Abstract
Rationale: Differences in patient characteristics and outcomes have been observed among current, former, and never-smokers with lung cancer, but most prior studies included few never-smokers and were not prospective.
Objectives: We used data from a large, prospective study of lung cancer care and outcomes in the United States to compare characteristics of never-smokers and smokers with lung cancer and to examine survival among the never-smokers.
Methods: Smoking status at diagnosis was determined by self-report and survival was determined from medical records and cancer registries, with follow-up through June 2010 or later. Cox regression was used to examine the association between smoking and survival, and to identify predictors of survival among never-smokers.
Measurements and Main Results: Among 3,410 patients with lung cancer diagnosed between September 1, 2003 and October 14, 2005 who completed a baseline patient survey, there were 274 never-smokers (8%), 1,612 former smokers (47%), 1,496 current smokers or smokers who quit recently (44%), and 28 with missing information about smoking status (<1%). Never-smokers appeared more likely than former and current/recent smokers to be female and of Asian or Hispanic race/ethnicity, and to have adenocarcinoma histology, fewer comorbidities, private insurance, and higher income and education. Compared with never-smokers, the adjusted hazard of death from any cause was 29% higher among former smokers (hazard ratio, 1.29; 95% confidence interval, 1.08–1.55), and 39% higher among current/recent smokers (hazard ratio, 1.39; 95% confidence interval, 1.16–1.67). Factors predicting worse overall survival among never-smokers included Hispanic ethnicity, severe comorbidity, undifferentiated histology, and regional or distant stage. Never-smoking Hispanics appeared more likely to have regional or advanced disease at diagnosis and less likely to undergo surgical resection, although these differences were not statistically significant.
Conclusions: Never-smokers with lung cancer are more likely than ever-smokers to be female, Asian or Hispanic, and more advantaged socioeconomically, suggesting possible etiologic differences in lung cancer by smoking status. Among never-smokers, Hispanics with lung cancer had worse survival than non-Hispanic whites.
Keywords: lung cancer, cigarette smoking, survival, Hispanic ethnicity, Asian ethnicity
Lung cancer is the leading cause of cancer death in the world (1, 2). Most disease is detected at an advanced stage and, despite gradual therapeutic advances, prognosis is still poor, with an overall 5-year survival rate of 15% (3, 4). Tobacco smoke exposure is the primary risk factor for lung cancer, but some patients do not have a history of smoking. In part because of a lack of smoking information in most cancer registries, population-based information about lung cancer in never-smokers and how it differs from the disease in those who are current or former smokers is limited. Although controversial, studies have reported that lung cancer in never-smokers is more frequently diagnosed among women and in Asian populations (5) and is more likely to be of adenocarcinoma histology (6–8).
Previous studies of prognosis have reported conflicting results, with some showing better survival among never-smokers (9), whereas others have not found a survival difference (10–12). Some of the heterogeneity in prior study results may be due in part to differences in clinical populations, data quality, data availability, and statistical power. In this study, we performed a secondary analysis of data from the Cancer Care Outcomes Research and Surveillance (CanCORS) study, a large, prospective, observational study of lung and colorectal cancer care and outcomes, to compare patient characteristics and survival between never-smokers and ever-smokers with lung cancer, and to identify independent predictors of survival among CanCORS participants with lung cancer who were never-smokers.
Methods
CanCORS is a prospective observational study of cancer care and outcomes among racially and ethnically diverse patients with lung and colorectal cancer. The methods of the study have been reported previously in detail (13). Patients newly diagnosed with lung cancer were enrolled from four large, population-based cancer registries representing Alabama, Iowa, Northern California, and Los Angeles County; five integrated health care delivery systems (located in Boston, MA; Detroit, MI; Hawaii; Portland, OR; and Seattle, WA); and 13 Veterans Health Administration (VHA) facilities (located in Atlanta, GA; Baltimore, MD; Biloxi, MS; Chicago, IL [Lakeside and Hines]; Durham, NC; Houston, TX; Indianapolis, IN; Minneapolis, MN; Nashville, TN; New York, NY; Seattle, WA; and Tucson, AZ). In 2000, these diverse settings captured approximately 10% of the total U.S. population, including almost 10% of all U.S. lung cancer cases. The demographic and tumor characteristics of CanCORS participants with lung cancer are similar to those of patients included in the Surveillance, Epidemiology and End Results (SEER) tumor registry, except that CanCORS included slightly fewer patients with advanced disease (14).
Patients
Eligible patients were diagnosed with incident, first primary lung cancer between September 1, 2003, and October 14, 2005. Overall, 5,525 patients with lung cancer were eligible and enrolled in the CanCORS study, out of 14,327 individuals who were identified by rapid case ascertainment within 6 months of diagnosis. For this analysis, there were 3,891 participants for whom medical records were abstracted and a telephone-based survey was completed, either by the patient or a surrogate. Of these, we excluded 26 patients with incomplete follow-up at one CanCORS site, 450 patients who were administered a brief version of the survey that skipped questions related to smoking status, and 5 patients with missing information about vital status, leaving 3,410 patients in this analysis (Figure 1).
Figure 1.
Cohort assembly. CanCORS = Cancer Care Outcomes Research and Surveillance.
