Tobacco Documents Online

Cancer Causes and Control 10: 475-482, 1999. ' © 1999 Klumer i4ademic Publishers. Primed in the nuuc..artds. Prospective study of smoking, antioxidant intake, and lung cancer in middle-aged women (USA) 475 F.E. Speizerl•"`, G.A. Colditz', D.J. Hunter', B. Rosner' & C. Hennekens' 'Channing Laboratory, Department of Medicine. Brigham and Women's Hospital, Harvard Medical School, and the Departments of Nutrition and Epidemiology, Harvard School of Public Health, Boston, M.4, USA Received 1I August 1998; accepted in revised form 25 May 1999 Key words: cigarette smoking, dietary antioxidants, lung cancer. Abstract Background: Although substantial evidence suggests that higher intake of fruits and vegetables can reduce the adverse impact of smoking on lung cancer risk, great uncertainty exists regarding the specific foods and their constituents that are protective. We therefore examine prospectively the relation between cigarette smoking and lung cancer incidence among women, and quantify the associations between dietary antioxidants, other nutrients, and lung cancer risk. Methods: In a 16-year prospective cohort study (the Nurses' Health Study), 593 cases of lung cancer were confirmed during 1,793,327 person-years of follow-up. Dietary data, including vitamin supplement use and food intake, were collected in 1980 using a validated semiquantitative food frequency questionnaire. Results: The risk of lung cancer increased with the number of cigarettes smoked and with early onset of cigarette smoking. The risk decreased rapidly with the discontinuation of smoking but took 15 years to fall to about the level of risk for women who had never smoked. Dietary intake of fat was not related to the risk of lung cancer. Although Q-carotene intake was not related to risk, intake of carrots showed a strong inverse relation:: women who reported consuming five or more carrots per week had a relative risk of 0.4 (95% CI = 0.2-0.8) compared with the risk for women who never ate carrots. Conclusions: Smoking is the most important risk factor for lung cancer in women, as it is in men. Higher vegetable consumption, particularly of carrots, may significantly reduce the risk of lung cancer, but dietary modification cannot be considered a substitute for smoking prevention and cessation. Introduction ~' Although the overwhelming _ positive association of cigarette smoking with lung cancer has been recognized for over 40 years, lung cancer remains the most common fatal cancer in the United States. Since 1990, lung cancer killed more women each year than breast cancer. While prospective data among women are limited, the follow-up of the American Cancer Society CPS-I and II and the cohort of Kaiser Permanente ' Correspondence to: Frank E. Speizer, MD, Co-Director, Chan- ning Laboratory, 181 Longwood Avenue, Boston, MA 02115, USA; representing the Nurses' Health Study Research Group. e-mail to frankspeizer(a3channing.harvard.edu. participants show strong relations between smoking and lung cancer [I, 21. There is no question that cigarette smoking is the major cause of lung cancer in both men and women. National statistics indicate that the gender- related difference in rates of lung cancer in the United States is explained, in large part, by the diflerent times at which men and women took up smoking [31. Generally, smoking became prevalent among women in the United States in the mid-20th century; men began smoking cigarettes in large numbers some 25 years earlier. Contemporary data from large prospective studies are important in tracking the epidemic as it evolves. More recently, risk factors for lung cancer other than active smoking have been considered. These risk factors include occupational exposures, family history of lung

476 disease [4] and environmental exposures (e.g., air pollu- tion and passive smoking) [5, 6]. An excess risk of lung cancer has been associated with the level of intake of saturated fat [7]. Carotenoids have been hypothesized to have anticar- cinogenic activity, in part due to their antioxidant properties [8, 91. Dietary intake of carotenoids, calcu- lated from fruit and vegetable consumption, and (f-carotene levels in prospectively collected blood spec- imens have been inversely associated with risk of subsequent lung cancer in a number of studies [10]. However, the lack of any beneficial effect of high doses of /3-carotene supplements in three large randomized trials [ 11-13], and indications of actual adverse effects on lung cancer incidence in two of these studies, has strongly suggested that (i-carotene may not be the protective factor in fruits, and vegetables. Alterna- tively, (i-carotene may not be protective at the ranges studied, or Q-carotene may require the presence of other factors also found in fruits and vegetables. Recent use of data in epidemiologic studies on specific carotenoids suggests a-carotene is at least as protective as /3-carotene in relation to risk of lung cancer [14, 151. Because of the relative lack of studies of women, one of the early specific aims in the Nurses' Health Study was to quantify the impact of smoking on the occur- rence of lung cancer. Since the population was relatively young at the beginning of the study, it was anticipated that few cases of lung cancer would develop in the first 10 years. During that period, additional exposure data were obtained that allowed us to examine a number of questions related to the effects of smoking and passive smoking and potential modifying risk factors. Previous analyses of the health effects of tobacco use in this cohort have addressed cardiovascular diseases [1(r18] and all-cause mortality [191. In this report from the Nurses' Health Study, based on 16 years of follow-up, we document the