|Home | About | Journals | Submit | Contact Us | Français|
Increased physical activity has been associated with decreased lung cancer risk. However, no previous investigation has examined physical activity in relation to lung cancer histologic types by smoking status. The authors investigated these relations in the National Institutes of Health–AARP Diet and Health Study among 501,148 men and women aged 50–71 years at baseline in 1995–1996. During follow-up to 2003, 6,745 lung carcinomas occurred (14.8% small cell, 40.3% adenocarcinoma, 19.7% squamous cell, 6.1% undifferentiated large cell, 7.2% non-small cell not otherwise specified, and 11.8% carcinoma not otherwise specified). Among former smokers, the multivariate relative risks of small cell, adenocarcinoma, squamous cell, and undifferentiated large cell carcinomas comparing the highest with the lowest activity level (≥5 times/week vs. inactive) were 0.93 (95% confidence interval (CI): 0.67, 1.28), 0.79 (95% CI: 0.67, 0.94), 0.73 (95% CI: 0.57, 0.93), and 0.61 (95% CI: 0.38, 0.98), respectively. Among current smokers, corresponding values were 0.77 (95% CI: 0.58, 1.02), 0.76 (95% CI: 0.61, 0.95), 0.85 (95% CI: 0.65, 1.11), and 1.10 (95% CI: 0.69, 1.78). In contrast, physical activity was unrelated to lung carcinoma among never smokers (Pinteraction between physical activity and smoking for total lung carcinomas=0.002). The inverse findings among former and current smokers in combination with the null results for physical activity among never smokers may point toward residual confounding by cigarette smoking as an explanation for the relations observed.
Lung cancer is the leading cause of cancer-related deaths among men and women in the United States, accounting for 28% of cancer deaths in this country (1). Cigarette smoking is the major determinant of lung cancer, accounting for approximately 90% of cases (2). Primary prevention is the most favorable strategy for lung cancer prevention, and smoking cessation remains the most favorable way to prevent lung cancer among smokers (2).
Most (3–13), but not all (14–20), of the numerous studies that have investigated the association between physical activity and lung cancer are consistent with an inverse relation between the 2. Information on whether physical activity differentially affects histologic lung carcinoma types is much more limited, with only 5 available studies on the topic (9–13). One study (10) found an inverse relation of physical activity to small cell and adenocarcinoma but no association with squamous cell carcinoma, and another study (11) noted an inverse association with small cell and squamous cell carcinoma and no relation with large cell and adenocarcinoma. In contrast, 3 studies (9, 12, 13) reported a statistically nonsignificant inverse association with physical activity that did not appear to vary by histologic type.
The relation of smoking to lung cancer differs markedly according to histologic type, with from 4- to 10-fold greater risks from smoking seen for small cell and squamous cell carcinomas than for adenocarcinoma and undifferentiated large cell carcinoma (21). Given differential strengths of the effects of smoking on lung carcinoma histologic types, smoking and physical activity may plausibly interact to influence the occurrence of lung carcinoma histologic types. However, no previous study has investigated physical activity in relation to lung carcinoma histologic types according to smoking status.
In a large cohort of US men and women, we examined the effect of smoking on the relation of physical activity to major lung carcinoma histologic types. Our study differs from previous prospective investigations in being the largest available study on the topic, with 6,745 lung carcinoma cases and several hundred to several thousand cases of each major histologic type.
The National Institutes of Health–AARP Diet and Health Study was established in 1995–1996 when an initial mailed questionnaire on medical history, diet, and physical activity was mailed to 3.5 million AARP (formerly known as the American Association of Retired Persons) members aged 50–71 years and residing in 1 of 6 US states (California, Florida, Louisiana, New Jersey, North Carolina, and Pennsylvania) or 2 metropolitan areas (Atlanta, Georgia, and Detroit, Michigan) (22). A total of 566,402 persons satisfactorily returned the questionnaire and, of these, we excluded individuals with a previous diagnosis of cancer other than nonmelanoma skin cancer (n=52,561), as well as those with missing information on physical activity (n= 5,911) or missing or inconsistent information on smoking habits (n=6,782). The analytical cohort comprised the remaining 501,148 subjects. The study was approved by the Special Studies Institutional Review Board of the US National Cancer Institute.
