Cancer Causes Control (2012) 23:1113–1126 DOI 10.1007/s10552-012-9980-3
ORIGINAL PAPER
Body mass index and lung cancer risk: results from the ICARE study, a large, population-based case–control study Chloe´ Tarnaud • Florence Guida • Alexandra Papadopoulos • Sylvie Ce´ne´e • Diane Cyr • Annie Schmaus • Loredana Radoı¨ • Sophie Paget-Bailly • Gwenn Menvielle • Antoine Buemi • Anne Sophie Woronoff • Daniele Luce • Isabelle Stu¨cker
Received: 23 August 2011 / Accepted: 23 April 2012 / Published online: 19 May 2012 Ó Springer Science+Business Media B.V. 2012
Abstract Background The association between body mass index (BMI) and lung cancer is still disputed because of possible residual confounding by smoking and preclinical weight loss in case–control studies. We examined this association using data from the multicenter ICARE study in France, a large, population-based case–control study. Methods A total of 2,625 incident lung cancer cases and 3,381 controls were included. Weight was collected at interview, 2 years before the interview, and at age 30. C. Tarnaud F. Guida A. Papadopoulos S. Ce´ne´e I. Stu¨cker (&) Inserm, CESP Centre for Research in Epidemiology and Population Health, U1018, Environmental Epidemiology of Cancer Team, UMRS 1018, Equipe 6, Baˆtiment Inserm 15/16, 16 avenue Paul Vaillant-Couturier, 94807 Villejuif Cedex, France e-mail:
[email protected] C. Tarnaud F. Guida A. Papadopoulos S. Ce´ne´e I. Stu¨cker Univ. Paris-Sud, UMRS 1018, 94807 Villejuif, France D. Cyr A. Schmaus L. Radoı¨ S. Paget-Bailly G. Menvielle D. Luce Inserm, CESP Centre for Research in Epidemiology and Population Health, U1018, Epidemiology of Occupational and Social Determinants of Health, 94807 Villejuif, France D. Cyr A. Schmaus L. Radoı¨ S. Paget-Bailly G. Menvielle D. Luce Universite´ de Versailles St-Quentin, UMRS 1018, 94807 Villejuif, France A. Buemi Registre des cancers du Haut-Rhin, Mulhouse, France A. S. Woronoff Registre des tumeurs du Doubs, CHU Saint Jacques, Besanc¸on, France
Lifetime smoking exposure was calculated using the comprehensive smoking index (CSI). Adjusted odds ratios (aORs) and 95 % confidence intervals were estimated by unconditional logistic regression and controlled for age, area, education, CSI, occupational exposure, previous chronic bronchitis, and parental history of lung cancer. We also examined the role of weight change. Analyses were stratified by smoking status and sex. Results When compared with that of men with normal BMI 2 years before the interview, lung cancer aORs (95 % CI) among men with BMIs of \18.5, 25–29.9, 30–32.4, and C32.5 kg/m2 were 2.7 (95 % CI 1.2–6.2), 0.9 (95 % CI 0.7–1.1), 0.8 (95 % CI 0.6–1.1), and 0.8 (95 % CI 0.6–1.0), respectively (ptrend = 0.02). Results were more pronounced among current smokers and were similar in men and women. Weight gain over time was associated with a significant decreased risk of lung cancer. Conclusions We found an inverse dose-dependent association between lung cancer risk and BMI 2 years prior to interview in current smokers. Impact statement BMI might be an individual factor impacting the risk of lung cancer related to smoking’s carcinogen-induced DNA damage. Keywords Lung cancer Body mass index Weight change Case–control study
Background Lung cancer remains a leading cause of mortality and morbidity worldwide, with more than 12 million new cases and 7.5 million deaths in 2008 [1]. In France, lung cancer ranks second among all cancers, with 32,430 new cases in 2008, but is the most common cause of death by cancer,
123
1114
with 27,793 deaths [1]. The overwhelming cause of lung cancer is tobacco smoking, and lung cancer rates reflect the historical prevalence of smoking among men and women [2]. Second-hand smoke has also been established as causally related to lung cancer, even though the strength of the association is much weaker [3]. Smoking is also causally related to pulmonary disease, such as chronic obstructive pulmonary diseases (COPD), and the associations between lung cancer and COPD [3–9], and emphysema [4, 10] have been known for 60 years and continue to be discussed [3]. Occupational exposures (e.g., asbestos, silica, heavy metals, polycyclic aromatic hydrocarbons) constitute the other major determinants of lung cancer [11]. The World Health Organization (WHO) has recently highlighted the influence of body mass index (BMI) on the risk of developing several cancers [12]. Obesity is associated with an increased risk of colorectal and post-menopausal breast cancers [13], whereas underweight is associated with pre-menopausal breast [14] and oral and pharyngeal cancers [15–17]. Since the 1990s, case–control and cohort studies have suggested that underweight people are at an increased risk of lung cancer [18–33], especially those who are current smokers [18, 22–25, 28, 31, 32], whereas no association is usually found among nonsmokers [18, 20, 22–25, 28, 34]. Despite numerous studies supporting an association between low BMI and lung cancer among smokers, the hypothesis of a direct relationship is still controversial. The debate focuses on two main methodological pitfalls that could explain these results. First, residual confounding by smoking could explain why the association is found in smokers but not