impact of smoking on the incidence of lung cancer in this large cohort of middle-aged women. In addition, we used dietary data collected initially in 1980 to assess the relation with adjustment for smoking, of both antioxidants and of dietary fat and specific foods on risk of lung cancer over the 12-year period from June 1980 to June 1992. Methods The Nurses' Health Study cohort was established in 1976, when 121,700 female US registered nurses between the ages of 30 and 55 years responded to a mailed questionnaire that inquired about risk factors for cancer and heart disease, with a specific focus on reproductive F.E. Speizer et aL history, menopausal status, contraceptive practices, hormone use, cigarette smoking, and use of permanent hair dyes. The details of the establishment of the cohort have been reported elsewhere [20]. Follow-up question- naires every 2 years have updated information on exposure status and on the occurrence of a number of medical conditions, including lung cancer. Repeated questionnaires are sent to non-respondents. Attempts are made to contact persistent non-responders by telephone. Deaths in the cohort are reported by family members, or the postal service, or identified by a search of the National Death Index. We estimate that mortality ascertainment is 98% complete [21, 22]. Sntoking history At baseline we asked each participant the age at which she started smoking, the age at which she stopped, whether she was a current smoker and the aver number of cigarettes she smoked per day, From onward we also-collected information on the specific brand of cigarettes smoked, which made it possible to estimate the nicotine and tar content of cigarettes smoked by current smokers. Although this information has been updated every 2 years, in the analyses presented we used the cigarette brand reported in 1978, as it best reflects long-term tar exposure. In 1982 we collected information about environmental tobacco- smoke exposure in both childhood and adulthood. Questions asked included "Did your parents smoke while you were living with them?" and "As an adult, how many years have you lived with someone who smoked regularly?" and "Are you currently exposed to cigarette smoke from other people?" n ldentifieation of lung cancer cases For each case of lung cancer reported by participants ^• identified on their death certificates, efforts were made:~~ obtain hospital records and pathology reports. Physi- cians reviewed these records (see below), and, to the extent that it was possible, were not aware of the details of smoking history. Cases were considered confirmed only if a pathology report indicated that the lesion was a primary lung tumor. These cases were subsequently classified by predominant cell type. No case was classified as confirmed on the basis of death certificate data only. Detailed dietary assessments In 1980 we collected dietary information on a semi- quantitative food-frequency questionnaire (SFFQ), for which several validation and reproducibility studies 2505586062

Smoking, antioxidant intake and lung cancer have been reported [23, 24]. In brief, on the initial SFFQ we collected information on the intake of typical portions of 61 common food items; the women recorded the frequency that each item was consumed over the past year using nine categories ranging from never to six or more times per day. The women were also asked to indicate whether their consumption of the specific item had changed significantly over the previous 10 years. Nutrients were calculated by using the products of the frequency of use of each food and its nutrient content, primarily using USDA food composition sources. "Car- otene" intake was calculated to reflect all vitamin A activity in plant foods, as well as a portion of vitamin A activity in dairy products that contain both carotenoids and retinol (e.g., milk). This measure does not differen- tiate among the various carotenoids, nor is it a summary measure of all carotenoids. Recently compiled analyses have allowed us to calculate the intake of specific carotenoids [25]. Energy-adjusted nutrient levels were determined by regression analysis and were_categorized into quintiles of intake (for details see Willett and Stampfer 126]). For the nutrient analyses, 89,284 women without a prior history of cancer (other than non-melanoma skin cancer) completed SFFQ questionnaires in 1980; in the subsequent 12 years there were-399 confirmed cases of lung cancer. The age and smoking effects in this subgroup were similar to those for the entire cohort in which 593 confirmed incident cases occurred between 1976 and 1992. Data analysis For each exposure of interest, person-years were accu- mulated and incidence rates were calculated by dividing the number of events in that exposure category by the number of person-years of follow-up in the same exposure category. For most of the smoking compari- sons the relative risk was calculated by dividing the incidence rate in the particular exposure category by the rate among never smokers. In an analysis of the effect of smoking cessation, current smokers were used as the referent group, and age-adjusted rates (5-year category) were calculated. Multivariate logistic models were used 477 Results Smoking and lung cancer Among 118,351 women who were free from cancer (except non-melanoma skin cancer) in 1976, 593 con- firmed cases of lung cancer were diagnosed during 1,793,089 person-years of follow-up, up to June 1992. As expected, rates increased strikingly from no cases among women 30-34 years old to 144 cases/100,000 person-years among women 65-69 years of age at the time of diagnosis (Table 1). Similar trends with increas- ing age were noted for adenocarcinomas and for other cell types (squamous and small