Study participants were followed up through December 31, 2003, by annual linkage of the cohort to the National Change of Address database maintained by the US Postal Service and its processing of undeliverable mail, various address change update services, and directly from cohort members’ notifications. In addition, vital status was ascertained by annual linkage of the cohort to the Social Security Administration's Death Master File in the United States. Follow-up searches of presumed deaths in the National Death Index Plus provided verification and information on cause of death.
Incident cases of lung carcinoma were identified by probabilistic linkage to the state cancer registries serving our cohort. We recently expanded our cancer registry ascertainment area by 3 states (Texas, Arizona, and Nevada) to capture cancer cases occurring among participants who moved to those states during follow-up. The North American Association of Central Cancer Registries certifies all 11 cancer registries (23). We conducted a validation study comparing registry findings with self-reports and medical records and found that approximately 90% of all cancer cases in our cohort were validly identified by using linkage to cancer registries (24).
The endpoint in the present analysis was first primary incident lung carcinoma. We also investigated the main histologic types of lung carcinoma, defined by anatomic site and histologic code of the International Classification of Diseases for Oncology, Third Edition (ICD-O codes 34.0–34.9) (25). We considered the following (histologic code): small cell (8002, 8041, 8042, 8044, 8045), adenocarcinoma (8140, 8200, 8231, 8250, 8251, 8260, 8290, 8310, 8323, 8430, 8480, 8481, 8490, 8550), squamous cell (8050, 8070, 8071, 8072, 8073, 8074), undifferentiated large cell (8012, 8020, 8021, 8022, 8031, 8032), non-small cell not otherwise specified (8046, 8560), and lung carcinomas not otherwise specified (8010, 8011, 8123, 8562). Non-small cell carcinomas included adenocarcinoma, squamous cell, undifferentiated large cell, and non-small cell not otherwise specified carcinomas.
Cases of fatal lung cancer that had not been diagnosed with incident lung cancer (and for which we hence lacked cancer registry data) contributed person-time to the study up to their date of death, at which point they were censored and not included as cases. In addition to our main analyses, we conducted a separate analysis of physical activity in relation to lung cancer mortality. In that subanalysis, cases of fatal lung cancer contributed person-time to the study up to their date of death.
The baseline questionnaire inquired about physical activity during the previous year, defined as the frequency each week spent at activities that lasted 20 minutes or more and caused either increases in breathing or heart rate or working up a sweat. There were 6 possible response options: never, rarely, 1–3 times per month, 1–2 times per week, 3–4 times per week, and 5 or more times per week. Our physical activity assessment corresponds to the American College of Sports Medicine's physical activity guidelines that recommend at least 20 minutes of continuous vigorous exercise 3 times per week for improving cardiorespiratory fitness (26).
Although our measure of physical activity has not been directly compared with referent instruments, a questionnaire very similar to the one used in our study showed good reliability (percentage agreement=0.76; kappa=0.53) and reasonable validity (percentage agreement=0.71; kappa= 0.40) as assessed by an activity monitor (27).
In a subset of study participants (n=310,105), we collected information on light and moderate to vigorous intensity physical activity. We used those data to assess whether the observed relations with vigorous activity also held for less vigorous forms of activity.
Participants reported if they had smoked more than 100 cigarettes during their lifetime, the number of cigarettes they smoked per day, whether they were currently smoking or had quit smoking, and the number of years since quitting, if applicable. We used that information to create categories of smoking status (never, former, current), smoking intensity (1–10, 11–20, 21–30, 31–40, 41–60, ≥61 cigarettes per day), and time since quitting (≥10 years, 5–9 years, 1–4 years, <1 year).
Cox proportional hazards regression (28) with person-time of follow-up as the time scale was used to estimate relative risks and the corresponding 95% confidence intervals of lung carcinoma. Using age as the time scale yielded similar results. We tested for and found no departures from the proportional hazards assumption. Follow-up time was calculated from the scan date of the baseline questionnaire until the first occurrence of 1 of the following events: diagnosis of lung carcinoma, move out of the registry ascertainment area, death, or the end of follow-up on December 31, 2003.
