0.22–0.92). There was no trend
in the relationship between EGFR percent positivity and survival
(P
trend
5 0.23) (data not shown). Percent COX-2 positivity was as-
sociated with a statistically significant increase in the risk of death
among patients with tumors staining positive for 11–50% of the cells
(HR 1.98, 95% CI 1.04–3.76).
Discussion
While the relationships between COX-2 and EGFR and survival have
been studied independently, only one study examining the relation-
ship between dual COX-2/EGFR tumor status and survival among
lung cancer patients has been published previously (34). It reported
no association between combined COX-2/EGFR tumor status and
survival. As it involved lung cancer samples from 23 COX-2/
EGFR, 9 COX-2/EGFRþ, 14 COX-2þ/EGFR and 7 COX-
2þ/EGFRþ patients (n 5 53), their analysis may not have had suf-
ficient power to detect statistically significant differences. Our con-
siderably larger sample size (n 5 157), however, yielded similar
results.
These findings in lung cancer are in direct contrast with the studies
of cervical and esophageal cancer that suggest that COX-2 and EGFR
expression may be associated with tumor progression and survival
among cancer patients. A comparison of patients with normal tissue,
Barrett’s esophagitis and esophageal adenocarcinoma revealed in-
creasing expression of COX-2 and EGFR from normal tissue to
Barrett’s esophagitis to adenocarcinoma (36). Patients with COX-
2þ/EGFRþ squamous carcinoma of the uterine cervix had a poorer
5 year disease-free survival rate and higher probability of locoregional
recurrence in comparison with patients with COX-2þ/EGFR,COX-
2/EGFRþ or COX-2/EGFR tumors (37). These results support
Table III. Estimated risk of death associated with EGFR and COX-2 positivity in women with adenocarcinoma of the lung
Strata All stages African-American Caucasian
HR
a
(95% CI) HR
b
(95% CI) HR
b
(95% CI)
COX-2 1.67 (1.01–2.78)
c
5.58 (0.64–48.37) 1.29 (0.72–2.30)
EGFR 0.56 (0.32–0.98)
d
0.93 (0.20–4.23) 0.53 (0.27–1.04)
COX-2þ/EGFRþ versus other 1.10 (0.65–1.85) 1.68 (0.39–7.26) 0.80 (0.41–1.58)
COX-2þ/EGFRþ versus COX-2/EGFR 0.72 (0.29–1.77) — 0.58 (0.21–1.60)
COX-2þ/EGFRþ versus COX-2þ/EGFR 0.64 (0.32–1.31) 0.80 (0.16–3.94) 0.68 (0.27–1.70)
COX-2/EGFRþ versus other 0.43 (0.20–0.92)
e
0.33 (0.03–3.30) 0.63 (0.25–1.58)
a
Adjusted for age at diagnosis, race, surgery, stage, family history of lung cancer, smoking pack-year history, history of chronic obstructive pulmonary disease and
radiation.
b
Adjusted for age at diagnosis, surgery, stage, family history of lung cancer, smoking pack-year history, history of chronic obstructive pulmonary disease and
radiation.
c
P 5 0.0470.
d
P 5 0.0431.
e
P 5 0.0298.
Survival in Months
Proportion Surviving
0102030405060
0.0 0.2 0.4 0.6 0.8 1.0
p=0.0470
COX-2 Negative
COX-2 Positive
Fig. 1. Cox proportional hazards estimates of overall survival among
patients with COX-2þ versus COX-2 tumors adjusted for age at diagnosis,
race, surgery, stage at diagnosis, family history of lung cancer, smoking
pack-year history, history of chronic obstructive pulmonary disease and
radiation.
A.L.Van Dyke et al.
1784
the interactions between COX-2 and EGFR in progression of esoph-
ageal and cervical cancer; however, this conclusion may not apply to
adenocarcinoma of the lung as both cancers differ from NSCLC in
etiology and pathogenesis.