cell). Current smoking status in 1976 and amount smoked were strong predictors of lung cancer risk over the subsequent 16 years (Figure 1). With adjustment for age, women who reported smoking more than 20 cigarettes per day in 1976 had approximately a 20-fold greater risk of lung cancer than never-smokers. The risk increased exponentially with increasing amount smoked up to two packs per day (p trend < 0.0001). We assigned tar values to the brand of cigarettes reported in 1978 by each current smoker, using the brand-specific tables provided by the US Federal Trade Commission [27]. The assigned tar-content values were divided into tertiles. Although the 1978 level of tar appeared to relate to lung cancer risk with adjustment for age and age first smoked (RR for top vs. bottom quartile = 2.0 [L5-2.8]), the effect of tar content was no longer significant. after additional adjustment for current number of cigarettes smoked (RR = 1.0 [0.7-1.4]). When the effect of stopping smoking was compared withh that of continuing to smoke, and an adjustment was made for age, a significant trend toward reduction Table I. Age-specific incidence of adenorarcinoma of the lung, other lung cancer, and total lung cancer diagnosed in the Nurses' Health Study from 1976 to 1992 Age Person- Total Adenocarcinoma Othercelltypes" (years) years Cases° Ratese Cases Rates° Cases Ratcs° to control simultaneously for age and smoking in 30-34 77.394 0 - 0 - 0 - assessments of dietary variables Neither diet nor 35-39 188,389 7 3.7 4 2.1 2 1.1 . 40-44 306,286 28 9.1 t t 3-6 13 4.2 smoking was updated: In analyses of time since quitting 45-49 362,508 66 18.2 33 4l 30 8.3 smoking and duration of use of vitamins, the data were 50-54 364,364 116 31.8 51 t4.0 60 16.5 updated every 2 years. 55-59 280,582 158 56.3 58 20.7 94 33.5 For all variables of interest we calculated 95% 60-64 162,243 144 88.8 58 35.7 81 49.9 confidence intervals. When dietary variables were as- 65-69 51,323 74 144.2 27 516 46 89.6 N sessed, we tested for trend across quintiles of intake, using the midpoint of each quintile in the trend test. ° Cell type missing for 25 cases. ^ Rate per 100,000 person-years. O M M M 0) O w

478 - L_ ~ 5-14 1s-24 2581 Cigarettes per Day Fig. 1. Relative risk of lung cancer, by number of cigarettes smoked per day. of relative risk of lung cancer with increasing number of years since stopping was observed (Table 2). Although the effect of smoking appeared to persist for a number of years after stopping, quitters had a relative risk approx- imately 50% lower than the figure for continuing smokers (with adjustment for age) after 2-5 years and a relative risk approaching that of non-smokers after 10-14 years (Table 2). In 1976 we asked about the age at which women had started smoking. After adjustment for number of cigarettes per day and current age, the effect of starting to smoke at an age - 17 years was generally greater than that of starting at age 22 or older. Compared to women who started smoking at ages 18-19, the relative risk was 1.1 (0.9-1.5) for women starting to smoke before age 18, and 0.8 (0.6-1.1) for those starting to smoke after age 21. In 1982 we inquired about environmental-tobacco smoke exposure in both childhood and adulthood. We documented a total of 58 cases of lung cancer among never smokers. Thirteen of these cases were diagnosed E.E. Syeizer et aiL before 1982. In another 10 cases, information on environmental tobacco smoke (which was not collected until 1982) was missing. Among the remaining 35 cases we found only two with no environmental tobacco- smoke exposure as adults. The age-adjusted relative risk for passive smoke exposure in adulthood was 1.5 (95% CI 0.3-6.3). Dict and lung cancer We found no significant association between energy- adjusted total dietary fat or intake of type of fat and risk of lung cancer (Table 3). If anything, we found a trend toward an inverse association with increasing intake of vegetable fat and linoleic acid. Risk of lung cancer was slightly higher among women with a greater cholesterol intake, and the test for trend was nearly significant. Results remained unchanged when we repeated the analysis without control for total energy intake. For total carotene, vitamins C and E, and folate (including supplements), we found modest inverse asso- ciations with lung cancer risk in the age- and energy- adjusted models. However, these associations were essentially nullified by inclusion in the model of the current number of cigarettes smoked and the age at starting to smoke (Table 4). Examining specific carotenoids, we noted a significant association for a-carotene compared with the lowest quintile, the highest quintile of intake had an inverse association of 0.6 (CI 0.4-0.8; p trend = 0.003) (Table 5). In contrast, the association for specific /i-carotene was close to null, relative risk = 0.8 (CI 0.6-1.1; y trend = 0.20). We also examined the duration of vitamin supplement use, Compared with women who never had taken a vitamin E supplement, women using a vitamin F supplement for 10 or more years had an adjusted I Table 2. Time since quitting smoking in relation to risk of lung cancer in 30- to 55-year-old women in the Nurses' Health Study !m Smoking status No. of cases Relative risk' Relative risk-' Current smoker 391 489,993 1.0 1.0 Years since last smoking <2 24 55,232 06 (0.4-0.9) 0.4 (0.2-0.7) 2-4.9 34 63,060 0.6 (0.440.8) 0.6 (0.4-1.0) 5-9.9 41 95,585 0.50 (0.4--0.7) 0.6 (0.4-0.9) t0-14.9 17 93,933 0.2 (O.t-04) 0.1 (0.1-0.3) .t5+ 28 214,271 0.1 (0_1-0.4) 0-1 (0_1-02) . Never smoker 58 776,300 0.1 (0.1-0.1) ' Adjusted for age and time period. •• Adjusted for age, 2 year follow-up interval, and age started smoking. ' Missing data for 0 cases during 4,947 person-ycars of follow-up.