Participants were divided into 5 categories according to their physical activity level: 0 (inactive), less than 1, 1–2, 3–4, and 5 or more times per week. The group of inactive participants served as the reference group. Tests of linear trend across increasing categories of physical activity were conducted by assigning the mean level of physical activity for categories and treating that term as a single continuous variable. We assessed lung carcinoma risk in 3 models: one adjusting for age and gender; one adjusting for age, gender, smoking status, smoking dose, and time since quitting; and one additionally adjusting for body mass index, race/ethnicity, marital status, family history of any cancer, education, and intakes of fruit and vegetables, red meat, and alcohol. In extensive initial analyses, we had ruled out confounding by numerous variables, including by dietary supplement use. Missing values for covariates were included in the models as a separate category. Risk estimates were calculated for total lung carcinoma and histologic types of lung carcinoma separately.
To examine whether the association between physical activity and risk of lung carcinoma was modified by other potential risk factors for lung carcinoma, we conducted both stratified analyses and formal tests of interaction; the statistical significance of the latter was evaluated by using likelihood-ratio tests. All relative risks are presented with 95% confidence intervals, and reported P values are based on 2-sided tests.
All analyses were conducted by using SAS, release 8.2, software (SAS Institute, Inc., Cary, North Carolina).
The mean ages at baseline and at the end of follow-up were 61.9 (standard deviation (SD), 5.4) and 69.1 (SD, 5.5) years, respectively. At baseline, 18% of the cohort reported not being physically active, and 19% indicated engaging in physical activity 5 or more times per week. Physical activity was positively associated with elements of a healthy lifestyle, including less smoking, lower body mass index, greater dietary intakes of fruit and vegetables, and less consumption of red meat. In addition, active participants were more likely to report a college education and to be married than their less active counterparts (Table 1).
Lung carcinoma risk was inversely associated with body mass index and educational level, and it was suggestively inversely related to intakes of fruit and vegetables. In contrast, lung carcinoma risk was positively associated with smoking, family history of cancer, and intakes of red meat and alcohol (data not tabulated).
During 3,600,331 person-years of follow-up (mean follow-up, 7.2 years; SD, 1.4), we documented 6,745 lung carcinoma cases, of which 14.8% were small cell, 40.3% were adenocarcinoma, 19.7% were squamous cell, 6.1% were undifferentiated large cell, 7.2% were non-small cell not otherwise specified, and 11.8% were lung carcinoma not otherwise specified. The risk of total lung carcinoma decreased in a linear fashion with increasing physical activity level (Table 2). In analyses that were adjusted for age and gender, participants who reported engaging in physical activity 5 or more times per week had a relative risk of 0.50 (95% confidence interval (CI): 0.46, 0.54; Ptrend < 0.001) as compared with their inactive counterparts. After additional control for smoking status (current, former, or never smoking), the inverse association was substantially attenuated (relative risk (RR) = 0.68, 95% CI: 0.63, 0.74; not shown in Table 2). When we further adjusted for the combination of smoking intensity and time since quitting, the relation became slightly weaker, but it remained statistically significant (RR = 0.77, 95% CI: 0.71, 0.83). Additional control for other potential confounding variables had little impact. When we examined physical activity in relation to mortality from lung cancer (n=4,793 cases), the multivariate relative risk was 0.79 (95% CI: 0.72, 0.87).
Undiagnosed lung carcinoma may have caused subjects to report a lower physical activity level at the time the baseline questionnaire was administered, which would bias our results. After we excluded all cases of lung carcinoma that occurred during the first 4 years of follow-up (n=3,260 lung carcinoma cases excluded), results were not materially altered (multivariate RR comparing extreme categories= 0.82, 95% CI: 0.74, 0.92). Findings were also virtually unchanged when we further minimized any impact that undiagnosed lung carcinoma may have had on physical activity levels by additionally excluding subjects who reported poor health at entry (n=3,392 lung carcinoma cases excluded; RR=0.83, 95% CI: 0.74, 0.93). When we repeated our analysis of excluding the first 4 years of follow-up, this time using mortality from lung cancer as an endpoint, the corresponding relative risk was 0.80 (95% CI: 0.71, 0.93).
Using data from a subcohort of study participants for whom we had a separate assessment of physical activity that included information on light and moderate to vigorous physical activity (n=3,836 cases), we found that both light activity (multivariate RR for >7 hours of activity per week vs. no activity=0.85, 95% CI: 0.76, 0.95) and moderate to vigorous activity (multivariate RR for >7 hours of activity per week vs. no activity=0.82, 95% CI: 0.74, 0.90) were inversely related to lung carcinoma.