The relationship between COX-2 expression and survival for
NSCLC has not been evaluated previously in African-Americans
and only sparingly in women. Previous studies included between 20
and 50% women in their analyses, and the majority of these studies
were conducted in populations outside of USA. Only one study con-
ducted in Japan examined this association stratified by sex (18); they
reported a significant decrease in survival among men, but not
women, with COX-2þ tumors. We observed that COX-2 expression
was associated with a poorer prognosis overall, with higher risk seen
among African-Americans than Caucasians; however, these results
were not statistically significant. While the number of African-
Americans included in this sample was small, these results suggest
that further race-specific research needs to be conducted. Among lung
cancer patients, African-Americans have a poorer prognosis in com-
parison with Caucasians (38). This difference in survival rates has
been associated with decreased probability of having surgical treat-
ment and poorer performance status (39,40); however, when universal
care is available, racial disparities in survival among NSCLC patients
are eliminated (41). At the same time, potential racial differences in
mechanisms underlying lung cancer development and progression
have not been explored. Our data suggest that COX-2-positive tumor
status is associated with decreased survival, and potentially more
aggressive disease, in African-American women. The increased
percent COX-2-positive staining and intensity of staining among
African-Americans may be associated with smoking behavior. Ciga-
rette smoking is associated with induction of COX2 gene expression.
While African-Americans smoke fewer cigarettes per day than their
Caucasian counterparts and start smoking later in life, they are less
likely to quit smoking and more likely to smoke non-filtered cigarettes
with a high tar yield (42). In our sample, African-Americans reported
smoking fewer cigarettes per day in comparison with Caucasians
(16.08 ± 10.28 versus 25.44 ± 12.12, P , 0.0001). In contrast with
the literature, in our sample there was no difference between African-
American women and Caucasian women in the number of years
smoked (34.68 ± 11.97 versus 36.31 ± 11.70, P 5 0.43) or in the
age smoking began (16.80 ± 4.33 versus 17.02 ± 4.28, P 5 0.77).
The increased COX-2 tumor staining among African-American
women may be related to their continued smoking. At the time of
interview, African-American women reported a lower smoking pack-
year history (29.12 ± 24.71 versus 46.82 ± 27.17, P 5 0.0002) and
more frequently reported being current smokers than Caucasian
women (69.6 versus 53.8%, P 5 0.0651). Few studies have examined
the association between cigarette smoking and COX-2 expression in
people with and without cancer. Cigarette smoking, but not number of
cigarettes smoked per day, is related to an increase in COX-2 mRNA
levels in patients with bladder cancer and controls (43). In contrast,
Laga et al. (19) report no association between COX-2 protein expres-
sion as measured by IHC and smoking pack-year history among
NSCLC patients. Thus, this relationship between current smoking
status and COX-2 expression and racial differences in outcome needs
to be explored further.
An association between EGFR protein expression and survival
among NSCLC patients has been shown previously (29,44); however,
other studies have found either no significant association (25,31,45,46)
or an inverse relationship (47,48) between EGFR protein expression
and survival. These findings remain controversial with only a handful of
studies using quantitative methods to measure protein expression. The
majority of studies used IHC to measure protein expression by scoring
percent positivity and/or staining intensity, and at present there is no
standard scoring method or definition of positivity. Furthermore, anti-
body detection method varied across studies, and many of these studies
did not adjust for factors such as stage, age at diagnosis, comorbidities
and smoking status. In contrast, our study did address the relationship
between IHC score and clinical–epidemiological variables.
The improvement in survival among patients with EGFRþ tumors
when tumor expression (positive/negative) was analyzed may have
been associated with their enhanced responsiveness to EGFR tyrosine
kinase inhibitors (TKIs). EGFR TKI therapy has been shown to be
most effective among women, Asians, never smokers and adenocar-
cinoma cases. As our study included only women with adenocarci-
noma, this argument may hold some validity. However, as
chemotherapy information was not collected through this study, this
question could not be addressed directly. Among NSCLC patients,
increased EGFR gene copy number has been associated with en-
hanced protein expression (49,50), and gene copy number, mRNA
expression and protein expression have been associated with im-
proved response to treatment with a TKI and overall survival
(51–53). Hirsch et al. (52) report that NSCLC cases with high EGFR
copy number who were not treated with gefitinib had a far worse
prognosis than cases with high EGFR copy number who were treated
with gefitinib. However, some studies report an association between
EGFR copy number and survival among NSCLC patients treated with
gefitinib but not between EGFR protein expression and prognosis or
survival among these cases (53,54). An analysis of cases with adeno-
carcinoma only revealed a significant association between a response
to treatment with gefitinib and moderate to intense EGFR protein
expression (32). While some studies have found no association be-
tween EGFR protein expression and response to TKI therapy or sur-
vival among patients treated with TKIs, these univariate analyses did
not take into account the important prognostic factors such as stage
and age at diagnosis (50,55). One study by Clarke et al. (56) that did
utilize multivariate analysis reported a greater decrease in the risk of
death with erlotinib treatment among EGFRþ NSCLC cases (HR
0.65) than among EGFR cases (HR 0.83); however, this interaction
between treatment and EGFR expression was not statistically signif-
icant. The converse pattern has also been shown; patients with tumors
negative for EGFR amplification and EGFR protein expression have
not responded to gefitinib or erlotinib in clinical trials (51).