Informed consent was provided by all patients or an appropriate surrogate. The study protocol for the present analysis was approved by the institutional review board at Stanford University (Stanford, CA), and the parent study protocol was approved by all of the CanCORS primary data collection sites.
Medical Records Data
Data from medical records of CanCORS participants were collected by professional chart abstractors. Variables of interest included patient characteristics (e.g., comorbidities), tumor characteristics (e.g., histology, stage), and vital status at the time of last contact. Histology was coded as adenocarcinoma, squamous cell carcinoma, large cell carcinoma, small cell carcinoma, or undifferentiated/other. Stage at diagnosis for most patients was coded according to the eighth edition of the American Joint Committee on Cancer (AJCC, Chicago, IL) coding manual. However, because stage was recorded as local, regional, or distant in a minority of participants, we mapped AJCC stage into these categories, treating patients with stage I and II disease as “local,” stage III as “regional,” and stage IV as “distant.” Treatment variables captured all treatment initiated within 6 months of diagnosis. Insurance coverage was classified as private, Medicare, Medicaid, other public (including VHA), none, or unknown.
To measure comorbidities, the CanCORS study used the Adult Co-morbidity Evaluation-27 (ACE-27). This validated, medical records–based instrument ranks patients as having none, mild, moderate, or severe comorbidity, based on the most severe comorbid conditions identified (15).
Survey Data
Information about age, sex and race/ethnicity was obtained, in order of priority, from the baseline patient interview, medical record abstraction, or patient tracking records. To facilitate analysis and interpretation, we recategorized self-reported race (17 categories) and ethnicity (7 categories) into a single variable with five mutually exclusive categories: white, Hispanic or Latino (regardless of race), African American, Asian/Pacific Islander, and multiple races/other/unknown. Smoking habits, annual household income, and education were collected at the time of the baseline interview. Ever-smokers were defined as patients who answered “yes” to the question “Have you ever smoked cigarettes regularly, that is, at least a few cigarettes per day?” Former smokers were patients who answered “yes” to the preceding question, and “no” to the question “Do you smoke cigarettes regularly now?” Never-smokers were defined as patients who answered “no” to the first question. Post hoc, we grouped “recent quitters” (those whose age at the time of quitting was within 1 yr of their age at diagnosis) with “current smokers,” because these two groups of patients had similar baseline characteristics and outcomes, and it is likely that at least some of these individuals had symptomatic lung cancer at the time of quitting. We refer to this group as current/recent smokers.
Follow-Up
Follow-up was measured in days from the date of diagnosis until the date of death, last contact, or last vital status update, whichever occurred latest. Follow-up for vital status was ascertained from health plan records (for integrated system and VA sites) and state and national death records through linkage to population-based cancer registries (for geography-based sites). Vital status was last updated between October 2011 and April 2012, except for one site at which updating was last performed in June 2010.
Statistical Analysis
Differences between patients with lung cancer who were smokers at the time of survey or up to 1 year before diagnosis, former smokers, and never-smokers were examined with respect to sociodemographic and clinical characteristics, including age, sex, race, comorbidity, insurance, income, education level, health care setting, histology, stage, receipt of surgery, receipt of chemotherapy, and receipt of radiotherapy, using chi-square tests. Unadjusted differences in survival among these three groups of patients by smoking status were compared by the Kaplan–Meier method. Cox proportional hazards regression analysis, with days since diagnosis as the time scale, was used to identify factors independently associated with survival among never-smokers. We tested for violations of the proportional hazards assumption by entering time-dependent covariates into models and using the proportionality test function in SAS version 9.1 (SAS Institute, Cary, NC). Levels of categorical covariates that violated the proportional hazards assumption and had crossing survival curves were combined.
Multiple imputations were used to handle missing data from the patient survey (16), including insurance (4.6% of observations), income (8.4%), education level (0.9%), and smoking status (0.8%). We did not impute data missing due to survey type, because we were unable to acquire smoking status information for the 450 patients who completed the brief version of the survey. Further details regarding the multiple imputation models have been previously reported (17).
All analyses were performed with SAS. We did not adjust for multiple comparisons. We considered a P value less than 0.05 to be statistically significant.
Results
The sample included 3,410 patients: 2,044 men (60%) and 1,366 women (40%). There were 274 (8%) never-smokers, 1,612 (47%) former smokers, and 1,496 (44%) current/recent smokers at the time of diagnosis (Table 1). At the time of most recent follow-up, 207 never-smokers (76%), 1,353 former smokers (84%), and 1,277 current/recent smokers (85%) had died.
Table 1.
