Tobacco Documents Online

I I ! I I I I I I I I I I I 1 I AITRIBUTABLE RISK OF LUNG CANCER IN NONSMOKING WOMEN Michael C.R. Alavania*, Ross C. Brownson**, Jacques Benichou* Christine Swanson* and John D. Boice, Jr.* * Epidemiology and Biostatistics Program, National Cancer Institute, Bethesda, Maryland, USA ** Department of Community Health, Saint Louis University School of Public Health, St. Louis, Missouri, USA Abstract Back rg ound In 1992, approximately 13,000 lung cancers occurred in nonsmoking U.S. women, but the etiology of these cancers is not well understood. Methods A population-based, case-control study of incident lung cancer among nonsmoking women in Missouri was conducted between 1986 and 1992. The study included 618 lung cancer cases and 1402 population-based, age-matched controls. Information on lung cancer risk factors was obtained by personal interview, or next-of-kin interviews (36% and 64% respectively). Year-long radon measurements were also sought in every dwelling occupied for the previous 5-30 years. Population attributable risks (PAR) for specific risk factors were computed for all subjects, for lifetime nonsmokers, for long-term ex-smokers, and by histologic cell type. Results The mean age of lung cancer diagnosis was 71 years, and nearly 50% of the lung cancers were histologically confirmed adenocarcinomas. Almost 40% of all lung cancers among lifetime nonsmokers and almost 50% of lung cancers among all subjects could be explained by the risk factors under study. Dietary intake of saturated fat and nonmalignant lung disease were the two leading identified risk factors for lung cancer among lifetime nonsmokers in Missouri, followed by environmental tobacco smoke, and occupational exposures to known carcinogens. Although an association with domestic radon exposure was not clearly demonstrated, it could be estimated that the PAR is less than about 5%. A similar pattern of risk was identified among former smokers, but in this group the lingering effect of a history of smoking was also very important. Along with saturated fat intake, the combined effect of previous active and passive smoking even after 15 years of active smoking cessation was responsible for more lung cancer than any other risk factor under study. A history of lung cancer among first degree relatives was a risk factor for exsmokers but not for lifetime nonsmokers. I

Conclusion Nearly forty percent of lung cancer cases among lifetime nonsmokers could be prevented if identified diet, occupational and general environmental factors were controlled. Genetic or familial factors seem to be most important to former smokers (and possibly to current active smokers) with little excess risk being seen among lifetime nonsmokers. The etiologic link between some of these factors (i.e., saturated fat and domestic radon) has not been examined in many other studies so a cautious interpretation of the population attributable risks presented for these exposures seems warranted. Introduction Cigarette smoking is by far the major cause of lung cancer, accounting for more than 80% of the 145,000 lung cancer deaths that occur each year in the United States. Lung cancer in nonsmokers, however, is also important and may account for more deaths than any other cancer except colon and breast in women and colon and prostate in men (1). Between June 1, 1986 to April 1, 1991, 19 k of all female lung cancer cases in Missouri occurred among nonsmokers (2). Despite its large public health impact, the etiology of lung cancer among nonsmokers is poorly defined. In this population, we previously determined the risk of various factors for lung cancer in a large, population-based, case-control study of lifetime nonsmokers and former smokers who had ceased smoking for at least 15 years (2-7). Here we present population attributable risk estimates to characterize, to the extent possible, the proportion of lung cancer that might be caused by each of the identified risk factors. Methods Population The study design and methods have been described previously (2-7). Briefly, white nonsmoking women 30-84 years of age who were residents of Missouri between June 1, 1986 and June 1, 1991 were eligible for inclusion. Lifetime nonsmokers consisted of those women who had not smoked more than 100 cigarettes or used any other tobacco products for more than 6 months in their lifetime. Former smokers were defined as women who ceased using all tobacco products 15 or more years prior to interview. Of the 3,475 women with lung cancer reported to the Missouri Cancer Registry, 650 were eligible for this study of whom 618 (95 %) agreed to participate. In addition to the registry-reported diagnosis of lung cancer, tissue slides were reviewed for histologic verification for 468 (76 %) of the cases by a panel of respiratory pathologist (10). A population-based sample of white, nonsmoking women control subjects were selected by frequency-matching on age from driver's license files provided by the Missouri Department of Revenue and for those over age 65, from lists of Missouri women provided by the Health Care Financing Administration (11). A total of 1527 nonsmoking control women responded to the initial screening interview; 1402 (92%) agreed to enroll in the study. Information on residential history, passive smoking exposure, family history, occupation, diet, previous lung disease or prior active smoking history was obtained from a structured questionnaire -2- I I I I I I I I I I I I I I I I I