Strengths and limitations
Our study has a number of strengths. It examines the relationship
between COX-2 and EGFR tumor expression and survival in women
and by race. As lung tumors of different histology have significant
differences in clinical and biological features, analyses were limited
to adenocarcinomas instead of including cases with any NSCLC his-
tology. Only in-person interviews regarding risk factors were per-
formed, follow-up was standardized and all cancer diagnoses were
confirmed histologically by the same pathologist, eliminating incon-
sistencies secondary to interobserver variability. The collection of risk
Survival in Months
Proportion Surviving
0 1020304050
0.0 0.2 0.4 0.6 0.8 1.0
p=0.0431
EGFR Negative
EGFR Positive
Fig. 2. Cox proportional hazards estimates of overall survival among
patients with EGFRþ versus EGFR tumors adjusted for age at diagnosis,
race, surgery, stage at diagnosis, family history of lung cancer, smoking
pack-year history, history of chronic obstructive pulmonary disease and
radiation.
COX-2, EGFR and survival among women with lung cancer
1785
factor data and clinical data allowed for adjustment of these variables
in the survival analysis. Moreover, the substantial proportion of cases
who were African-American (19% for COX-2 analysis and 22% for
EGFR analysis) permitted race-specific analysis. Additionally, our
sample size was sufficient to stratify COX-2 and EGFR analyses by
stage and race. However, while we were able to include presence or
absence of surgery and radiation treatment history in the proportional
hazards models, we did not collect data on specific chemotherapeutic
treatments. In addition, our results may only apply to adenocarcino-
mas of the lung rather than to all NSCLC histological types.
In our study, women with COX-2þ tumors had a poorer prognosis
than did women with COX-2 tumors. On stratified analysis, a greater
reduction in survival time was observed among African-American
women with COX-2-positive tumors than among Caucasian women.
EGFRþ tumor status was associated with improved survival time.
Combined COX-2/EGFR tumor expression did not predict survival
among women with adenocarcinoma of the lung in our analysis.
While our results and those of Brattstrom et al. (34) suggest that
combined COX-2/EGFR tumor expression does not predict survival
among lung cancer patients, larger studies should be conducted to an-
alyze whether this relationship varies by stage at diagnosis. Additional
studies with larger numbers of African-Americans need to be conducted
to more fully evaluate the relationship between COX-2 tumor status and
survival stratified by race, stage at diagnosis and smoking status. In
summary, while combined COX-2/EGFR expression does not predict
survival among women with adenocarcinoma of the lung, COX-2 ex-
pression and EGFR expression may predict survival separately.
Funding
National Institutes of Health (R01-CA87895, N01-PC35145,
P30CA22453).
Acknowledgements
Conflict of Interest Statement: None declared.
References
1. Ries,L., Melbert,D., Krapcho,M., Mariotto,A., Miller,B.A., Feuer,E.J.,
Clegg,L., Horner,M.J., Howlader,N., Eisner,M.P., Reichman,M. and
Edwards,B.K. (eds) (2007) SEER Cancer Statistics Review, 1975–2004.
National Cancer Institute. Bethesda, MD, http://seer.cancer.gov/csr/
1975_2004/based on November 2006 SEER data submission, posted to SEER
web site, 2007.
2. Liu,M. et al. (2007) EGFR signaling is required for TGF-b1-mediated
COX-2 induction in human bronchial epithelial cells. Am. J. Respir. Cell
Mol. Biol., 37, 578–588.