in non-smokers. Although smokers have lower body weights than non-smokers, this association is complex; heavy smokers tend to have greater body weights than do light smokers, which likely reflects a clustering of risky behaviors (e.g., low degree of physical activity, poor diet, smoking) that is conducive to weight gain [35]. Careful adjustment for tobacco smoking is thus required to avoid any residual confounding by smoking. The second issue is the fact that weight loss can pre-date lung cancer shortly before diagnosis. Case–control studies therefore encounter difficulty if BMI refers to weight at inclusion in the study, which usually corresponds to the time of diagnosis. From 25 % [36, 37] to 50 % [38, 39] of patients lose about 5 % [38] of their usual body weight before diagnosis. Considering a patient’s usual weight (i.e., at least 1 year before diagnosis) is thus crucial to ascertaining the independent association of weight with lung cancer. Only three case–control studies have considered weight before diagnosis [21, 25, 34], of which one reported an increased lung cancer risk for underweight people [21], one found no association [25], and one showed an increased lung cancer risk for overweight people [34]. To solve this problem,
123
Cancer Causes Control (2012) 23:1113–1126
cohort studies have excluded cases diagnosed in the first years following enrollment [18, 23, 27, 28, 32, 33] or used past usual weight [20, 22]. These studies still found an increased risk of lung cancer for underweight individuals [18, 20, 23, 27, 28, 32, 33], with the exception of one study [22]. The influence of BMI in lung cancer is thus still controversial and requires clarification. The ICARE study is a large, population-based case–control study set up in France to explore the roles of lifestyle and environmental, occupational, and genetic susceptibility risk factors in lung and upper aerodigestive tract (UADT) cancers. Using data from this study, we examined the association between BMI and lung cancer while overcoming the limitations of previous studies, especially bias due to preclinical weight loss and insufficient adjustment for tobacco smoking. In our questionnaire, subjects were asked to declare their weight at age 30, 2 years before the interview, and at inclusion in the study, allowing us to examine patients’ BMIs long before the disease developed and the role of weight change during adulthood in the risk of lung cancer, an issue that has rarely been examined [21–23].
Methods Study design The ICARE study, conducted from 2001 to 2006, is a large, multicenter, population-based case–control study of respiratory cancers [40]. The study was conducted in 10 of the 11 French de´partements1 (administrative areas) with population-based general cancer registries and covered a population of 7.6 million people (about 13 % of the French population). All new cases of lung and UADT cancers aged 18–75 years identified during the study period in each registry were eligible for inclusion in the study. These analyses focus on lung cancer cases and controls. Cases were all diagnosed and histologically confirmed incident primary lung cancers. The interviewers regularly contacted the pathologists and clinical services in the registry catchment area of each de´partement to identify eligible cases. Permission for each patient to be included in the study was requested from physicians. Pathology reports were reviewed by registry physicians to determine the topography and morphology type of the tumors according to the International Classification of Diseases for Oncology, 3rd edition [41], codes C33–C34. All histological types were included. Each case was interviewed within 3 months of diagnosis, on average. Of the 4,865 eligible cases identified, 781 (16 %) died before any contact could 1
France is divided into 96 administrative units called de´partements.
Cancer Causes Control (2012) 23:1113–1126
be made, and 486 (10 %) could not be located. Among the 3,598 remaining cases, 238 could not be contacted because of their health status, leaving 3,360 potential participants. Four hundred and thirty-four (13 %) refused to participate. Population controls were randomly selected through incidence-density sampling [42] from residential telephone directories. Unlisted numbers were reached by increasing the last digit of each number by one. Controls were frequency-matched to cases by age, gender, and de´partement. Additional stratification ensured that controls were representative of the de´partement’s population in terms of socioeconomic status (based on the last job held) and gender. Among the 4,676 eligible controls, 4,411 were contacted, and 3,555 agreed to participate (81 % overall; 81 % for men, 77 % for women). The study thus included 6,481 participants: 2,926 cases and 3,555 controls. Men accounted for 78 % of the study population. The present analysis includes the 6,006 (93 %) participants who answered the complete questionnaire with height and weight data. Indeed, when subjects were too tired to answer the complete questionnaire, interviewers used a summary version of the questionnaire that included mainly lifetime information on both smoking and occupational history, but not weight, at the different time periods. This