Smoking, antioxidant intake and lung cancer Table 3. Energy-adjusted dietary fat and relative risk of lung cancer in US women followed from 1980 to 1992 Type of Pat Quintile of fat intake Q5 vs QI (95% CI) Test for trend value 1 2 3 4 5 p Total 1.0 1.0 0.9 0.9 1.1 (0.8-L4) 0.78 Vegetable 1.0 0.8 0.8 0.9 0.8 (0.6-1.1) 0.28 Animal 1.0 1.0 1.1 1.1 Ll (0.8-1.5) 0.37 Saturated 1.0 1.0 1.l 1.1 1.1 (0-8-1.5) 0.43 Linoleic 1.0 0.9 0.9 1.0 0.8 (0.6-1.1) 0.32 Oleic 1.0 1.1 L0 1.0 1.2 (0.9-1.6) 0.44 Trans lA 0.9 0.9 0.9 1.0 (0.7-1.3) 0.32 Cholesterol I.0 1.0 1.3 1.3 1.3 (0.9-1.7) 0.06 Adjusted for age, total energy intake, smoking (past, current amount in 1980; 1-4, 5-14, 15-24, 25-34, 35-44,-45+) and age of starting to smoke (< 17, 18-I9, 20-21, 22+). relative risk of 0.7 (CI 0A-1.4); for 15+ years of vitamin C supplement use the relative risk was 0.7 (CI 0.3-1.6); and for 5 or more years of vitamin A supplement use the relative risk was. 1.0 (CI 0.5-2.2). Among foods with known high carotene content (e.g., sweet potatoes, carrots, broccoli, and other green vegetables), only carrots and broccoli had a significant inverse association after adjustment for smoking vari- ables (Table 6). Compared with women who "never" or "almost never" consumed carrots (Table 6) (and after adjustment for cigarette smoking), those consuming two to four servings per week had a relative risk of 0.6 (CI 479 0.4-L0), and those consuming five or more servings per week had a relative risk of 0.4 (Cl 0.2-0.8; p trend = 0.003). For broccoli the corresponding fig- ures were 0.9 (Cl 0.6-1.3) and 0.2 (Cl 0.0-0.7; p trend = 0.03). The association for broccoli became non-significant when carrots were included in the model. We repeated these analyses, excluding 10,000 women whose carrot consumption had changed in the 10 years prior to 1980. The monotonic decrease in risk with increasing carrot consumption became somewhat stronger. Compared with women who never consumed carrots, women with stable consumption of five or more servings of carrots per week over the previous 10 years had a smoking-adjusted relative risk of 0.3 (CI 0.2-0.7). When we repeated the analyses updating cigarette smoking, the association with lung cancer remained direct and strong. Similarly, when analyses of individual fruits and vegetables were conducted using updated average intake, the magnitude of associations was not materially different from those using only baseline data. Discussion The risks of lung cancer associated with smoking in this large cohort of middle-aged women are consistent with results that have been obtained on numerous occasions. Smoking of more cigarettes per day and longer duration of smoking (as calculated from age at starting and number of years smoking) have consistently been linked Table 4. Relative risk.of lung cancer in Nurses' Health Study cohort, 1980-1992, according to quintile for intake of vitamin C, folate, vitamin E, and carotene ;I Micronutrient Quintile for intake Trend p value I (low) 2 3 4 5 Vitamin C' (ing/day) <93 93-131 132-182 183-358 > 359 Multivariate 1.0 0.92 1.15 0.6 1.35 0.03 (95% CI) - (07-1.2) (0.9-1.5) (0.4-0.9) (L.0-1.8) Carotene° <3600 5336 7645 11268 > 11268 Multivariate 1.0 1.2 0.95 0.94 0.9 0.30 (95 % CI) - (0.9-1.6) (0.7-1.3) (0_7-1.3) (0 7-1.31 Vitamin E' (IU/day) <3.9 3.9-4.9 5.0-7.3 7.4-24.0 >-24.1 Multivariate 1.0 1.1 0.7 0.91 1.2 0.10 (95% Cl) - (0.8-L5) (0.5-1.0) (0.7-1.2) (0.9-L6 Folate" (Ftg/day) < 173 173-237 237-321 322-550 > 550 Multivariate 1.0 1.1 0-9 0.7 LI 0.58 (95% CI) (0.8-L4) (0.7-1.2) (0.5-0.9) (0.8-1.5) ' Including supplements. . ° Carotene index - reflects all vitamin A activity in plant foods as well as a portion of the vitamin A activity from foods known to contain both carotenoids and retinol (e.g. milk). ` Age, total energy intake, smoking (past and current amount in 1980; 1-4, 5-14, 15-24, 25-34, 35-44, 45+) and age of starting to smoke (~ 17, 18-19, 20-21, 22+). I

480 Tablr 5. Dietary intake of specific carotenoids and relative risk of lung cancer, adjusted for age, total energy intake, and cigarette smoking Carotenoid Quintile for intake QI v.r Q5 (95% CI) Trend p value 1 2 3 4 5 u-Carotene 1.0 0.8 0.7 0.7 0.6 (0.4-0.8) 0.003 J3-Carotene 1.0 1.1 0.8 1.0 0.8 (0.6-1.1) 0.17 Cryptoxanthin 1.0 0.8 0.7 0.8 0.8 (0.6-LI) 0.44 Lycopcnc 1.0 0.7 0.6 0.9 0.8 (0.6-t.l) 0.76 Lutein L0 1.1 11 0.9 1.1 (0.8-1.5) 023 Adjusted for age, total energy intake, smoking (past, current amount in 1980; 1-4, 5-14, 15-24, 25-34, 35-44, 45+) and age of starting to smoke (517, 18-19, 20 21, -22+). to lung cancer. This study suggests that the findings in women do not differ from those of men. We found no effect of tar content, perhaps because the large majority of these women smokers were smoking low-tar cigarettes. The effect of stopping smoking was strong and became