We next investigated physical activity in relation to histologic types of lung carcinoma (Table 2).As in our analysis of total lung carcinoma, adjustment for smoking accounted for most of the difference between the age- and gender-adjusted models and the multivariate models. Physical activity showed an inverse or suggestively inverse relation with all histologic subtypes. The relative risks for small cell, adenocarcinoma, squamous cell, and undifferentiated large cell carcinomas were 0.82 (95% CI: 0.67, 1.01), 0.80 (95% CI: 0.71, 0.91), 0.78 (95% CI: 0.65, 0.93), and 0.86 (95% CI: 0.62, 1.21), respectively.
When we repeated the histology-specific analyses among cases of fatal lung cancer for which we had both incidence and mortality data, the corresponding relative risks for small cell, adenocarcinoma, squamous cell, and undifferentiated large cell carcinomas were 0.83 (95% CI: 0.66, 1.05), 0.86 (95% CI: 0.73, 1.02), 0.74 (95% CI: 0.59, 0.94), and 0.88 (95% CI: 0.59, 1.32), respectively.
Because lung carcinomas among ever smokers and never smokers may be differentially influenced by physical activity, we conducted additional analyses that were stratified by smoking status (Table 3). Increased physical activity was similarly related to decreased risk of total lung carcinoma among both current and former smokers (Pinteraction = 0.301). In contrast, no relation of physical activity to total lung carcinoma was noted among never smokers, and that null association differed significantly from the inverse relation with physical activity observed among ever smokers (Pinteraction =0.002).
On evaluation of lung carcinoma subtypes by smoking status, we observed a similar pattern of an inverse association with physical activity among current and former smokers and no relation among never smokers for most histologic types, although the difference in the relation of physical activity to lung carcinoma by smoking status (ever vs. never smokers) was statistically significant only for total non-small cell lung carcinoma (Pinteraction =0.006) and, within that group, for adenocarcinoma (Pinteraction = 0.019) (Table 3). For the group of current smokers, the inverse association with physical activity was most apparent for adenocarcinoma.
To evaluate whether the association between physical activity and lung carcinoma was modified by gender, age, race, education, body mass index, history of emphysema, intakes of fruit and vegetables, red meat, and alcohol, and use of nonsteroidal antiinflammatory drugs, we repeated our analyses within subgroups defined by those variables (Table 4). Physical activity was related to decreased lung carcinoma risk in almost all subgroups, suggesting no important effect modification (all Pinteraction > 0.05).
In this prospective study of 501,148 men and women, increased physical activity appeared to be associated with a decrease in the risk of total lung carcinoma of 22%. The association persisted after controlling for established or suspected risk factors for lung carcinoma, including smoking. The inverse relation was apparent for all histologic subtypes of lung carcinomas.
Our findings for total lung carcinoma confirm the majority of previous reports that physical activity is inversely related to risk of lung cancer. Eight prospective (3, 4, 7–10, 12, 13) and 3 case-control (5, 6, 11) studies observed a 23%–42% decreased risk of lung cancer for high versus low levels of total physical activity (7, 8, 10, 12) or certain types of physical activity (4, 13), although inverse relations were limited to subgroup analyses in some studies (3, 4, 9). Risk reduction tended to be more pronounced in studies with less comprehensive adjustment for smoking (5, 7). In contrast, 4 prospective investigations (14, 16–18), 1 retrospective cohort study (19), and 1 case-control study (15) found no association between physical activity and lung cancer. One case-control study (20) observed a positive association of physical activity and lung cancer risk. Inconsistent findings from previous reports may be due to limited sample sizes (3, 4, 7, 8, 14, 17, 18), imprecise assessments of physical activity (15, 20), insufficient variability in physical activity (10), variation in the magnitude of residual confounding by smoking (3–18, 20), or potential recall bias (5, 6, 11, 15, 19, 20).
Reasonably comprehensive information on smoking enabled us to examine the association between physical activity and risk of lung carcinoma according to specific subgroups defined by the combination of smoking status, smoking intensity, and time since quitting smoking. Consistent with most (6, 11, 12), but not all (7), previous studies that presented data stratified by smoking status, our study found no association between physical activity and total lung carcinoma among never smokers. It has been suggested that the etiology of lung cancer among never smokers is distinct from that among smokers (29, 30).