short version was used for 5 % of men and 3 % of women. A total of 2,625 cases (2,029 men and 596 women) and 3,381 controls (2,641 men and 740 women) were available for analysis. The ICARE study was approved by the Institutional Review Board of the French National Institute of Health and Medical Research (IRB-Inserm, n° 01–036). Confidentiality was guaranteed, and informed consent to participate in the study was obtained from patients and controls as recommended by the French Data Protection Authority (Commission Nationale Informatique et Liberte´), which also approved the ICARE study (CNIL n° 90120). Data collection Trained interviewers used standardized questionnaires in face-to-face interviews to collect information about demographic characteristics, education, personal and familial medical history, alcohol use, and residential history. Lifetime occupational history, including all jobs held for at least 1 month, was collected and compiled using lists of occupations and/or industries known (list A) or suspected (list B) to be associated with lung cancer. These lists are based on previous epidemiologic studies and on evaluations of carcinogenicity by the International Agency for Research on Cancer [43] and are regularly updated [44, 45], providing a standardized tool to quantify the burden of occupational lung cancer [46]. Smoking history was collected for all periods of consumption and included the
1115
number of cigarettes smoked/day, the type (blonde vs. dark) and brand of cigarettes, whether or not filters were used, and the inhalation pattern. Modification of any of one of these variables constituted a new period of consumption. Smokers were defined as subjects who had smoked at least 100 cigarettes in their lifetime, and former smokers were defined as those who had not smoked for at least 2 years. BMI and weight change Subjects were asked to report their current weight and height as well as their weight at 30 years old and 2 years before the interview. BMI was computed as weight (kg) divided by the square of height (m2). BMI is a continuous variable and was grouped according to the WHO international classification [47] as follows: underweight (BMI \ 18.5 kg/m2), normal (18.5 B BMI \ 25 kg/m2, used as the reference group), overweight (25 B BMI \ 30 kg/m2), and obese (BMI C 30 kg/m2) [48]. A higher cutoff (BMI C 32.5 kg/m2) [47] for obesity was used only for men because the sample included few obese women (37 women and 156 men were obese at age 30). We calculated change in weight during adulthood as the difference between weight 2 years before the interview (at a mean age of 59 years) and weight at age 30. We considered a threshold of 3 kg, corresponding to a BMI difference of one unit for an individual of the mean height of 1.71 m. The resulting continuous variable was categorized as follows: gained more than 3 kg; minimal change in weight (gained or lost \3 kg, used as the reference group); or lost 3 kg or more. Tobacco smoking: definition of the comprehensive smoking index (CSI) Lifetime tobacco smoking exposure was assessed using the CSI. The CSI aggregates the mean number of cigarettes smoked/day, total years of smoking, and time since cessation [49]. The CSI of never smokers is null. In our data, the CSI varies linearly with lung cancer risk and was used for adjustment as a continuous variable. Statistical analysis Unconditional logistic regression was used to estimate adjusted odds ratios (aORs) and 95 % confidence intervals (CIs) for associations between lung cancer risk and BMI prior to interview. Multivariable-adjusted models were adjusted for age (B50, 50–60, 60–70, [70), de´partement, tobacco smoking consumption (CSI, as a continuous variable), educational level (general secondary education, vocational secondary education, or primary education),
123
1116
occupational exposure (yes/no, based on whether a subject had held at least one job included on list A, the list of industries and occupations known to be associated with lung cancer), previous chronic bronchitis (yes/no), and parental history of lung cancer (yes/no). Analyses were conducted separately by sex and performed both in the total population and separately for current, former, and nonsmokers. We examined associations between lung cancer risk and weight change only among participants who had a normal BMI at age 30 to avoid any confusion between normal weight change during adulthood and weight change induced by already pathologic weight at age 30. We investigated the influence of tobacco smoking (CSI, pack-years, and intensity of smoking) on BMI and change in weight by calculating a Spearman’s correlation coefficient. We also examined the modification effect of BMI on the association between smoking status and lung cancer with likelihood ratio test on two nested models, one including the interaction term and the other without. In order to explore competitive mortality bias, analyses were repeated for participants younger than 50 years old, who are less at risk of cardiovascular mortality, especially women. Because COPD has been associated with a decreased BMI [50, 51] and could be associated with an increased risk of lung cancer [3–9], we adjusted our analyses for chronic bronchitis and also repeated the main analyses after exclusion of subjects who reported this antecedent. All statistical tests were two-sided, and all analyses were performed using the STATA 10.0 software system.