Characteristics of study patients
| Variable | Level | n | % |
|---|---|---|---|
| Smoking status | Never-smoker | 274 | 8.04 |
| Former smoker | 1,612 | 47.3 | |
| Current/recent smoker | 1,496 | 43.9 | |
| Missing | 28 | 0.8 | |
| Age, yr | 0–59 | 747 | 21.9 |
| 60–69 | 1,031 | 30.2 | |
| 70–79 | 1,112 | 32.6 | |
| 80+ | 520 | 15.2 | |
| Sex | Male | 2,044 | 59.9 |
| Female | 1,366 | 40.1 | |
| Race/ethnicity | White, non-Hispanic | 2,593 | 76.0 |
| Hispanic or Latino | 145 | 4.3 | |
| African American | 365 | 10.7 | |
| Asian/Pacific Islander | 138 | 4.1 | |
| Multiple/other/unknown | 169 | 5.0 | |
| Comorbidity | None | 592 | 17.4 |
| Mild | 1,277 | 37.4 | |
| Moderate | 754 | 22.1 | |
| Severe | 787 | 23.1 | |
| Insurance | Private/HMO | 2,255 | 66.1 |
| Medicare | 681 | 20.0 | |
| Medicaid | 86 | 2.5 | |
| Other public | 166 | 4.9 | |
| Unknown | 3 | 0.1 | |
| None | 65 | 1.9 | |
| Missing | 154 | 4.5 | |
| Annual household income | Less than $20,000 | 1,144 | 33.6 |
| $20,000–$40,000 | 1,019 | 29.9 | |
| $40,000–$60,000 | 477 | 14.0 | |
| $60,000 or more | 484 | 14.2 | |
| Missing | 286 | 8.4 | |
| Education | Not applicable | 7 | 0.2 |
| Some grade school | 772 | 22.6 | |
| High school graduate | 1,209 | 35.4 | |
| Some college | 850 | 24.9 | |
| College graduate | 332 | 9.7 | |
| Graduate or professional school | 210 | 6.2 | |
| Missing | 30 | 0.9 | |
| Health care setting | Nonintegrated | 2,122 | 62.2 |
| Integrated | 828 | 24.3 | |
| VHA | 440 | 12.9 | |
| Unknown | 20 | 0.6 | |
| Number of cigarettes per day (among current/recent and former smokers) | Not applicable | 274 | 8.0 |
| 1–10 | 321 | 9.4 | |
| 11–20 | 1,015 | 29.8 | |
| 21–30 | 463 | 13.6 | |
| 31–40 | 743 | 21.8 | |
| 41+ | 459 | 13.5 | |
| Missing | 135 | 4.0 | |
| Years since quitting smoking (among former smokers) | 1–5 | 329 | 9.7 |
| 6–10 | 263 | 7.7 | |
| 11–20 | 446 | 13.1 | |
| More than 20 | 583 | 17.1 | |
| Missing | 1,789 | 52.5 | |
| Histology | Adenocarcinoma | 1,005 | 29.5 |
| Squamous cell | 671 | 19.7 | |
| Small cell | 411 | 12.1 | |
| Large cell | 131 | 3.8 | |
| Undifferentiated or other | 708 | 20.8 | |
| Missing | 484 | 14.2 | |
| Stage | Local | 998 | 29.3 |
| Regional | 930 | 27.3 | |
| Distant | 1,390 | 40.8 | |
| Missing | 92 | 2.7 | |
| Surgery | Yes | 1,067 | 31.3 |
| No | 2,215 | 65.0 | |
| Unknown/not applicable | 128 | 3.8 | |
| Chemotherapy | Yes | 1,780 | 52.2 |
| No | 1,630 | 47.8 | |
| Radiotherapy | Yes | 1,593 | 46.7 |
| No | 1,578 | 46.3 | |
| Unknown/not applicable | 239 | 7.0 |
Definition of abbreviations: HMO = health maintenance organization; VHA = Veterans Health Administration.
As shown in Table 2, there were statistically significant associations between smoking status and age, sex, race/ethnicity, comorbidity, insurance coverage, household income, education, health care setting, histology, and treatment with chemotherapy or radiotherapy (P < 0.0001 for all comparisons). Compared with former and current/recent smokers, never-smokers with lung cancer appeared more likely to be female, to be Hispanic or Asian, to have mild or no comorbidities, and to have at least some college education. In addition, they were far more likely to have adenocarcinoma histology. Compared with current smokers, never-smokers apparently were more likely to have private health insurance, to have an annual household income greater than $60,000, and to receive care in a nonintegrated health system.
Table 2.