3. Reckamp,K.L. et al. (2006) A phase I trial to determine the optimal bi-
ological dose of celecoxib when combined with erlotinib in advanced non-
small cell lung cancer. Clin. Cancer Res., 12, 3381–3388.
4. O’Byrne,K.J. et al. (2007) Combination therapy with gefitinib and rofecox-
ib in patients with platinum-pretreated relapsed non-small-cell lung cancer.
J. Clin. Oncol., 25, 3266–3273.
5. Gadgeel,S.M. et al. (2007) Phase II study of gefitinib, an epidermal growth
factor receptor tyrosine kinase inhibitor (EGFR-TKI), and celecoxib, a cy-
clooxygenase-2 (COX-2) inhibitor, in patients with platinum refractory
non-small cell lung cancer (NSCLC). J. Thorac. Oncol., 2, 299–305.
6. Argwala,A.K. et al. (2006) Gefitinib plus celecoxib in chemotherapy-naive
patients with stage IIIB/IV non-small cell lung cancer (NSCLC): a phase II
study from the Hoosier Oncology Group. J. Clin. Oncol., 24, 18S abstract
7066.
7. Fidler,M.J. et al. (2006) Gastrointestinal hemorrhage in advanced non-
small cell lung cancer (NSCLC) patients treated with erlotinib and cele-
coxib. J. Clin. Oncol., 24, 18S abstract 7172.
8. Gadgeel,S.M. et al. (2007) Response to dual blockade of epidermal growth
factor receptor (EGFR) and cyclooxygenase-2 in nonsmall cell lung cancer
may be dependent on EGFR mutational status of the tumor. Cancer, 110,
2275–2284.
9. Zang,E.A. et al. (1996) Differences in lung cancer risk between men and
women: examination of the evidence. J. Natl Cancer Inst., 88, 183–192.
10. Tam,I.Y.S. et al. (2006) Distinct epidermal growth factor receptor and
KRAS mutation patterns in non-small cell lung cancer patients with differ-
ent tobacco exposure and clinicopathologic features. Clin. Cancer Res., 12,
1647–1653.
11. Liu,X.H. et al. (1998) NS398, a selective cyclooxygenase-2 inhibitor, in-
duces apoptosis and down-regulates bcl-2 expression in LNCaP cells. Can-
cer Res., 58, 4245–4249.
12. Masferrer,J.L. et al. (2000) Antiangiogenic and antitumor activities of cy-
clooxygenase-2 inhibitors. Cancer Res., 60, 1306–1311.
13. Goodwin,J.S. et al. (1983) Regulation of the immune response by prosta-
glandins. J. Clin. Immunol., 3, 295–315.
14. Yuan,A. et al. (2005) Total cyclooxygenase-2 mRNA levels correlate
with vascular endothelial growth factor mRNA levels, tumor angiogenesis
and prognosis in non-small cell lung cancer patients. Int. J. Cancer, 115,
545–555.
15. Hida,T. et al. (1998) Increased expression of cyclooxygenase 2 occurs
frequently in human lung cancers, specifically in adenocarcinomas. Cancer
Res., 58, 3761–3764.
16. Fang,H.Y. et al. (2003) Cyclooxygenase-2 in human non-small cell lung
cancer. Eur. J. Surg. Oncol., 29, 171–177.
17. Lu,C. et al. (2004) Prognostic factors in resected stage I non-small-cell lung
cancer: a multivariate analysis of six molecular markers. J. Clin. Oncol., 22,
4575–4583.
18. Tsubochi,H. et al. (2006) Combined analysis of cyclooxygenase-2 expres-
sion with p53 and Ki-67 in nonsmall cell lung cancer. Ann. Thorac. Surg.,
82, 1198–1204.
19. Laga,A.C. et al. (2005) Prognostic significance of cyclooxygenase 2 ex-
pression in 259 cases on non-small cell lung cancer. Arch. Pathol. Lab.
Med., 129, 1113–1117.
20. Hosomi,Y. et al. (2000) Increased cyclooxygenase 2 (COX-2) expression
occurs frequently in precursor lesions of human adenocarcinoma of the
lung. Lung Cancer, 30, 73–81.