Results Table 1 presents sociodemographic characteristics, tobacco smoking consumption of cases and controls, and histological subtypes of lung cancer cases. Cases were older than controls in men (mean age: 60.1 vs. 58.1, p \ 0.001) but younger in women (57.4 vs. 60.4, p \ 0.001). Only male cases had a lower educational level than controls. Male and female cases more frequently had held at least one job included in the list of industries and occupations known to be associated with lung cancer (list A). Both men and women more frequently reported suffering from chronic bronchitis and having a parental history of lung cancer. Smoking status was strongly associated with lung cancer risk, with an overriding proportion of never smokers among female cases compared with male cases (29.2 and 2.6 %, respectively, p \ 0.001). Increasing CSI was associated with a steady increase in the mean number of packyears, cigarettes smoked/day, and total years of smoking but with a steady drop-off of total years since quitting. A CSI greater than two corresponded to the consumption of a mean of 63 pack-years, or 30 cigarettes/day over
123
Cancer Causes Control (2012) 23:1113–1126
44 years, with time since cessation \1 year. CSI was strongly associated with lung cancer risk, with a linear dose–effect relationship (ptrend \ 0.001). Adenocarcinoma was the predominant histological type among female cases, whereas adenocarcinoma and squamous cell carcinoma were the main subtypes for men. Weight 2 years before inclusion in the study was missing for 6 subjects, whereas weight at age 30 years was missing for 408 participants (7 %), who were more frequently cases (64 %). In comparison with cases who reported weight at 30 years, those who did not were preferentially current smokers (57 vs. 66 %, p \ 0.001) and stopped education earlier (19 vs. 25 % had a general secondary level, p = 0.048). However, among cases, we found no statistical difference between individuals with and without a reported weight at age 30 with respect to mean age (59.6 vs. 59.7 years, p = 0.77), height (170.6 vs. 170.2 cm, p = 0.26), weight at the interview (70.3 vs. 69.2 kg, p = 0.43), weight 2 years prior to the interview (74.8 vs. 74.6 kg, p = 0.92), and CSI (1.49 vs. 1.53, p = 0.38). The same comparisons between controls yielded similar results: mean age (58.7 vs. 59.4 years, p = 0.21), height (171.2 vs. 171.2 cm, P = 0.97), weight at interview (77.5 vs. 77.5 kg, p = 0.10), weight 2 years before (77.4 vs. 75.9 kg, p = 0.25), and CSI (0.57 vs. 0.59, p = 0.99). The main determinants of BMI among controls were studied (Table 2). BMI was associated with sex, age, educational level, chronic bronchitis, and tobacco smoking status. Among controls, we examined the correlation between CSI and BMI separately for current smokers and former smokers. Although the Spearman’s coefficient correlations were weak, they were positive and significant (r = 0.11, p = 0.004, and r = 0.11, p \ 0.001, for current and former smokers, respectively). Results were similar when considering pack-years or intensity of smoking. Finally, CSI was not correlated with the weight change between 2 years before the interview and age 30 years (r = 0.05, p = 0.18). Figure 1 shows trends in the evolution of average BMI from age 30 years to the date of interview according to smoking status. Average BMI similarly increased from age 30 years until 2 years before the interview for both cases and controls, although the slope was somewhat weaker for current smoker cases compared with their controls. Two years before the interview, the figures reveal a break in the trends between cases and controls. Whereas the average BMI of controls seems to plateau until the interview, the average BMI of cases dropped dramatically within the 2 years prior to the interview. Table 3 presents the association between lung cancer risk and BMI prior to interview. Two years before the interview, BMI was inversely associated with lung cancer risk for both men and women although the trend was more
Cancer Causes Control (2012) 23:1113–1126
1117
Table 1 Background characteristics of cases and controls and histological subtypes of lung cancer cases Men (n = 3,381) Controls (n = 2,641) n (%)
ORa (95 % CI)
Cases (n = 2,029) n (%)
Women (n = 1,336) Controls (n = 740) n (%)
ORa (95 % CI)
Cases (n = 596) n (%)
De´partements Calvados
356 (13.5)
239 (11.8)
102 (13.8)
77 (12.9)
Doubs, Belfort He´rault
127 (4.8)
99 (4.9)
48 (6.5)
41 (6.9)
309 (11.7)
195 (9.6)
73 (9.9)
68 (11.4)
Ise`re
365 (13.8)
299 (14.7)
87 (11.8)
83 (16.9) 76 (12.8)
Loire-Atlantique
328 (12.4)
263 (13.0)
102 (13.8)
Manche
246 (9.3)
240 (11.8)
65 (8.9)
57 (9.6)
Bas-Rhin
343 (13.0)
275 (13.6)
108 (14.6)
90 (15.1)
Haut-Rhin
66 (2.5)
56 (2.8)
11 (1.5)
16 (2.7)
Somme Vende´e
370 (14.0)
254 (12.5)
106 (14.3)
51 (8.6)
131 (5.0)
109 (5.4)
38 (5.1)
37 (6.2)
B50
630 (23.9)
291 (14.3)
0.6 (0.5; 0.7)
161 (21.8)
149 (25.0)
1.4 (1.1; 1.9)
50–60 60–70
807 (30.6) 877 (33.2)
704 (34.7) 715 (35.2)
1.1 (0.9; 1.3) 1. 0 reference
146 (19.7) 252 (34.1)
207 (34.7) 157 (26.3)
2.2 (1.7; 3.0) 1.0 reference