Characteristics of study patients, stratified by smoking status (with imputed data)
| Variable | Level | Never-Smokers (%) | Former Smokers (%) | Current Smokers (%) | P Value |
|---|---|---|---|---|---|
| Age, yr | 51–60 | 29.3 | 10.9 | 32.4 | <0.0001 |
| 61–70 | 19.3 | 27.0 | 35.7 | ||
| 71–80 | 25.9 | 39.9 | 26.1 | ||
| >80 | 25.6 | 22.3 | 5.8 | ||
| Sex | Male | 37.5 | 63.8 | 59.9 | <0.0001 |
| Female | 62.5 | 36.2 | 40.1 | ||
| Race/ethnicity | White, non-Hispanic | 59.6 | 80.0 | 74.8 | <0.0001 |
| Hispanic | 10.6 | 4.0 | 3.4 | ||
| African American | 9.8 | 8.0 | 13.8 | ||
| Asian/Pacific Islander | 15.2 | 3.7 | 2.4 | ||
| Multiple/other/unknown | 4.9 | 4.5 | 5.5 | ||
| Comorbidity | None | 24.0 | 14.6 | 19.1 | <0.0001 |
| Mild | 42.9 | 35.6 | 38.4 | ||
| Moderate | 20.9 | 23.7 | 20.6 | ||
| Severe | 12.2 | 26.0 | 21.9 | ||
| Insurance | Private/HMO | 73.9 | 72.0 | 59.4 | <0.0001 |
| Medicare | 13.8 | 21.1 | 20.1 | ||
| Medicaid/other public | 6.2 | 4.1 | 11.3 | ||
| Unknown/none | 6.2 | 2.8 | 9.2 | ||
| Annual household income | Less than $20,000 | 30.3 | 30.5 | 41.0 | <0.0001 |
| $20,000–$40,000 | 28.7 | 36.7 | 31.0 | ||
| $40,000–$60,000 | 14.8 | 17.5 | 15.9 | ||
| More than $60,000 | 26.3 | 15.3 | 12.2 | ||
| Education | Some grade school | 18.1 | 22.6 | 24.5 | <0.0001 |
| High school graduate | 26. 8 | 34.0 | 38.9 | ||
| Some college or more | 55.2 | 43.4 | 36.7 | ||
| Health care setting | Nonintegrated | 67.7 | 62.0 | 62.3 | <0.0001 |
| Integrated | 27.8 | 26.4 | 21.7 | ||
| VHA | 4.5 | 11.6 | 16.0 | ||
| Number of cigarettes per day (among current/recent and former smokers) | 1–10 | 12.5 | 8.5 | ||
| 11–20 | 34.2 | 32.1 | |||
| 21–30 | 15.5 | 16.0 | |||
| 31–40 | 22.0 | 27.6 | |||
| 40+ | 15.8 | 15.7 | |||
| Years since quit smoking (among former smokers) | 1–5 | 20.3 | |||
| 6–10 | 16.2 | ||||
| 11–20 | 27.5 | ||||
| >20 | 36.0 | ||||
| Histology | Adenocarcinoma | 61.3 | 35.6 | 28.1 | <0.0001 |
| Squamous cell | 8.4 | 24.1 | 24.3 | ||
| Small cell | 3.0 | 11.7 | 18.6 | ||
| Larger cell | 3.8 | 4.3 | 4.8 | ||
| Undifferentiated or other | 23.5 | 24.3 | 24.2 | ||
| Stage | Local | 27.4 | 32.6 | 27.8 | 0.01 |
| Regional | 26.3 | 28.3 | 28.1 | ||
| Distant | 46.3 | 39.1 | 44.1 | ||
| Surgery | Yes | 34.9 | 34.0 | 30.6 | 0.10 |
| No | 65.1 | 66.1 | 69.4 | ||
| Chemotherapy | Yes | 49.7 | 47.9 | 57.3 | <0.0001 |
| No | 50.3 | 52.1 | 42.7 | ||
| Radiotherapy | Yes | 39.3 | 47.7 | 54.9 | <0.0001 |
| No | 60.7 | 52.3 | 45.1 |
Definition of abbreviations: HMO = health maintenance organization; VHA = Veterans Health Administration.
Stage distribution was most favorable for former smokers, but differences among the three groups were relatively modest (P = 0.01). Never-smokers less frequently received radiation therapy than former and current/recent smokers (39, 48, and 55%, respectively), and also appeared less likely to have received chemotherapy than current/recent smokers (50 vs. 57%, respectively), but not former smokers (48%).
Unadjusted median survival was longer in never-smokers (507 d; interquartile range [IQR], 116–2467) than in former smokers (330 d; IQR, 101–1298) and current/recent smokers (323 d; IQR, 104–1070) (P < 0.0001) (Figure 2). In an analysis adjusted for age, sex, race, histology, stage, comorbidities, treatment, insurance, education, income, and health care setting, the hazard of death was 29% greater among former smokers compared with never-smokers (HR, 1.29; 95% CI, 1.08–1.55; P = 0.005), and 39% greater among current/recent smokers compared with never-smokers (HR, 1.39; 95% CI, 1.16–1.67; P < 0.001). Results were similar when we excluded the three treatment variables from the model. Results were also similar when we grouped “recent quitters” with “former smokers” instead of “current smokers,” although the magnitude of the association between smoking and worse survival became greater for former/recent smokers than current smokers in the full model.
Figure 2.
Kaplan–Meier survival curves for never-smokers (dashed green line), former smokers (dashed red line), and current/recent smokers (solid blue line). Tic marks indicate censored observations. The x axis represents time in days from date of diagnosis to date of most recent vital status assessment.
In an analysis restricted to never-smokers (Table 3), factors associated with worse survival included Hispanic ethnicity (compared with non-Hispanic white ethnicity), severe comorbidity (compared with no comorbidity), some college education or higher (compared with some grade school), undifferentiated histology (compared with adenocarcinoma histology), regional or distant stage (compared with local stage), and no treatment (compared with surgery or chemotherapy).
Table 3.