We noted a progressive attenuation of the relation between physical activity and lung carcinoma with increasing control for smoking. In addition, physical activity was inversely associated with lung carcinoma among current and former smokers. Because smoking is associated with both physical activity levels and lung cancer risk and is imperfectly measured, residual confounding by cigarette smoking is a reasonable explanation for the apparently protective effect of physical activity seen in smokers, because one would expect to observe an inverse association between physical activity and lung cancer among both smokers and never smokers if the physical activity and lung cancer relation were causal.
In theory, physical activity may decrease lung carcinoma risk by increasing pulmonary ventilation and perfusion (31–34), thereby reducing the amount of time potential carcinogens, including tobacco-specific nitrosamines, reside in the airways (35, 36). Physical activity may also reduce the risk of lung carcinoma by attenuating a smoking-related decline in lung function (32), which represents a strong predictor of lung cancer (37, 38). This mechanism may be operative even after smoking cessation, although we do not believe that this hypothesis alone can explain the similarity in the relations between physical activity and lung carcinoma that we observed among current and former smokers. In addition, physical activity may protect against lung carcinoma risk by counteracting oxidative DNA damage (39–41) or enhancing DNA repair capacity, both of which play a critical role in lung carcinogenesis (42, 43). The beneficial effect of physical activity on lung carcinoma risk also may be due to its influence on immune destruction of errant cells, because regular physical activity enhances immune function (44, 45).
We observed a similar risk reduction of total lung carcinoma afforded by physical activity in both women and men, among younger and older subjects, for highly educated and less highly educated individuals, for lean and overweight participants, and for those with low and high intakes of fruits and vegetables, red meat, and alcohol. In contrast, some studies have noted that the association between physical activity and lung cancer varies according to gender (3, 4), body mass index (11), or age (3, 9).
An important strength of our study is the substantial size of the cohort, yielding more than 4 times the number of cases than any previous prospective study on the topic. This resulted in relatively more precise risk estimates for physical activity. Subjects with preexisting cancer at baseline were excluded from the analyses in order to reduce the influence that malignant disease may have had on physical activity levels at entry. In secondary analyses, we further minimized the potential for bias due to preexisting but undiagnosed lung carcinoma by excluding the first 4 years of follow-up.
Measurement error in the assessment of physical activity was a potential concern, in particular because our questionnaire has not been directly compared with validation instruments (46). However, a physical activity instrument very similar to ours has demonstrated reasonable validity and reproducibility (27). Moreover, our prospective study design precluded bias attributable to differential recall of physical activity by participants with and without lung carcinoma. The large size of our cohort with the associated costs prohibited us from using more accurate measures of physical activity, such as activity monitors (47). Similarly, we relied on self-reported data from questionnaires for other potential confounding variables such as dietary intake and supplement use, which were also subject to measurement errors.
In conclusion, our data show that increased physical activity is associated with reduced risk of total lung carcinoma and individual lung carcinoma histologic types among current and former smokers. These relations, however, may be due to residual confounding by cigarette smoking. Our study's finding of no association between physical activity and lung carcinoma among never smokers is consistent with a noncausal relation with physical activity. The most important individual and public health strategy for lung cancer prevention is to discourage smoking initiation among adolescents and to advocate for smoking cessation among tobacco users of any age.
Author affiliations: Nutritional Epidemiology Branch, Division of Cancer Epidemiology and Genetics, National Cancer Institute, National Institutes of Health, Bethesda, Maryland (Michael Leitzmann, Corinna Koebnick, Christian C. Abnet, Neal D. Freedman, Yikyang Park, Arthur Schatzkin); Department of Research and Evaluation, Kaiser Permanente Southern California, Pasadena, California (Corinna Koebnick); Cancer Prevention Fellowship Program, Division of Cancer Prevention, National Cancer Institute, National Institutes of Health, Bethesda, Maryland (Neal D. Freedman); AARP, Washington, District of Columbia (Albert Hollenbeck); and Division of Cancer Control and Population Sciences, National Cancer Institute, National Institutes of Health, Bethesda, Maryland (Rachel Ballard-Barbash).
This research was supported by the Intramural Research Program of the National Cancer Institute, National Institutes of Health.
The authors thank Sigurd Hermansen and Kerry Grace Morrissey from Westat for study outcomes ascertainment and management, Leslie Carroll at Information Management Services for data support and analysis, and Tawanda Roy at the Nutritional Epidemiology Branch for research assistance.
Conflict of interest: none declared.