[70
327 (12.4)
319 (15.7)
1.2 (1.0; 1.5)
181 (24.5)
82 (13.9)
0.7 (0.5; 1.0)
Age
Education level General secondary education and higher
1,023 (39.6)
448 (22.9)
1.0 reference
232 (31.8)
179 (30.9)
1.0 reference
Vocational secondary education
1,057 (40.9)
855 (43.6)
1.8 (1.6; 2.1)
259 (35.5)
204 (35.2)
1.0 (0.8; 1.3)
Primary education or less
505 (19.5)
656 (33.5)
3.0 (2.5; 3.5)
239 (32.7)
197 (34.0)
1.1 (0.8; 1.4)
Ever held a job in list Ab No
2,324 (88.0)
1,622 (80.3)
1.0 reference
733 (99.1)
573 (96.6)
1.0 reference
Yes
317 (12.0)
399 (19.7)
1.8 (1.5; 2.1)
7 (0.9)
20 (3.4)
3.7 (1.5; 8.7)
Previous chronic bronchitis No
2,484 (94.2)
1,679 (82.9)
1.0 reference
689 (93.6)
472 (79.5)
1.0 reference
Yes
154 (5.8)
347 (17.1)
3.3 (2.7; 4.1)
47 (6.4)
122 (20.5)
3.8 (2.7; 5.4)
Parental history of lung cancer No
2,343 (94.7)
163 (91.8)
1.0 reference
653 (94.1)
463 (87.9)
1.0 reference
130 (5.3)
147 (8.2)
1.6 (1.3; 2.1)
41 (5.9)
64 (12.1)
2.2 (1.5; 3.3)
Never smoker
771 (29.2)
52 (2.6)
1.0 reference
496 (67.0)
173 (29.1)
1.0 reference
Former smoker
1,289 (48.8)
786 (38.8)
8.3 (6.2; 11.2)
129 (14.4)
103 (17.3)
2.2 (1.6; 3.0)
Current smoker
581 (22.0)
1,186 (58.6)
37.2 (27.4; 50.5)
115 (15.5)
319 (53.6)
8.8 (6.5; 12.0)
0.0
771 (29.2)
52 (2.6)
1.0 reference
496 (67.2)
173 (29.2)
1.0 reference
0.0–0.5
589 (22.4)
59 (4.4)
2.2 (1.5; 53.1)
88 (11.9)
30 (15.1)
0.9 (0.5; 1.4)
0.5–1.0
482 (18.3)
212 (10.5)
6.5 (4.7; 9.0)
45 (6.1)
33 (5.6)
2.0 (1.2; 3.3)
1.0–1.5
374 (14.2)
364 (18.1)
15.4 (11.2; 21.1)
61 (8.3)
100 (16.9)
4.7 (3.2; 7.0)
1.5–2.0
315 (12.0)
701 (34.8)
35.2 (25.7; 48.2)
39 (5.3)
186 (31.4)
12.6 (8.4; 18.8)
[2.0
102 (3.9)
598 (29.7)
88.3 (61.9; 126)
9 (1.2)
71 (12.0)
21.7 (10.5; 44.6)
Yes Tobacco smoking status
Tobacco smoking consumption (CSI)
Histology Squamous cell carcinoma
719 (32.4)
100 (16.8)
Adenocarcinoma
706 (34.8)
316 (53.0)
Large cell carcinoma
179 (8.8)
43 (7.2)
123
1118
Cancer Causes Control (2012) 23:1113–1126
Table 1 continued ORa (95 % CI)
Men (n = 3,381) Controls (n = 2,641) n (%)
Cases (n = 2,029) n (%)
ORa (95 % CI)
Women (n = 1,336) Controls (n = 740) n (%)
Cases (n = 596) n (%)
Small cell carcinoma
289 (14.2)
Other specified carcinomas
90 (4.4)
87 (14.6) 33 (5.5)
Other neoplasmsc
31 (1.5)
12 (2.0)
Multiple lung tumors
15 (0.7)
5 (0.8)
a
ORs are adjusted for age (except for variable age) (B50, 50–60, 60–70,[70) and de´partements (Calvados, Doubs-Belfort, He´rault, Ise`re, LoireAtlantique, Manche, Bas-Rhin, Haut-Rhin, Somme, Vende´e) b
Ever held at least one job included on the list of industries and occupations known to be associated with lung cancer (list A) Other specified malignant neoplasms (ICD-O morphology codes: 8033, 8082, 8249, 8263, 8980) or unspecified malignant neoplasm (ICD-O morphology codes: 8000–8005)
c
Table 2 Associations between BMI at interview and sociodemographic characteristics and tobacco smoking status in controls BMI at interview (kg/m2)
All
Control subjects BMI \ 18.5 n (%)
18.5 B BMI \ 25 n (%)
25 B BMI \ 30 n (%)
BMI C 30 n (%)
37 (1.1)
1,328 (39.3)
1,441 (42.7)
573 (17.0)
p
Sex Men
11 (0.4)
977 (37.0)
1,210 (45.8)
443 (16.8)
Women
26 (3.5)
351 (47.6)
231 (31.3)
130 (17.6)
\0.001
Age B55
10 (0.9)
541 (48.1)
439 (39.0)
136 (12.1)
[55
27 (1.2)
787 (34.9)
1,002 (44.5)
437 (19.4)
Educational level CGeneral secondary education
20 (1.6)
596 (47.5)
499 (39.8)
140 (11.2)
\General secondary education
17 (0.8)
716 (34.8)
913 (44.3)
413 (20.1)
\0.001
\0.001
Previous chronic bronchitis No
31 (1.0)
1,263 (39.8)
1,350 (42.6)
527 (16.6)
Yes
5 (2.5)
62 (30.9)
88 (43.8)
46 (22.9)
0.006
Ever held a job in list Aa No
37 (1.2)
1,203 (39.3)
1,301 (42.6)
514 (16.8)
Yes
0 (0.0)
125 (38.6)
140 (43.2)
59 (18.2)
0.18
Tobacco smoking status
a
Never smoker
18 (1.4)
552 (43.6)
507 (40.1)
189 (14.9)
Former smoker
12 (0.9)
469 (33.1)
653 (46.1)
283 (20.0)
Current smoker
7 (1.0)
307 (44.1)
281 (40.4)
101 (14.5)
\0.001
Ever held at least one job included in the list of industries and occupations known to be associated with lung cancer (list A)
moderate and non-significant in women. Furthermore, this relationship was specific to the male and female current smokers, and no association was found in former and never smokers. Interaction between BMI and smoking status in the risk of lung cancer was statistically significant in men (p \ 0.001), not in women (p = 0.2). The analysis, with the weight at age 30, showed similar feature, although the increase in risk for underweight subjects was weaker. When BMI was analyzed as a continuous variable, the risk
123
of lung cancer decreased with an increment of one unit of BMI in both male and female current smokers (aOR = 0.93; 95 % CI 0.90–0.96 and aOR = 0.91; 95 % CI 0.86–0.98, respectively). There was no association among former and never smokers. Table 4 shows the association between lung cancer risk and change in weight during adulthood among people with normal BMI (i.e., 18.5 B BMI \ 25 kg/m2) at age 30. When compared with male current smokers who had a
Cancer Causes Control (2012) 23:1113–1126
Women current smokers
in
in
e ag
Women former smokers
in
ag
in
e
t.-
2 in t.
t.-
30 e ag
27.5 27 26.5 26 25.5 25 24.5 24 23.5 23 22.5 22 21.5 21 20.5 20 30
average BMI
27.5 27 26.5 26 25.5 25 24.5 24 23.5 23 22.5 22 21.5 21 20.5 20
2 in t.