Multivariate analysis of factors related to survival among nonsmokers
| Variable | Level | Hazard Ratio | 95% CI | P Value |
|---|---|---|---|---|
| Age, yr | 51–59 | 1.00 | Reference | |
| 60–70 | 1.05 | 0.60–1.86 | 0.86 | |
| 70–79 | 1.06 | 0.60–1.85 | 0.85 | |
| 80+ | 1.08 | 0.56–2.08 | 0.81 | |
| Sex | Male | 1.00 | Reference | |
| Female | 0.76 | 0.52–1.13 | 0.18 | |
| Race/ethnicity | Non-Hispanic white | 1.00 | Reference | |
| Hispanic or Latino | 2.06 | 1.07–3.99 | 0.03 | |
| African American | 1.75 | 0.96–3.18 | 0.07 | |
| Asian/Pacific Islander | 0.83 | 0.47–1.49 | 0.53 | |
| Multiple/other/unknown | 0.80 | 0.30–2.16 | 0.66 | |
| Comorbidity | None | 1.00 | Reference | |
| Mild | 1.49 | 0.87–2.55 | 0.15 | |
| Moderate | 1.67 | 0.92–3.01 | 0.09 | |
| Severe | 3.21 | 1.52–6.79 | 0.002 | |
| Insurance | Private/HMO | 1.00 | Reference | |
| Medicare | 1.06 | 0.60–1.88 | 0.84 | |
| Medicaid/other public | 1.56 | 0.67–3.63 | 0.30 | |
| Unknown/none | 1.28 | 0.52–3.18 | 0.59 | |
| Income | <$20,000 | 1.00 | Reference | |
| $20,000–$40,000 | 0.82 | 0.45–1.46 | 0.49 | |
| $40,000–$60,000 | 0.71 | 0.32–1.57 | 0.39 | |
| $60,000 or more | 0.56 | 0.26–1.20 | 0.14 | |
| Education | Some grade school | 1.00 | Reference | |
| High school graduate | 1.22 | 0.65–2.29 | 0.54 | |
| Some college or more | 2.00 | 1.02–3.93 | 0.04 | |
| Health care setting | Nonintegrated | 1.00 | Reference | |
| Integrated | 0.79 | 0.53–1.19 | 0.26 | |
| VHA | 0.72 | 0.27–1.90 | 0.50 | |
| Histology | Adenocarcinoma | 1.00 | Reference | |
| Squamous cell | 1.35 | 0.73–2.51 | 0.33 | |
| Small cell | 1.80 | 0.58–5.56 | 0.31 | |
| Large cell | 2.33 | 0.90–6.07 | 0.08 | |
| Undifferentiated/other | 1.73 | 1.12–2.68 | 0.01 | |
| Stage | Local | 1.00 | Reference | |
| Regional/distant | 3.92 | 2.08–7.37 | <0.0001 | |
| Surgery | Yes | 1.00 | Reference | |
| No | 5.25 | 3.00–9.21 | <0.0001 | |
| Chemotherapy | Yes | 1.00 | Reference | |
| No | 2.39 | 1.50–3.81 | 0.0003 | |
| Radiotherapy | Yes | 1.00 | Reference | |
| No | 1.48 | 0.98–2.23 | 0.06 |
Definition of abbreviations: CI = confidence interval; HMO = health maintenance organization; VHA = Veterans Health Administration.
Results were similar in multivariable models that adjusted for insurance, income, and education separately (see Tables E1a–E1c in the online supplement), although some college education was less strongly associated with the outcome (and not statistically significant) when modeled separately. Results were also similar when we excluded treatment variables, except that Hispanic ethnicity and education were no longer associated with survival (Table E2). In addition, without adjustment for treatment, an annual income of $20,000 to $40,000 (compared with <$20,000) was associated with better survival, and large cell histology (compared with adenocarcinoma) was associated with worse survival.
In post hoc analyses performed to explore reasons for worse survival among never-smoking Hispanics, we found that, compared with non-Hispanic whites, they were more likely to have regional or advanced disease at diagnosis and less likely to undergo surgical resection; although these differences appeared to be clinically important in magnitude, they were not statistically significant (Table E3).
Discussion
In this large, prospective cohort study of practices and outcomes for patients with lung cancer, we found that never-smokers with lung cancer have a different clinical and demographic profile than smokers, with more adenocarcinoma, higher socioeconomic status, fewer comorbidities, and proportionally more women, Asians, and Hispanics (5, 8–10, 12).
Previous studies have reported conflicting results for associations between smoking status and age or stage at diagnosis (5, 9). We found that current/recent smokers were younger than never-smokers and former smokers with lung cancer, perhaps related to competing risks of death from other causes, and that former smokers had a slightly more favorable stage distribution. The latter finding is difficult to explain, and probably not related to greater use of screening in former smokers, given that enrollment in the CanCORS study concluded several years before screening by low-dose computed tomographic scanning began to disseminate into clinical practice. It is possible that there was selectively greater participation by former smokers with less advanced disease, or that the difference was due to chance.
We also found that current smokers were more likely to receive chemotherapy and/or radiotherapy, and that never-smokers were less likely to receive radiotherapy, whereas stage distribution was similar between these two groups. Although it is not clear why there were differences in treatment across groups, it suggests that worse survival in current smokers probably was not related to less intensive treatment, although the slightly less frequent use of surgery in current smokers may have contributed.