evident relatively soon (within 2-5 years). These data provide additional evidence of an excess risk of lung cancer among women exposed to environ- mental-tobacco smoke exposure in adult life defined as exposure both at home and at work. The estimated Table 6- Food intake and relative risk of lung cancer, Nurses' Health Study, 198a1992' Food Frequency af consumption Trend value Never 1-3 I/ 2-4/ 5+f months week week week I+/day p Vegetables Broccoli 1 A LO 0.8 0.9 0.2 0.03 Carrots 10 0.8 0_7 0.6 0.4 0.003 Spinach and 1.0 1.0 0.9 0.9 1.1 0.69 othcr greens Sweet 1-0 t.l 10 0.56 potatoes Cabbage 10 1.0 1 0 1-1 0-6 Corn 1.0 0.8 0.8 0.8 0.5 0.18 Peas 1 0 1.0 0_9 1-0 07 0.65 Beans 1.0 1.0 1.0 1.1 0.23 Frvit Oranges 1.0 0.7 0.6 0.8 0.7 0.01 Orange juice 10 0.7 0.6 0-8 0 7 0.17 Peaches 1-0 0.9 0.8 0.8 0.9 0.26 Bananas L0 0.9 0.8 L0 1.0 0.77 Other fruit 10 0.8 0.7 47 0.8 0.17 Tomatoes 1.0 0.9 0.8 L0 0.8 0.46 ' Adjusted for age, smoking (past, current amount in 1980; 1-4, 5-14, 15-24, 25-34, 35-44, 45+) and age of starting to smoke (<17, 18-19, 20-21. 22+). F.E. Speizer et al. relative risk was 1.5, which is somewhat stronger than the risks reported in other studies [28, 29], but the confidence interval was wide because only two eases occurred among unexposed women. Although animal experiments [30] and some human studies [9] suggest a protective effect for foods rich in ji-carotene, the recently reported trials of (I-carotene supplementation on risk of lung cancer point out the possible difficulties of extrapolating from observational studies on food intake to the effect of a specific dietary supplement. Furthermore, van Poppet et alL recently reported a randomized study comparing oxidative DNA damage (as assessed by level of excretion of 8-oxo-7,8- dehydro-2-deocyguanosine, a measure used to demon- strate effects of smoking) in two groups of smokers: one given supplemental Q-carotene and the other given a placebo [31]. No effect of the supplement was detected. However, in an animal model, a strong proliferative effect and bronchial metaplasia was observed in lun~ tissue when fl-carotene-supplemented animals were ex- posed to tobacco smoke [321. We found no apparent protective effects of some specific foods believed to be high in [1-carotene, yet we did observe aa protective association with carrots - the major dietary source of a- carotene as well as /f-carotene [25, 33]. Carrots contrib- ute 3725 units of a-carotene per serving and are the only important source of a-carotene. The relation between cigarette smoking and intake of carotenoid-rich foods seen in this and other studies [34, 351 emphasizes the importance of controlling for level of cigarette smoking when examining dietary risk factors. Previous studies based on measures of diet and serum /i- carotene levels suggest that some protective association may persist after cigarette smoking is controlled for, although residual confounding cannot be excluded [36]. We observed little association between carotene or folate intake and risk of lung cancer after controlling for smoking; however, increased carrot consumption t. mained a strong predictor of reduced lung cancer risk. Furthermore, when we assessed in detail the specific dietary carotenoids in the foods consumed by women, we observed significant protective associations with (1-caro- tene_ These results are similar to case-control studies conducted by Ziegler [37] and Le Marchand er al. (14], who found fi-carotenoid-containing dark green vegeta- bles (high in lutein), and tomatoes (high in lycopene), as well as all vegetables combined, to be associated with lower risk of lung cancer. In a large case-control study of lung cancer among non-smoking women (dominated by adenocarcinomas), Alavanja et a!. [71 found no relation to total carotenoids, /3-carotene., or vitamin A, with or without supplements, as assessed by a dietary question- naire. Similarly, they found no relation to vitamin C or E 2505586066

1 Snxoking, antioxidant intake and lung cancer with or without supplements. These authors made no attempt to separate the various carotenoids. Because the various carotenoids tend to be found in the same foods, we cannot be confident that carotenoids other than fl-carotene are primarily responsible for the inverse associations observed in this cohort between lung cancer and consumption of vegetables, particularly carrots. Although other factors in these foods have anticancer effects [9], carotenoids retain a potential role. For example, some laboratory studies suggest greater anticarcinogenic potential for other carotenoids than for fl-carotene. Murakoshi et al. [38] found that a- but not. /3-carotene reduced the number of lung tumors in a two- stage model in mice. Findings were similar in liver and skin cancer models and in studies in which the growth