Men former smokers
Time
Time
Women never smokers
Men never smokers
average BMI
27.5 27 26.5 26 25.5 25 24.5 24 23.5 23 22.5 22 21.5 21 20.5 20
27.5 27 26.5 26 25.5 25 24.5 24 23.5 23 22.5 22 21.5 21 20.5 . 2 in t
30 ag e
ag
in
e
t.-
2 in t
30
.
20 t.-
average BMI
t.2
t. in
Time
Time
average BMI
t.
27.5 27 26.5 26 25.5 25 24.5 24 23.5 23 22.5 22 21.5 21 20.5 20 30
average BMI
ag
in
e
t.2
30
average BMI
Men current smokers 27.5 27 26.5 26 25.5 25 24.5 24 23.5 23 22.5 22 21.5 21 20.5 20
Time
in
Fig. 1 Changes in average BMI between age 30, 2 years before the interview (int. - 2), and the time of the interview (int.) for cases and controls
1119
Time Controls Cases
minimal change in weight (\3 kg), the risk of lung cancer was significantly decreased for subjects who gained more than 3 kg and was increased for subjects who lost more
than 3 kg, although non-significantly. We observed a similar but weaker trend among former smokers. No such association was observed among never smokers. The few
123
1120
Cancer Causes Control (2012) 23:1113–1126
Table 3 Associations between lung cancer risk and BMI at age 30 and 2 years before the interview aORa (95 % CI)
Men Controls
aORa 95 % CI
Women
Cases
Controls
Cases
BMI 2 years before the interview (kg/m2) All BMI \ 18.5 18.5 B BMI \ 25 25 B BMI \ 30
28
2.7 (1.2; 6.2)
BMI \ 18.5
18
36
1.5 (0.7; 2.9) 1.0 reference
916
712
1.0 reference
18.5 B BMI \ 25
338
288
1,059
660
0.9 (0.7; 1.1)
25 B BMI \ 30
193
108
0.9 (0.6; 1.2)
221
152
0.8 (0.6; 1.1)
BMI C 30
116
63
0.8 (0.6; 1.3)
174 2,384
124 1,676
0.8 (0.6; 1.0)
Total
665
495
30 B BMI \ 32.5 BMI C 32.5 Total
14
ptrend
0.02
0.20
Current smokers BMI \ 18.5
BMI \ 18.5
3
24
4.1 (1.1; 15.1)
5
29
18.5 B BMI \ 25
229
502
1.0 reference
18.5 B BMI \ 25
63
170
1.0 reference
25 B BMI \ 30
217
317
0.6 (0.5; 0.8)
25 B BMI \ 30
19
41
0.9 (0.4; 2.1)
30 B BMI \ 32.5
33
66
0.7 (0.4; 1.2)
BMI C 30
14
20
0.3 (0.1; 0.7)
BMI C 32.5
36
47
0.5 (0.3; 0.8)
Total
101
260
518
956
Total
\0.001
ptrend
2.0 (0.6; 6.5)
0.007
Former smokers BMI \ 18.5
4
18.5 B BMI \ 25
377
192
1.0 reference
25 B BMI \ 30
542
319
1.1 (0.8; 1.4)
30 B BMI \ 32.5 BMI C 32.5
139 102
83 74
0.9 (0.6; 1.3) 1.0 (0.7; 1.5)
BMI C 30 Total
1,165
672
Total ptrend
2.9 (0.6; 14.5)
BMI \ 18.5
5
4
3
0.5 (0.1; 3.0)
18.5 B BMI \ 25
55
50
1.0 reference
25 B BMI \ 30
36
25
0.7 (0.3; 1.5)
19 114
9 87
0.6 (0.2; 1.7)
0.68
0.42
Never smokers BMI \ 18.5
9
4
1.5 (0.4; 5.3)
310
18
1.0 reference
18.5 B BMI \ 25
219
68
1.0 reference
25 B BMI \ 30
300
24
1.4 (0.7; 2.7)
25 B BMI \ 30
139
42
1.0 (0.6; 1.5)
30 B BMI \ 32.5
49
3
0.7 (0.2; 2.7)
BMI C 30
83
34
12 (0.7; 2.0)
BMI C 32.5
36
3
1.3 (0.4; 4.9)
Total
450
148
695
48
Total
–
–
–
BMI \ 18.5
18.5 B BMI \ 25
ptrend
0.71
0.75
BMI at age 30 years (kg/m2) All BMI \ 18.5
BMI \ 18.5
29
26
1.5 (0.8; 3.0)
18.5 B BMI \ 25
1,644
1,090
1.0 reference
18.5 B BMI \ 25
25 B BMI \ 30 30 B BMI \ 32.5
544 53
401 32
1.1 (0.9; 1.4) 0.6 (0.3; 1.1)
25 B BMI \ 30 BMI C 30
35
9
0.3 (0.1; 0.7)
Total
2,305
1,558
BMI C 32.5 Total ptrend
59
66
495
344
1.0 reference
67 22
35 10
0.9 (0.6; 1.5) 0.7 (0.3; 1.7)
643
455
0.07
1.0 (0.7; 1.6)
0.44
Current smokers BMI \ 18.5
BMI \ 18.5
4
17
2.1 (0.6; 7.1)
18.5 B BMI \ 25
351
638
1.0 reference
18.5 B BMI \ 25
25 B BMI \ 30
128
210
0.9 (0.7; 1.3)
25 B BMI \ 30
123
14
50
74
176
1.4 (0.6; 3.0) 1.0 reference
9
10
0.6 (0.2; 1.9)
Cancer Causes Control (2012) 23:1113–1126
1121
Table 3 continued aORa (95 % CI)
Men Controls
Cases
Controls
30 B BMI \ 32.5
8
15
0.6 (0.2; 1.7)
BMI C 30
BMI C 32.5
8
3
0.2 (0.5; 0.9)
Total
499
883
Total ptrend
aORa 95 % CI
Women Cases
5
2
102
238
0.04
0.1 (0.0; 0.8)
0.014
Former smokers BMI \ 18.5
9
8
9
1.3 (0.4; 4.4)
788
425
1.0 reference
18.5 B BMI \ 25
89
65
1.0 reference
25 B BMI \ 30
274
177
1.2 (0.8; 1.4)
25 B BMI \ 30
11
7
0.7 (0.2; 2.3)
33 17
16 6
0.5 (0.6; 1.3) 0.4 (0.7; 1.5)
BMI C 30 Total
3 111
2 82
1.2 (0.1; 11.1)
1,124
633
30 B BMI \ 32.5 BMI C 32.5 Total ptrend
1.5 (0.6; 14.5)
BMI \ 18.5
12
18.5 B BMI \ 25
0.17
0.68
Never smokers BMI \ 18.5
–
505
27
1.0 reference