I i 1 I 1 I I I I I I I I I I i I I administered by a trained telephone interviewer. Next-of-kin interviews were conducted for 64 percent (n=396) of the cases and none of the controls. Current residential radon concentrations were measured by placing two alpha track detectors in each dwelling occupied for at least one year by the study subject during the preceding 30 years in the state of Missouri. One detector was placed in the bedroom and the other in the kitchen for 12 months. Extensive quality control procedures were implemented to assure reliable radon measurements (7). Odds Ratio and Attributable Risk Estimation Unconditional logistic regression was used to estimate adjusted odds ratios. The risk factors under study were saturated fat intake, history of active smoking, previous nonmalignant lung disease, passive smoking, occupational exposure to carcinogens, family history of lung cancer and domestic radon. Each logistic model included the risk factor under study as well as those variables that were associated with a significant increase or decrease in lung cancer risk (2-7). Saturated fat intake was further adjusted to account for the caloric content of the daily diet (5). Namely, age (in five categories, 0-54, 55-64, 65- 74, 75-79, ? 80 years) and daily caloric intake (in five categories defined by quintiles of intake in the controls) were controlled for in all models, while saturated fat intake (in five categories defined by quintiles of intake in the controls), history of smoking (ever/never) and previous nonmalignant lung disease (ever/never) were controlled for in models where they were not already part of the exposure under study. Estimates of populations attributable risks (PARs) were obtained by using an approach based on unconditional logistic regression (8,9). By combining adjusted odds ratio estimates and the observed prevalence of the risk factor under study in the cases, this approach yields adjusted PAR estimates. The same logistic models were used for odds ratio and PAR estimation, therefore allowing one to adjust PAR estimates for the same factors as odds ratio estimates. Since both the odds ratios and the prevalence of exposure affect PARs, they are both tabulated (table 3). For smoking history, nonmalignant lung disease, occupation (use of asbestos, pesticides or working in the dry-cleaning industry), and a family history of lung cancer both the odds ratio and PAR were computed based on the comparison of ever vs. never exposed. For variables such as passive smoking, saturated fat intake and domestic radon, where exposure is ubiquitous, judgements had to be made to define exposure cut points along the exposure continuum that might be achieved as preventive measures in Missouri. For passive smoking the exposed group were women with >40 pack years of smoking from a spouse, while the unexposed group was for women with <40 pack years of exposure. For saturated fat intake which showed a significant monotonic dose-response effect (5) we compared the upper half of the exposure continuum with the lower half, assuming that a dietary modification of this extent might be possible. Finally, for domestic radon exposure, we estimated PAR by defining the exposed group as those subjects with a time-weighted-average (25 years) domestic radon exposure of 4pCi/L or greater (the current EPA action level). Cut points for each of these exposures were associated with a significant excess relative risk of lung cancer in our earlier study (3,5,7). For two variables, a history of nonmalignant lung disease and residential history (for radon exposure), odds ratios and PARs based on in-person interviews only were used because they were -3- I

considered a more accurate source of information (for nonmalignant lung disease) and because they provided an upper limit of risk (for radon exposure). Results Most women in our series developed lung cancer after the age of 70 years, were married, and had completed high school (Table 1). There were few differences between the 618 cases and 1,402 controls in any of the demographic characteristics evaluated. However, the proportion of former smokers (women who had quit smoking more than 15 years previously : median period of cessation=26 years), was about twice as high among lung cancer cases (30 percent) as among controls (17 percent). Pathologic material from 468 cases was available for review. Adenocarcinoma was the most frequent lung cancer cell type (62 percent), followed by squamous cell carcinoma (6 percent), bronchoalveolar adenocarcinoma (4 percent), small cell carcinoma (3 percent), and all other cell types combined (25 percent) (Table 2). Women in the upper half of the saturated fat consumption continuum were at a seventy percent excess risk of lung cancer compared to women in the lower half. This excess relative risk translates into a population attributable risk of approximately 22 % since the exposed population in this case, constitutes 50% of the total population. We estimate that reducing the saturated fat intake below the 50th percentile (i.e., in this study estimated to be 18.8 grams/day) would be the single most effective action identified to reduce lung cancer incidence in a nonsmoking female population in Missouri (Table 3). Further reducing the saturated fat consumption to below the 20th percentile would reduce the risk of lung cancer even more, the PAR for saturated fat consumption above the 20th percentile being 48% (not shown in Table (3)(5). Both life-long nonsmokers and long term ex-smokers achieved a similar degree of benefit from a reduction in saturated fat intake. Fruit and/or vegetable consumption, which has been found to have a beneficial effect of reduced lung cancer incidence in some smoking and nonsmoking populations (12), did not have a measurable impact on lung cancer risk in this study. The population attributable risk of saturated fat intake was slightly higher among nonadenocarcinoma cell types (25 %) than adenocarcinoma (19%). The picture of risk seems to change, however, when more extreme saturated fat intakes are compared. The relative risk of lung adenocarcinoma was much greater than the risk for nonadenocarcinoma when extreme quintiles of intake of saturated fat are examined (5). Even after 15 years of smoking cessation, former smokers were at over twice the risk of lung cancer (OR=2.3) as were lifelong nonsmokers. This lingering risk to former smokers accounted for approximately 17% of all lung cancers in this population (Table 3). If all ex-smokers (including those who quit smoking 1-15 years) were included in this study the percent of risk attributed to a history of smoking would have increased substantially. Prior active smoking was associated with 22% of the nonadenocarcinoma compared to 13 % of adenocarcinoma. Based on in-person interviews only, a history of nonmalignant lung disease such as pneumonia, asthma and tuberculosis was associated with a significant excess lung cancer relative risk of 50% overall and in lifetime nonsmokers, but only 30% in long-term ex-smokers in our study. This was slightly more than when next-of-kin interviews were also included. Nonmalignant lung disease occurred in over one- third of the women in our control group and was associated with 16% of all lung cancer among -4- I I 11 I I I I I I I I I I I