Our analysis extends the findings of several previous studies that reported better survival after lung cancer among never-smokers than ever-smokers. In an analysis of 132 never-smokers and 522 current smokers with non–small cell lung cancer (NSCLC), Nordquist and colleagues found that survival at 5 years was worse among smokers than never-smokers after adjustment for other prognostic factors (HR, 1.3; 95% CI, 1.04–1.69; P = 0.02) (11). Bryant and Cerfolio reported similar results for smokers versus never-smokers after adjustment for sex and completeness of resection (HR, 1.21; 95% CI, 0.98–1.39; P = 0.07), especially among those with stage I disease (5-yr survival, 75 vs. 62%, respectively; P = 0.02) (10). Another study of 98 patients with NSCLC (68 smokers and 30 never-smokers) found that 2-year overall survival rates were better for never-smokers than smokers (52 vs. 17%; P = 0.002) (18). Likewise, in an analysis of data from 883 patients with NSCLC, including 286 never-smokers (32%), Toh observed worse survival among smokers (HR, 1.30; 95% CI, 1.04–1.62) (19). Finally, Tsao and colleagues found worse survival among smokers (P < 0.0001) in an analysis of 1,370 patients with stage III and IV NSCLC treated with chemotherapy or chemoradiation (20).
Two prior studies did not show an association between smoking status and survival. Subramanian and colleagues observed no difference in survival between nonsmokers and smokers with NSCLC in a case–control study of 254 never-smokers who were matched to 254 smokers by sex, histology, stage, and year of diagnosis (5-yr survival: 27% for never-smokers vs. 31% for smokers; P = 0.73) (12). Matching for stage, histology, and other clinical variables may have obscured differences between groups. In another investigation of 317 patients with NSCLC (117 never-smokers) in Singapore, unadjusted survival was worse for ever-smokers than never-smokers, although it was not different after adjustment for stage, treatment, weight loss, and performance status (HR, 0.98; 95% CI, 0.70–1.38; P = 0.92) (21).
In our analysis of survival among never-smokers, we found that compared with non-Hispanic whites, Hispanics had significantly worse survival and African Americans showed a trend toward worse survival. Worse outcomes for African Americans with lung cancer have been documented previously (22, 23), including worse outcomes for African American nonsmokers (24). Hispanics with stage I lung cancer have been shown to have worse survival than non-Hispanic whites, likely attributed in large part to a lower frequency of resection, which we also observed as a nonsignificant trend (25). Similarly, members of our group previously analyzed data from the Greater Bay Area Cancer Registry for nonsmoking women in particular, and reported worse survival among foreign-born Hispanic women than non-Hispanic white women (26).
In contrast, an analysis of data from the SEER cancer registries found that the hazard of death was 15% lower among Hispanic smokers and nonsmokers with any stage of lung cancer, when compared with non-Hispanic whites (27). The present analysis extends the scope of these findings by including men and patients with lung cancer from diverse geographic locations. Of interest, our finding that Hispanic never-smokers fared worse than white never-smokers was apparent only when we adjusted for treatment, indicating that treatment is an important source of confounding of this relationship.
We did not find a significant difference in lung cancer survival between never-smoking Asians and non-Asians. In contrast, Ou and colleagues examined registry data from Southern California and reported that Asian background was an independent and favorable prognostic factor for survival among both smokers (HR, 0.87; 95% CI, 0.81–0.93; P = 0.0001) and never-smokers (HR, 0.84; 95% CI, 0.73–0.97; P = 0.0180) with lung cancer (28). In fact, our findings are consistent with those of Ou and colleagues, given that the HR comparing never-smoking Asians with non-Asians was 0.83 in our study and 0.84 in theirs. If Asians do experience better survival, regardless of smoking status, the survival advantage may be due in part to the higher prevalence of certain mutations in the epidermal growth factor receptor tyrosine kinase (EGFR-TK) among Asians, conferring a better response to therapy with EGFR inhibitors, or a better prognosis independent of treatment.
Unique strengths of our study include the large sample; prospective collection of detailed information regarding patient and tumor characteristics, including sociodemographics and comorbidities; limited amounts of missing data addressed by using advanced methods of multiple imputation; and enrollment of a geographically, racially, and ethnically diverse cohort of patients that is similar to the SEER registry in its representation of the U.S. lung cancer population.
The study has several limitations. First, self-report of smoking status could have resulted in misclassification, although previous studies have shown that self-reported smoking status has good concordance with serum or urine cotinine levels (29). Also, because smoking status was reported prospectively regarding lung cancer follow-up, any misclassification was likely nondifferential with respect to prognosis.
Another limitation is that patients with advanced disease were somewhat less likely to participate in CanCORS, and our results may not, therefore, be as applicable to patients with advanced lung cancer. Although participation may have differed somewhat by smoking status, selective participation of “healthier” smokers would have predictably attenuated the observed survival advantage associated with never smoking. Selective participation by individuals with better access to care also may have influenced the results in unpredictable ways. For example, among never-smokers with advanced disease, participation rates may have been positively associated with years of education, which would explain the counterintuitive finding that participants with at least some college education had an increased hazard of death. Finally, we did not include data about treatment completion or complications of treatment, which may have accounted for some of the differences we observed.
Conclusions
In a large and diverse, national sample of patients with lung cancer, we found that never-smokers with lung cancer were more frequently female and Asian, and more likely to have adenocarcinoma and less severe comorbidities than smokers. We also found that survival was better for never-smokers with lung cancer, when compared with both former smokers and current/recent smokers. Among never-smokers, Hispanics with lung cancer had worse survival than non-Hispanic whites, but it remains unclear whether this is related to underlying genetic differences in either cancer biology or treatment response, disparities in treatment or access to care, or a combination of these factors.
Additional material
Supplementary data supplied by authors.