18.5 B BMI \ 25
25 B BMI \ 30
142
14
1.7 (0.7; 2.7)
25 B BMI \ 30
12
1
1.3 (0.2; 2.7)
BMI C 30
–
–
– (0.4; 4.9)
Total
659
42
30 B BMI \ 32.5 BMI C 32.5 Total ptrend
–
BMI \ 18.5
–
18.5 B BMI \ 25
0.35
37
8
332
103
0.6 (0.2; 1.3) 1.0 reference
47
18
1.0 (0.5; 1.9)
14
6
1.1 (0.4; 3.1)
430
135 0.39
a
Adjusted for age (B50, 50–60, 60–70, [70), area of residence (Calvados, Doubs-Belfort, He´rault, Ise`re, Loire-Atlantique, Manche, Bas-Rhin, Haut-Rhin, Somme, Vende´e), tobacco smoking consumption: CSI (comprehensive smoking index, continuous variable), educational level (general secondary education, vocational secondary education, primary education), occupational exposure (yes/no), previous chronic bronchitis (yes/no), parental history of lung cancer (yes/no)
number of women who lost weight during adulthood made the results unstable, particularly in the subgroup analysis by smoking status. Among current smokers, we examined the associations between BMI 2 years before the interview and the risk of lung cancer according to histological subtypes (Table 5). No clear inverse trend was observed for small cell carcinoma among both men and women, although underweight subjects displayed a strong increase in risk. However, the results for non-small cell carcinoma subtypes were similar to our previous findings despite having few cases for some subtypes, especially in women.
Discussion In this large, population-based case–control study, we found an inverse, dose-dependent association between BMI measured at least 2 years before the interview and risk of lung cancer in current smokers. The association was stronger when considering BMI 2 years before the interview than when considering BMI at age 30. No association was found in never smokers. Weight gain during adulthood
was associated with a significant decreased risk of lung cancer in male and female current smokers. Cases and controls were stratified by sex and age, using a single control group for both types of cancer (lung and UADT cancers). This design explains the statistically significant difference in the age distributions between cases and controls. However, the large number of subjects in each age group allowed satisfactory adjustment. Collaboration with the French network of cancer registries allowed us to recruit lung cancer cases in almost all of the healthcare establishments in the de´partements covered by the registries. Overall, the participation rate was high and comparable between cases and controls. Because missing weight at age 30 was more frequent among cases, we cannot rule out the possibility of selection bias related to missing data. However, given the low percentage of missing data (7 %) and the similarities between participants with and without missing data, we assume that our findings are not likely due to such a bias. Smoking and high BMI are strong risk factors for cardiovascular mortality, which could occur prior to death from lung cancer. To verify that the lower percentage of overweight subjects among cases was not due to
123
1122
Cancer Causes Control (2012) 23:1113–1126
Table 4 Association between lung cancer risk and weight change between age 30 and 2 years before the interview among men and women with a normal BMI at age 30 (18.5–25 kg/m2) Weight change
Men (18.5 B BMI \ 25 kg/m2) Controls (n = 1,800)
Cases (n = 1,275)
aORa (95 % CI)
Women (18.5 B BMI \ 25 kg/m2) Controls (n = 534)
Cases (n = 396)
aORa (95 % CI)
All Weight gain (at least 3 kg) Minimal weight change (\3 kg) Weight loss (at least 3 kg)
1,081
576
0.6 (0.5; 0.8)
336
175
0.6 (0.4; 0.9)
514
436
1.0 reference
132
146
1.0 reference
1.8 (1.0; 3.1)
0.8 (0.4; 1.8)
38
67
1,633
1,079
Weight gain (at least 3 kg)
209
255
Minimal weight change (\3 kg)
133
324
2.0 (0.8; 5.1)
Total
20
19
488
340
0.4 (0.3; 0.5)
42
64
0.2 (0.1; 0.5)
1.0 reference
26
95
1.0 reference 0.6 (0.1; 3.3)
Current smokers
Weight loss (at least 3 kg)
7
49
349
628
Weight gain (at least 3 kg)
560
301
Minimal weight change (\3 kg)
212
107
Weight loss (at least 3 kg) Total
13 785