I I t I I I I I I I I I I I nonsmoking women in Missouri. Little difference in risk was experienced between long-term ex-smokers and lifetime nonsmokers, or between adenocarcinoma and other cell types. Exposure to environmental tobacco smoke (ETS) (>_40 pack-years) from a smoking spouse was experienced by one-fifth of all women in our study. The thirty percent excess relative risk among these women was responsible for approximately 6% of all lung cancers in this population (Table 3). This number rose to 8% in lifetime nonsmokers. Other sources of ETS might increase the population attributable risk even further but the Missouri Women Health Study was unable to assess the effect of ETS in most public places. A small additional increment of risk might be expected if a more comprehensive assessment of ETS related risk could be made. Ten percent of all nonadenocarcinoma cases could be attributed to spousal sources of ETS while only about 1% of the adenocarcinoma cases could be attributed to ETS. The combined effect of previous active smoking and passive smoking was responsible for 22 % of lung cancer in this population, and the figure rose to 30% for nonadenocarcinoma cell types. Working with asbestos or pesticides or in dry-cleaning facilities was associated with a moderate excess risk of lung cancer (OR=2.0). However, since exposure to these substances or workplace environments was uncommon in Missouri (approximately 5% of the female population) it was responsible for only about 5% of all lung cancer among nonsmokers. Both adenocarcinoma and nonadenocarcinoma cases were equally affected by these occupational factors. A family history of lung cancer among first degree relatives resulted in a small increased risk of lung cancer (RR = 1.4). Approximately 10% of the controls in our study population had such a history resulting in a population attributable risk of 4%. It should be noted, however, that the risk was not uniformly distributed, rather most of the risk was associated with former smokers (OR=3.9, not shown in table) and no excess risk was observed among lifetime nonsmokers (OR= 1.0, not shown in Table 3). A family history of lung cancer was about equally common in both adenocarcinoma and nonadenocarcinoma cases. Only 6% of the women in Missouri had a history of radon exposure exceeding 4pCi/L that spanned a 25-year period. This pattern of radon exposure is similar to that observed in the United States as a whole (13). In Missouri the mean radon level found in homes was 1.6pCi/L. In our study interviewing living cases resulted in a slightly more elevated estimated risk of lung cancer associated with domestic radon exposure than did interviewing next-of-kin. The reason for this discrepancy is unclear but we based our attributable risk computation on the experience of cases who were interviewed while still alive. This decision resulted in a larger radon associated lung cancer risk. For those living in dwellings with over a 4pCi/1 exposure the excess risk was 60%, resulting in an (nonsignificantly elevated) attributable risk of 4% in nonsmoking Missouri women, with little difference in risk found between lifetime nonsmokers and long-term ex-smokers. Seven percent of adenocarcinoma cases were associated with radon exposure but no excess risk was found among nonadenocarcinoma cases. N 0 00 ~ -4 00 -5- W ~ W I