Footnotes
Supported by the National Cancer Institute. This analysis was supported by the Veterans Health Administration (VHA) Health Services Research and Development Service (IIR 05-101-3). The sponsors played no role in data collection, analysis, interpretation, manuscript preparation, or the decision to submit for publication. At the time this work was performed, Dr. Gould was employed by the U.S. Department of Veterans Affairs. The views expressed in this manuscript are those of the authors and not necessarily those of the National Cancer Institute or the Veterans Health Administration.
Author Contributions: Conception and design, C.C.-D., E.T.C., S.L.G., H.A.W., and M.K.G. Data collection, S.S., X.X., and D.W.W. Data analysis, C.C.-D., S.S., X.X., and M.K.G. Data interpretation and presentation, C.C.-D., S.S., X.X., E.T.C., S.L.G., D.W.W., H.A.W., and M.K.G. Drafting of manuscript, C.C.-D., E.T.C., S.L.G., H.A.W., and M.K.G. Revision of manuscript for important intellectual content, C.C.-D., S.S., X.X., E.T.C., S.L.G., D.W.W., H.A.W., and M.K.G. Final approval of manuscript, C.C.-D., S.S., X.X., E.T.C., S.L.G., D.W.W., H.A.W., and M.K.G.
This article has an online supplement, which is accessible from this issue’s table of contents online at www.atsjournals.org
Author disclosures are available with the text of this article at www.atsjournals.org.
References
- 1.Siegel R, Naishadham D, Jemal A. Cancer statistics, 2012. CA Cancer J Clin. 2012;62:10–29. doi: 10.3322/caac.20138. [DOI] [PubMed] [Google Scholar]
- 2.Parkin DM, Bray F, Ferlay J, Pisani P. Global cancer statistics, 2002. CA Cancer J Clin. 2005;55:74–108. doi: 10.3322/canjclin.55.2.74. [DOI] [PubMed] [Google Scholar]
- 3.Alberg AJ, Brock MV, Samet JM. Epidemiology of lung cancer: looking to the future. J Clin Oncol. 2005;23:3175–3185. doi: 10.1200/JCO.2005.10.462. [DOI] [PubMed] [Google Scholar]
- 4.Bunn PA, Jr, Thatcher N. Systemic treatment for advanced (stage IIIb/IV) non–small cell lung cancer: more treatment options; more things to consider [conclusion] Oncologist. 2008;13:37–46. doi: 10.1634/theoncologist.13-S1-37. [DOI] [PubMed] [Google Scholar]
- 5.Wakelee HA, Chang ET, Gomez SL, Keegan TH, Feskanich D, Clarke CA, Holmberg L, Yong LC, Kolonel LN, Gould MK, et al. Lung cancer incidence in never smokers. J Clin Oncol. 2007;25:472–478. doi: 10.1200/JCO.2006.07.2983. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Scagliotti GV, Longo M, Novello S. Nonsmall cell lung cancer in never smokers. Curr Opin Oncol. 2009;21:99–104. doi: 10.1097/CCO.0b013e328321049e. [DOI] [PubMed] [Google Scholar]
- 7.Subramanian J, Govindan R. Lung cancer in never smokers: a review. J Clin Oncol. 2007;25:561–570. doi: 10.1200/JCO.2006.06.8015. [DOI] [PubMed] [Google Scholar]
- 8.Sun S, Schiller JH, Gazdar AF. Lung cancer in never smokers—a different disease. Nat Rev Cancer. 2007;7:778–790. doi: 10.1038/nrc2190. [DOI] [PubMed] [Google Scholar]
- 9.Toh CK. The changing epidemiology of lung cancer. Methods Mol Biol. 2009;472:397–411. doi: 10.1007/978-1-60327-492-0_19. [DOI] [PubMed] [Google Scholar]
- 10.Bryant A, Cerfolio RJ. Differences in epidemiology, histology, and survival between cigarette smokers and never-smokers who develop non–small cell lung cancer. Chest. 2007;132:185–192. doi: 10.1378/chest.07-0442. [DOI] [PubMed] [Google Scholar]
- 11.Nordquist LT, Simon GR, Cantor A, Alberts WM, Bepler G. Improved survival in never-smokers vs current smokers with primary adenocarcinoma of the lung. Chest. 2004;126:347–351. doi: 10.1378/chest.126.2.347. [DOI] [PubMed] [Google Scholar]
- 12.Subramanian J, Velcheti V, Gao F, Govindan R. Presentation and stage-specific outcomes of lifelong never-smokers with non–small cell lung cancer (NSCLC) J Thorac Oncol. 2007;2:827–830. doi: 10.1097/JTO.0b013e318145af79. [DOI] [PubMed] [Google Scholar]
- 13.Ayanian JZ, Chrischilles EA, Fletcher RH, Fouad MN, Harrington DP, Kahn KL, Kiefe CI, Lipscomb J, Malin JL, Potosky AL, et al. Understanding cancer treatment and outcomes: the Cancer Care Outcomes Research and Surveillance Consortium. J Clin Oncol. 2004;22:2992–2996. doi: 10.1200/JCO.2004.06.020. [DOI] [PubMed] [Google Scholar]