of human neuroblastoma cells was inhibited [39]. Although /3-ca.rotene ranks ahead of a-carotene with regard to , singlet oxygen quenching, the tissue concentration of a- ~ carotene is I log lower [40]. Whether the way we eat and process foods containing these carotenoids subtly alters their potency is unknown. - Concern may be raised regarding the large number of analyses conducted to evaluate the relation between carotenoid intake and risk of lung cancer. However, we note that evaluation of relations for nutrients and foods is standard practice in nutritional epidemiology. When an association with disease is found for a specific nutrient, it is important to examine and report whether the major foods contributing to the nutrient are also related similarly to risk of disease [24]. Furthermore, when examining the association between carrot intake and lung cancer, exclusion of women who had greatly increased or decreased their intake over the 10 years prior to 1980,.refined the exposure measure by excluding those women with unstable intake, the association became stronger. ^ Although some previous studies, but not all, have ~- suggested a relation of dietary fat intake with risk of lung cancer among men and women [7, 15, 40-48] the findings in this study for fat are strikingly null. Previous reports have indicated approximately twofold differen- ces in risk for the extreme quartiles of intake. A somewhat stronger association was observed by Ala- vanja et a(L among non-smoking women [7]. However, several prospective cohort studies have failed to show such a relation [49-53]. While fat may act as a promoter of caroinogenesis, the inconsistent findings for total and- specific subtypes of fat across the numerous previous studies suggest the need for caution in the interpretation of this possible relation. Somewhat more consistent positive associations have been reported for dietary cholesterol, and the findings from the Nurses' Health Study are consistent with a modest positive association. 481 In women, as in men, smoking is the most powerful cause of lung cancer. Perhaps of equal importance is that, within as little as 10 years after stopping smoking, women have approximately a 50% reduction in the risk of lung cancer, and the risk continues downward after 15 years. From a public health perspective these data are encouraging. Lung cancer now kills more women than breast cancer and it is clear that this epidemic in women is related to cigarette smoking. Thus, stopping smoking (or even better, not starting) remains a crucial public health message. Diet modification may provide a modest measure of protection, but will not be sufficient to counterbalance the adverse effect of smoking. Acknowledgements Unfailing support from staff of the Nurses' Health Study has made this work possible. Special thanks go to Barbara Egan, Lisa Dunn, Karen Corsano, Gary Chase, and Sue-Wei Chiang for assistance with this project. We would also like to thank Sir Richard Peto for his helpful comments. The work was supported by Grant-CA 40356 from the National Institutes of Health, National Cancer Institute, Bethesda, MD, USA. References 1. Thun M, Dally-Lalty C, Myers D, et a(. (1997) Trends in tobacco smoking and mortality from cigarette use in Cancer Prevention Studies.i (1959 through 1965) and 11 (1982 through 1988)_ In: Burns D, Garfinkel L, Samet J, eds. Changes in Cigarette-related Disease Risk and their Implications for Prevention and Contr'ol. Washington, DC: NIH Publication no. 97-0213, pp. 305-333. 2. Friedman G, Tekawa I, Sadler M, Sidney S (1997) Smoking and mortality: the Kaiser Permanente experience. In: Burns D, Garfinkel L, Samet J, eds. Changes in Cigarette-related Disease Risk and their Implications for Prevention and Conbot. Washing- ton,DC: NIH Publication no. 97-4213, pp. 477-499. 3. US Department of Health and Human Services (1989) Surgeon General's Report. Tlre Fleahh Conseqaences of Smoking. Wash- ington, DC: Government Printing Office. 4. Wu A, Yu M, Thomas D, Pike M, Henderson B (1988) Persona] and family history of lung disease as risk factors for adenocarci- noma of the lung. Cancer Res 48: 7279-7284. 5. Speizer F, Samet J(1994) Air pollution and lung cancer. In: Samet . J, ed. Epidemiology of Lung Cancer. New York: Marcel Dekker, pp.131-150. 6. Wu AH, Henderson BE, Pike MC, Yu MC (1985) Smoking and . other.risk factors for lung cancer in women._J.Nn11 Cancer Inst 74: 747-751. 7. Alavanja MC, Brown CC, Swanson C, Brownson RC (1993) ,-- Saturated fat intake and lung cancer risk among non-smoking women in Missouri. J Natt Cancer Inst 85: 1906-1916. 8. Gerster H(1993) Anticarcinogenic effect of common carotenoids. J Vit Nufr Research 63: 93-121. 9. Steinmetz KA, Potter JD (1991) Vegetables, fruit, and cancer. II. Mechanisms Cancer Cautes Control 2: 427-442.