Discussion Overall, 48% of all lung cancers among current nonsmokers could be attributed to a history of smoking, saturated fat intake, nonmalignant lung disease, environmental tobacco smoke, occupational exposures especially to asbestos, pesticides or dry-cleaning environments, and a family history of lung cancer. In Missouri domestic radon exposure in excess of the EPA action level was associated with a small nonsignificant additional risk of lung cancer. For lifetime nonsmokers 36% of all lung cancer among nonsmokers could be attributed to these nonsmoking risk factors. The amount of evidence from other studies supporting the association between these factors and lung cancer varies greatly and thus cautious interpretation is warranted. The strongest etiologic links identified involved a history of active smoking (14), and occupational exposures to carcinogens such as asbestos (15), while causal relationships are strongly suspected for environmental tobacco smoke (16,17), and a family history of lung cancer (18). Evidence from other studies supporting the etiologic association of saturated fat intake (19,20) and domestic radon exposure (i.e., z4pCi/L)(21-24), on the other hand, is not yet adequate and is in need of additional investigations. Strengths and Weaknesses The major strengths of our investigation include the evaluation of incident cases of lung cancer in a population-based setting, the relatively large number of nonsmoking women available for study and the comprehensive effort to ascertain domestic radon measurements in homes occupied by the study subjects during a 30-year period prior to enrollment in the study. Finally, we conducted a pathology review of cases, which enhances our histologic-specific findings. The potential weaknesses of this study included the use of self-reported data on previous lung disease, family history of lung cancer, passive smoking, diet and a history of active smoking. Moreover, we had no information on exposure to ambient air pollution which has been associated with lung cancer in certain industrial urban centers. Although we could not eliminate these potential weaknesses from the current study, a second interview conducted in a sample of cases and controls suggested that the reporting of nonmalignant lung disease and smoking was highly reproducible (25). Although air pollution is likely to be an independent risk factor for lung cancer (26), it is not likely to seriously confound the results reported in this paper. Conclusion Cessation of cigarette smoking remains the most constructive action to reduce the occurrence of many serious chronic diseases, including lung cancer. Even among long-tetm former smokers, 17% of their lung cancers could be attributable with some confidence to their prior habit. Smoke inhaled involuntarily by a nonsmoking spouse also could account for nearly 7% of lung cancers. In contrast, other exposures among nonsmoking women appear less important, such as occupation and domestic radon. Occupational risks are low because women of this generation were unlikely to work in hazardous jobs with toxic exposures. This will likely change in the future as more employment opportunities have opened for women for most occupations. While radon exposures in underground mines are clearly carcinogenic (20), the picture is not as clear for domestic radon (21-23). Making the most liberal assumptions in our data about possible radon risks, however, it is estimated that the PAR is likely less than 5 45. This percentage is much lower than that estimated by extrapolation of risks from underground -6- I I I I I I I

I I I I I I I miners for which the attributable risk for radon-related lung cancer among nonsmokers would be about 12% based on a multiplicative model and over 30% based on a submultiplicative model between radon and smoking (27). Consumption of high levels of saturated fat and a history of prior lung diseases, especially pneumonia, were major contributors to population risk in this series. The etiologic link between saturated fat and lung cancer has been explained in only a few other studies so that a cautious interpretation of the high PAR seems warranted. Nonetheless, it seems prudent to assume that dietary factors could contribute to lung cancer risk, as they do other chronic diseases such as coronary heart disease, and thus a person should strive to reduce saturated fat and increase fruit and vegetable in their diets. -7- N O O V O W O ~ b