21. Yamaguchi,N.H. et al. (2004) COX-2, MMP-9, and Noguchi classification
provide additional prognostic information about adenocarcinoma of the
lung. A study of 117 patients from Brazil. Am. J. Clin. Pathol., 121, 78–86.
22. Achiwa,H. et al. (1999) Prognostic significance of elevated cyclooxygenase
2 expression in primary, resected lung adenocarcinomas. Clin. Cancer Res.,
5, 1001–1005.
23. Mascaux,C. et al. (2006) Has Cox-2 a prognostic role in non-small-cell
lung cancer? A systematic review of the literature with meta-analysis of the
survival results. Br. J. Cancer, 95, 139–145.
24. Prenzel,N. et al. (2001) The epidermal growth factor receptor family as
a central element for cellular signal transduction and diversification. En-
docr. Relat. Cancer, 8, 11–31.
25. Rusch,V. et al. (1993) Differential expression of the epidermal growth
factor receptor and its ligands in primary non-small cell lung cancers and
adjacent benign lung. Cancer Res., 53, 2379–2385.
26. Hirsch,F.R. et al. (2003) Epidermal growth factor receptor in non-small-
cell lung carcinomas: correlation between gene copy number and protein
expression and impact on prognosis. J. Clin. Oncol., 21, 3798–3807.
27. Veale,D. et al. (1987) Epidermal growth factor receptors in non-small cell
lung cancer. Br. J. Cancer, 55, 513–516.
28. Suzuki,S. et al. (2005) Protein overexpression and gene amplification of
epidermal growth factor receptor in nonsmall cell lung carcinomas: an
immunohistochemical and fluorescence in situ hybridization study. Cancer,
103, 1265–1273.
29. Shah,L. et al. (2004) Expression of syndecan-1 and expression of epidermal
growth factor receptor are associated with survival in patients with non-
small cell lung carcinoma. Cancer, 101, 1632–1638.
30. Fontanini,G. et al. (1998) Evaluation of epidermal growth factor-related
growth factors and receptors of neoangiogenesis in completely resected
stage I-IIIA non-small-cell lung cancer: amphiregulin and microvessel
count are independent prognostic indicators of survival. Clin. Cancer
Res., 4, 241–249.
31. Brabender,J. et al. (2001) Epidermal growth factor receptor and HER2-neu
mRNA expression in non-small cell lung cancer is correlated with survival.
Clin. Cancer Res., 7, 1850–1855.
32. Parra,H.S. et al. (2004) Analysis of epidermal growth factor receptor ex-
pression as a predictive factor for response to gefitinib (‘Iressa’, ZD1839) in
non-small-cell lung cancer. Br. J. Cancer, 91, 208–212.
33. Meert,A.P. et al. (2002) The role of EGF-R expression on patient survival in
lung cancer: a systematic review with meta-analysis. Eur. Respir. J., 20,
975–981.
34. Brattstrom,D. et al. (2004) HER-2, EGFR, COX-2 expression status cor-
related to microvessel density and survival in resected non-small cell lung
cancer. Acta Oncol., 43, 80–86.
A.L.Van Dyke et al.
1786
35. Schwartz,A.G. et al. (2007) Reproductive factors, hormone use, ER expres-
sion and risk of non-small cell lung cancer in women. J. Clin. Oncol., 25,
5785–5792.
36. Li,Y. et al. (2006) Cyclooxygenase-2 and epithelial growth factor receptor
up-regulation during progression of Barrett’s esophagus to adenocarci-
noma. World J. Gastroenterol., 12, 928–934.
37. Kim,G.E. et al. (2004) Synchronous coexpression of epidermal growth
factor receptor and cyclooxygenase-2 in carcinomas of the uterine cervix:
a potential predictor of poor survival. Clin. Cancer Res., 10, 1366–1374.
38. Albano,J.D. et al. (2007) Cancer mortality in the United States by education
level and race. J. Natl Cancer Inst., 99, 1384–1394.
39. Greenwald,H.P. et al. (1998) Social factors, treatment, and survival in
early-stage non-small cell lung cancer. Am. J. Public Health, 88, 1681–
1684.