482 10. Ziegler RG, Mayne ST, Swanson CA (1996) Nutrition and lung cancer. Cancer Causes Control 7: 157-177. 1 L Albanes D, Heinonen OP, Huttunen JK, et af. (1995) Effects of alpha-tocopherol and beta-carotene supplements on cancer inci- dence in the Alpha-Tocopherol Beta-Carotene Cancer Prevention Study. Am J Clin Nutr 62 (Suppl. 6): 51427S1430. 12. Hennekens CH, Buring JE, Manson JE, et al. (1996) Lack of etfect of long-term supplementation with beta carotene on the incidence of malignant neoplasms and cardiovascular disease. N Eng! J Med 334: 1145-1149. 13. Omenn GS, Goodman GE, Thornquist MD, e1 al. (1996) Effects of a combination of beta carotene and vitamin A on lung cancer and cardiovascular disease. N Engl J Med 334: 1 I50-1155. 14. Le Marchand L, Hankin 3H, Kolonel LN, Beecher GR, Wilkens LR, Zhao LP (1993) Intake of specific carotenoids and lung cancer risk. Cancer Epidemiol Biomarkers Prev 2: 183-187. - 15. Ziegler RG, Colavito EA, Hartge P, et al. (1996) Importance of n-carotene, /i-carotene, and other phytochemicals in the etiology of lung cancer. J Natl Cancer Inst 88: 612-615. 16. Willett WC, Green A, StampQer MJ, et al. (1987) Relative and absolute excess risks of coronary heart disease among women who smoke cigarettes. N Engl J Med 317: 1303-1309. 17. Kawachi 1, Colditz G, Stampfer M, et aL (1993) Smoking cessation and time course of decreased risk of coronary heart disease in women. Arch Intern Med 154: 169-175. 18. Cotditz GA, Bonita R, Stampfer MJ, et al. (1988) Cigarette smoking and risk of stroke in middle-aged women. N Eng/ J Med 318: 937-941. 19. Kawachi I, Colditz GA, Stampfer MJ, et al. (1993) Smoking cessation in relation to total mortality rates in women: a prospective cohort study. Ann Inlern Med 119: 992-1000. 20. Colditz GA (1995) The Nurses' Health Study: A cohort of US women followed since 1976. J Am Med Womens Assoc 50: 40-04 and 63. 21. Stampfer MJ, Willett WC, Speizer FE, et al. (1984) Test of the National Death Index. Am J Epidemio! 119: 837-839. - 22. Rich-Edwards 1, Corsano K, Stampfer M (1994) Test of the National Death Index and Equifax Nationwide Death Search. Am J Epidemiol140: 1016-1019. 23. Willett WC; Sampson L, Stampfer MJ, et al. (1985) Reproduc- ibility and validity of a semiquantitative food frequency question- naire. Am J Epidemia! 122: 51-65. 24. Willett WC (1998) Nutritional Epidemiology, 2nd edn. New York: Oxford University Press, 1998. 25. Chug-Ahuja JK, Holden JM, Forman MR, Mangels AR, Beecher GR, Lanza E (1993) The development and application of a carotenoid database for fruits, vegetables, and selected multicom- ponent foods. J Am Diet Assoc 93: 318-323. 26. Willett WC, Stampfer MJ (1986) Total energy intake: implications for epidemiologic analyses. Am J Epidemiol 124: 17-27. 27. Federal Trade Commisson (1978) Report of "Tar" and Nicotene Content of Smoke of 167 Varieties of Cigarettes. Washington, DC: Federal Trade Commission. 28. Dockery D, Trichopoulos D (1997) Risk of lung cancer from environmental exposures to tobacco smoke. Cancer Causes Control 8: 333-345. 29. Blot W, McLaughlin J (1998) Passive smoking and lung can- cer risk: what is the story now? J Natl Cancer Inst 90: t416- 1417. 30. Wattenherg LW (1985) Chemoprcvention of cancer. Cancer Res 45: 1-8. 31. Van Poppel G, Paulsen H, Shopt S, Verkagen H (1995) No influence of beta-carotene on oxidative DNA damage in male smokers. J Narl Cancer Inst 87: 310-31 L F.E. Speizer et al. 32. Wang X-D, Liu C, Bronson R, Smith D, Krinsky N, Russell R (1999) Retinoid signaling and activator protein-I expression in ferrets given b-carotene supplements and exposed to tobacco smoke. J Nail Cancer Inst 91: 60-66. 