I References 1. Schneiderman, MA; Davis, DL and Wagener, DK. Lung cancer that is not attributable to smoking. Letter. JAMA 1989;261:2635-6. 2. Alavanja, MCR; Brownson, RC; Boice, JD and Hock, E. Preexisting lung disease and lung cancer among nonsmoking women. Amer. J. E i~demiol. 1992;136:623-632. 3. Brownson, RC; Alavanja, MCR; Hock, Et and Loy, TS. Passive smoking and lung cancer in nonsmoking women. Amer. J. Public Health 1992;82:1525-1530. 4. Brownson, RC, Alavanja, MCR and Chang, JC. Occupational risk factors for lung cancer among nonsmoking women: a case-control study in Missouri (United States). Cancer Causes Control 1993;4:449-454. 5. Alavanja, MCR; Brown, CC; Swanson, C and Brownson, RC. Saturated fat intake and lung cancer risk among nonsmoking women in Missouri. J. Natl. Cancer Inst. 1993;85:1906-1916. 6. Brownson, RC; Alavanja, MCR; Berger, E and Chang, JC. Family history of cancer risk of lung cancer among nonsmoking women in Missouri. Amer. Journ. Epidemiol. (In review). 7. Alavanja, MCR; Brownson, RC; Lubin, JH; Brown, C; Berger, E and Boice, JD. Residential radon exposure and lung cancer among nonsmoking women. J. Natl. Cancer Inst. (In Press). 8. Bruzzi, P; Green, SB; Byar, DP; Brinton, LA and Schairer, C. Estimating the population attributable risk for multiple risk factors using case-control data. Amer. J. Epidemiol. 1985; 122:904-914. 9. Benichou, J and Gail, MH. Variance calculations and confidence intervals for estimates of the attributable risk based on logistic models. Biometrics 1990;46:991-1003. 10. Brownson, RC; Loy, TS; Ingram, E; Myers, JL; Alavanja, MCR; Sharp, DJ and Chang, JC. Histologic types of lung cancer among nonsmoking women: pathologic review and survival patterns. Cancer (In review). 11. Martin, G; Alavanja, MCR and Zahm, SH. Department of Health and Human Services epidemiology research 1989 data users conference proceedings. Baltimore, MD: Health Care Finance Administration, 1989:181-186. (HCFA publication no. 03293). 12. Steinmetz, KA and Potter, JD. Vegetables, fruit and cancer: Epidemiology. Cancer Causes Control 1991;2:325-357. 13. Nero, AV; Schwehr, MB; Nazaroff, WN and Revzan, KL. Distribution of airborne radon-222 concentrations in U.S. homes. cience 1986;234:992-997. ~ I I I I N O I tp ' ~ -4 tp W 8- ~ ~ ~

l ~ ~ 14. U.S. Department of Health and Human Services. Reducing the Health Consequence of Smoking:  25 Years of Progress. A Report of the Surgeon General. U.S. Department of Health and Human Services, Public Health Service, Centers for Disease Control, Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health, DHHS Publication No. (CDC) 89-8411, 1989. I 15 World Health Organization, Overall Evaluation of Carcinogenicity: An Updating of IARC . Monograph. Volume i to 42, Supplement 7 International Agency for Research on Cancer, Lyon, 1987. I 16. Respiratory Health Effects of Passive Smoking: Lung Cancer and Other Disorders U.S. I 17. Environmental Protection Agency Washington, DC, 1992. Fontham, ETH; Correa, P; Wu-Williams, A et al. Lung cancer in nonsmoking women: a multicenter case-control study. Cancer Enidemiol. Biomarkers Prev. 1991;1:35-43. 18. Ernster, VL. The epidemiology of lung cancer in women. Annals of Epidemiology 1994;4:102- 110. I 19. Byers, TE; Graham, S; Haughey, BP et al. Diet and lung cancer risk: Findings from the Western Diet Study. Amer. J. Enidemiol. 1967;125:351-363. I 20. Jain, M; Burch, JD; Howe, GR et al. Dietary factors and the risk of lung cancer: results from a case-control study, Toronto, 1981-1958. hu. J. Cancer 1990;45:287-293. I 21. National Research Council. Health risks of radon and other internally deposited alpha-emitters. I 22. BEIR IV. Washington, DC: National Academy Press, 1988. Pershagen, G; Akerblom, G; Axelson, 0; Clavensjo, B; Damber, L; Desai, G; Enflo, A; Lagarde, F; Mellander, H; Svartengren, M and Swedjemark, GA. Residential radon exposure and lung cancer in Sweden. N. Enyl. J. Med. 1994;330:159-64. I 23 et al. Indoor radon and lung cancer in China. JNCI JD Jr ZY; Boice Xu Blot WJ . . , , , ; 1990;82:1025-30 ! . 24. Schoenberg, JB; Klotz, JB; Wilcox, HB et al. Case-control study of residential radon and lung cancer among New Jersey women. Cancer Research 1990;50:6520-24. I 25. Brownson, RC; Alavanja, MCR and Hock, E. Reliability of passive smoke exposure histories I 26. in a case-control study of lung cancer. Int. J. of EQdemiol. 1993;22:804-08. Dockery, DW; Pope, A; Xiping, X; Spengler, JD; Ware, JH; Fay, ME; Ferris, BG and An association between air pollution and mortality in six U.S. cities. N. E FE iezer S N O 00 . , p Med. 1993;329:1753-59. s V I I 9- OD W O A V I

27. Lubin, JH; Boice, JD Jr.; Hornung, RW; Edling, C; Howe, G; Kunz, E; Jusiak, RA; Morrison, HI; Radford, EP; Samet, JM; Tirtnarche, M; Woodward, A; Ziang, YS and Pierce, DA. Radon and lung cancer risk: A joint analysis of 11 underground studies. NII-I Publication No. 94-3644, 1994. N O O ~ -4 O W O -10- ~ I I I I I I Z I I I