40. Blackstock,A.W. et al. (2002) Outcomes among African-American/Non-
African-American patients with advanced non-small-cell lung carcinoma:
report from the cancer and leukemia group B. J. Natl Cancer Inst., 94, 284–
290.
41. Mulligan,C.R. et al. (2006) Unlimited access to care: effect on racial dis-
parity and prognostic factors in lung cancer. Cancer Epidemiol. Biomarkers
Prev., 15, 25–31.
42. Kabat,G.C. et al. (1991) Comparison of smoking habits of blacks and
whites in a case-control study. Am. J. Public Health, 81, 1483–1486.
43. Badawi,A.F. et al. (2002) Influence of cigarette smoking on prostaglandin
synthesis and cyclooxygenase-2 gene expression in human urinary bladder
cancer. Cancer Invest., 20, 651–656.
44. Lee,J.S. et al. (1989) Expression of epidermal growth factor receptor
(EGFR): a favorable prognostic factor for surgically resected non-small
cell lung cancer (NSCLC). Proc. Annu. Meet. Am. Soc. Clin. Oncol., 8,
A878.
45. Rusch,V. et al. (1997) Overexpression of epidermal growth factor receptor
and its ligand transforming growth factor a is frequent in respectable non-
small cell lung cancer but does not predict tumor progression. Clin. Cancer
Res., 3, 515–522.
46. Kanematsu,T. et al. (2003) Phosphorylation, but not overexpression, of
epidermal growth factor receptor is associated with poor prognosis of
non-small cell lung cancer patients. Oncol Res., 13, 289–298.
47. Tateishi,M. et al. (1990) Immunohistochemical evidence of autocrine
growth factors in adenocarcinoma of the human lung. Cancer Res., 50,
7077–7080.
48. Veale,D. et al. (1993) The relationship of quantitative epidermal growth
factor receptor expression in non-small cell lung cancer to long term sur-
vival. Br. J. Cancer, 68, 162–165.
49. Dziadziuszko,R. et al. (2007) Epidermal growth factor receptor gene copy
number and protein level are not associated with outcome of non-small-cell
lung cancer patients treated with chemotherapy. Ann. Oncol., 18, 447–452.
50. Jeon,Y.K. et al. (2006) Clinicopathologic features and prognostic implica-
tions of epidermal growth factor receptor (EGFR) gene copy number and
protein expression in non-small cell lung cancer. Lung Cancer, 54, 387–
398.
51. Hirsch,F.R. et al. (2007) Combination of EGFR gene copy number and
protein expression predicts outcome for advanced non-small-cell lung can-
cer patients treated with gefitinib. Ann. Oncol., 18, 752–760.
52. Hirsch,F.R. et al. (2006) Molecular predictors of outcome with gefitinib in
a phase III placebo-controlled study in advanced non-small-cell lung can-
cer. J. Clin. Oncol., 24, 5034–5042.
53. Cappuzzo,F. et al. (2005) Epidermal growth factor receptor gene and pro-
tein and gefitinib sensitivity in non-small-cell lung cancer. J. Natl Cancer
Inst., 97, 643–655.
54. Cappuzzo,F. et al. (2007) Prospective study of gefitinib in epidermal
growth factor receptor fluorescence in situ hybridization—positive/phos-
phor-Akt—positive or never smoker patients with advanced non-small-cell
lung cancer: the ONCOBELL trial. J. Clin. Oncol., 25, 2248–2255.
55. Perez-Soler,R. et al. (2001) A phase II trial of the epidermal growth factor
receptor (EGFR) tyrosine kinase inhibitor OSI-774, following platinum-
based chemotherapy, in patients with advanced, EGFR-expressing, non-
small cell lung cancer (NSCLC). Proc. Am. Soc. Clin. Oncol., 20, 310A
abstract 1235.
56. Clark,G.M. et al. (2006) Smoking history and epidermal growth factor
receptor expression as predictors of survival benefit from erlotinib for pa-
tients with non-small-cell lung cancer in the National Cancer Institute of
Canada Clinical Trials Group Study BR.21. Clin. Lung Cancer, 7, 389–394.
Received January 31, 2008; revised April 24, 2008; accepted April 25, 2008
COX-2, EGFR and survival among women with lung cancer
1787