- 14.Catalano PJ, Ayanian JZ, Weeks JC, Kahn KL, Landrum MB, Zaslavsky AM, Lee J, Pendergast J, Harrington DP Cancer Care Outcomes Research Surveillance Consortium. Representativeness of participants in the cancer care outcomes research and surveillance consortium relative to the surveillance, epidemiology, and end results program. Med Care. 2013;51:e9–e15. doi: 10.1097/MLR.0b013e318222a711. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Piccirillo JF, Tierney RM, Costas I, Grove L, Spitznagel EL., Jr Prognostic importance of comorbidity in a hospital-based cancer registry. JAMA. 2004;291:2441–2447. doi: 10.1001/jama.291.20.2441. [DOI] [PubMed] [Google Scholar]
- 16.Little RJ. Statistical analysis with missing data. New York: Wiley; 1986. [Google Scholar]
- 17.He Y, Zaslavsky AM, Landrum MB, Harrington DP, Catalano P. Multiple imputation in a large-scale complex survey: a practical guide. Stat Methods Med Res. 2010;19:653–670. doi: 10.1177/0962280208101273. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Itaya T, Yamaoto N, Ando M, Ebisawa M, Nakamura Y, Murakami H, Asai G, Endo M, Takahashi T. Influence of histological type, smoking history and chemotherapy on survival after first-line therapy in patients with advanced non–small cell lung cancer. Cancer Sci. 2007;98:226–230. doi: 10.1111/j.1349-7006.2006.00379.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Toh CK, Gao F, Lim WT, Leong SS, Fong KW, Yap SP, Hsu AA, Eng P, Koong HN, Thirugnanam A, et al. Never-smokers with lung cancer: epidemiologic evidence of a distinct disease entity. J Clin Oncol. 2006;24:2245–2251. doi: 10.1200/JCO.2005.04.8033. [DOI] [PubMed] [Google Scholar]
- 20.Tsao AS, Liu D, Lee JJ, Spitz M, Hong WK. Smoking affects treatment outcome in patients with advanced nonsmall cell lung cancer. Cancer. 2006;106:2428–2436. doi: 10.1002/cncr.21884. [DOI] [PubMed] [Google Scholar]
- 21.Toh CK, Wong EH, Lim WT, Leong SS, Fong KW, Wee J, Tan EH. The impact of smoking status on the behavior and survival outcome of patients with advanced non–small cell lung cancer: a retrospective analysis. Chest. 2004;126:1750–1756. doi: 10.1378/chest.126.6.1750. [DOI] [PubMed] [Google Scholar]
- 22.Bach PB, Cramer LD, Warren JL, Begg CB. Racial differences in the treatment of early-stage lung cancer. N Engl J Med. 1999;341:1198–1205. doi: 10.1056/NEJM199910143411606. [DOI] [PubMed] [Google Scholar]
- 23.Greenwald HP, Polissar NL, Borgatta EF, McCorkle R, Goodman G. Social factors, treatment, and survival in early-stage non–small cell lung cancer. Am J Public Health. 1998;88:1681–1684. doi: 10.2105/ajph.88.11.1681. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Thun MJ, Hannan LM, Adams-Campbell LL, Boffetta P, Buring JE, Feskanich D, Flanders WD, Jee SH, Katanoda K, Kolonel LN, et al. Lung cancer occurrence in never-smokers: an analysis of 13 cohorts and 22 cancer registry studies. PLoS Med. 2008;5:e185. doi: 10.1371/journal.pmed.0050185. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Wisnivesky JP, McGinn T, Henschke C, Hebert P, Iannuzzi MC, Halm EA. Ethnic disparities in the treatment of stage I non–small cell lung cancer. Am J Respir Crit Care Med. 2005;171:1158–1163. doi: 10.1164/rccm.200411-1475OC. [DOI] [PubMed] [Google Scholar]
- 26.Gomez SL, Chang ET, Shema SJ, Fish K, Sison JD, Reynolds P, Clément-Duchêne C, Wrensch MR, Wiencke JL, Wakelee HA. Survival following non–small cell lung cancer among Asian/Pacific Islander, Latina, and non-Hispanic white women who have never smoked. Cancer Epidemiol Biomarkers Prev. 2011;20:545–554. doi: 10.1158/1055-9965.EPI-10-0965. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Saeed AM, Toonkel R, Glassberg MK, Nguyen D, Hu JJ, Zimmers TA, Robbins DJ, Koniaris LG, Lally BE. The influence of Hispanic ethnicity on nonsmall cell lung cancer histology and patient survival: an analysis of the Survival, Epidemiology, and End Results Database. Cancer. 2012;118:4495–4501. doi: 10.1002/cncr.26686. [DOI] [PubMed] [Google Scholar]
- 28.Ou SH, Ziogas A, Zell JA. Asian ethnicity is a favorable prognostic factor for overall survival in non–small cell lung cancer (NSCLC) and is independent of smoking status. J Thorac Oncol. 2009;4:1083–1093. doi: 10.1097/JTO.0b013e3181b27b15. [DOI] [PubMed] [Google Scholar]
- 29.Nyberg F, Agudo A, Boffetta P, Fortes C, González CA, Pershagen G. A European validation study of smoking and environmental tobacco smoke exposure in nonsmoking lung cancer cases and controls. Cancer Causes Control. 1998;9:173–182. doi: 10.1023/a:1008882227444. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.