33. Mangels AR, Holden JM, Beecher GR, Forman MR, Lanz1 E (1993) Carotenoid content of fruits and vcgetables: an evaluation of analytic data. J Am Diet Assoc 93: 284-296. 34. Subar A, Harlan L, Mattson M (1990) Food and nutrient intake differences between smokers and non-smokers in the United States. Am J Public Health 80: 1323-1329- 35. Rimm E, Colditz G(1993) Smoking, alcohol, and plasma levels of carotenes and vitamin E. Ann NY Acad Sci 686: 323-334. 36. Willett W, Colditz G(1998) Vitamin A and lung cancer. In: Willett W, ed- Nutritional Epidemiology. New York: Oxford University Press, pp. 357-376. 37. Ziegler RG (1991) Vegetables, fruits and carotenoids and the risk of cancer. Am J Clin Nutr 53: 251S-259S. 38. Murakoshi M, Nishino H, Satomi Y, er al. (1992) Potent preven- tive action of alpha-carotene against carcinogenesis: spontaneous liver carcinogenesis and promoting stage of lung and skin carcinogenesis in mice are suppressed more effectively by alpha- carotene than by beta-carotene. Cancer Rer 52: 6583-6587. - 39. Nishim H (1995) Cancer chemoprevention by natural carotenn_~ and their related compounds. J Cell Bioclsem 33 (Suppl.): 231-235. 40. Hinds MW, Kolonel LN, Hankin JH, Lee J (1983) Dietary cholesterol and lung cancer risk in a multiethnic population in Hawaii. /nt J Cancer 32: 727-732. 41. Goodman MT. Kolonel LN, Yoshrzawa CN, Hankin JH (1988) The effect of dietary cholesterol and fat on the risk of lung cancer in Hawaii. Am J Epidemiol 128: 1241-1255. 42. Byers TE, Graham S, Haughey BP, Marshall 7R, Swanson MK (1987) Diet and lung cancer risk: findings from the Western New York Diet Study. Am J Epidemiol 125: 351-363. 43. Wynder EL, Hebert JR, Kabat GC (1987) Association of dietary fat and lung cancer. J Ndt! Cancer Inst 79: 63I-637. 44. Jai,n M, Burch JD, Howe GR, Risch HA, Miller AB (1990) Dietary factors and risk of lung cancer: results from a case-control study, Toronto, 1981-1985. lnt J Cancer 45: 287-293. 45. Shekelle RB, Rossof AH, Stamler J(1991) Dietary cholesterol and incidence of lung cancer: the Western Electric Study. Am J Epidemiol 134: 480-484. 46. Knekt P, Seppanen R, Jarvinen R, et al. (1991) Dietary cholester- ol, fatty acids, and the risk of lung cancer among men_ Nutr Cancer 16: 267-275. 47. Xic JX, LesaRre E, Kesteloot H (1991) The relationship betv animal fat intake, cigarette smoking, and lung cancer. Cam.~ Cau.res Control 2: 79-83. 48. De Stefani E, Denco-PeBegrini H, Mendilaharsu M, Carzoglio J, Roneo A (1997) Dietary fat and lung cancer: a case~control study in Uruguay. Cancer Causes Control 8: 913-92I. 49. Heilbrun LK, Nomura AM, Stemmcrmann GN (1984) Dietary cholesterol and lung cancer risk among Japanese men in Hawaii. Am J Clin Nutr 39: 375-379. 50. Fraser GE, Beeson W L, Phillips RL (1991) Diet and lung cancer in California Seventh-day Adventists. Am J Epidemiol 133: 683-693. 51. Kvale G, Bjelke E, Gart JJ (1983) Dietary habits and lung cancer risk. InE J Cancer 31: 397-405. 52. Bandera E, Freudenheim J, Marshall J, el al. (1997) Diet and alcohol consumption and lung cancer risk in the New York State Cohort. Cancer Causes Cantrol7: 828-840. 53. Veierod M, Laake P, Thelle D(19_97) Dietary fat intake and risk of lung cancer: a prospective study of 51,452 Norwegian men and women. Eur J Cancer Prev 6: 540-549.