Weight gain (at least 3 kg) Minimal weight change (\3 kg)
Total
3
10
71
169
0.7 (0.5; 1.0)
61
38
0.6 (0.2; 1.5)
1.0 reference
23
21
1.0 reference
16 424
1.6 (0.7; 4.0)
5 89
5 64
1.0 (0.2; 4.5)
290
20
2.1 (0.7; 5.9)
229
70
0.9 (0.5; 1.5)
158
5
1.0 reference
81
28
1.0 reference
17
2
3.3 (0.5; 19.5)
12
4
1.7 (0.3; 3.9)
465
27
322
102
Former smokers
Never smokers
Weight loss (at least 3 kg) Total a
Adjusted for age (B50, 50–60, 60–70, [70, area of residence (Calvados, Doubs-Belfort, He´rault, Ise`re, Loire-Atlantique, Manche, Bas-Rhin, Haut-Rhin, Somme, Vende´e), tobacco smoking consumption: CSI (comprehensive smoking index, continuous variable), educational level (general secondary education, vocational secondary education, primary education), occupational exposure (yes/no), previous chronic bronchitis (yes/no), parental history of lung cancer (yes/no)
competitive mortality bias, we repeated the analyses presented in Table 3 and 4 in subjects aged \50 years. Although the number of subjects was smaller, the results for both men (580 cases and 247 controls) and women (145 cases and 120 controls) remained unchanged and significant (data not shown). Underweight patients with advanced-stage non-small cell lung cancer have significantly worse overall survival and time to progression [52]. We can assume that cases that were deceased before any contact could be made (16 %) had a lower BMI, slightly increasing the average BMI at inclusion for our cases, which cannot explain our findings. Furthermore, we are not aware of a possible relation between BMI prior to interview and survival in lung cancer. We used self-reported height and weight and can expect, based on a previous study that compared self-reported BMI and measured BMI [53], a slight overestimation of height and an underestimation of weight in our study. Examination of changes in weight limited this bias, assuming similar errors in the recall of the two weights. Furthermore,
123
a large, French national survey that measured BMI found that 48 % of men and 34 % of women aged from 55 to 75 years were overweight, a result similar to our findings, while the prevalence of obesity was slightly lower in our controls than in the French population (19 vs. 24 %) [54]. Stommel and Schoenborn [53] concluded that self-reported BMI can be used to estimate health risks if corrected for age, sex, and education of the survey respondents, as we did in our analysis. Our results were more pronounced for BMI 2 years before the interview compared with BMI at age 30, a result likely due to non-differential misclassification, which would have diluted the true association. Our analyses were conducted separately by smoking status and accurately adjusted for lifetime tobacco smoking exposure using CSI, which specifically models the effects on lung cancer of intensity, total years of smoking, and time since cessation [49]. Since the frequency of subjects in the lowest BMI category in the overall population is rather low, in men as in women, we observed very few subjects or no subject at all with a BMI \ 18.5 in each smoking status, leading to unstable associations. However, it would be more
Cancer Causes Control (2012) 23:1113–1126
1123
Table 5 Association between lung cancer risk and BMI 2 years before the interview according to histological subtypes among current smokers BMI 2 years before the interview (kg/m2)
Male current smokers Controls
aORa (95 % CI)
Female current smokers
Cases
Controls
aORa 95 % CI
Cases
Small cell carcinoma BMI \ 18.5
BMI \ 18.5
3
3
4.4 (0.7; 31.1)
3
9
6.6 (1.0; 45.4)
18.5 B BMI \ 25
229
63
1.0 reference
18.5 B BMI \ 25
61
33
1.0 reference
25 B BMI \ 30
217
64
0.9 (0.6; 1.4)
25 B BMI \ 30
18
13
3.7 (0.9; 14.9)
30 B BMI \ 32.5
33
19
1.8 (0.9; 3.6)
BMI C 30
14
4
0.1 (0.0; 0.7)
BMI C 32.5
36
12
0.9 (0.4; 2.1)
Total
96
59
518
161
Total ptrend
0.98
0.03
Non-small cell carcinoma BMI \ 18.5
BMI \ 18.5
3
20
3.4 (0.9; 13.1)
5
20
1.9 (0.6; 6.2)
18.5 B BMI \ 25
229
430
1.0 reference
18.5 B BMI \ 25
63
132
1.0 reference
25 B BMI \ 30
217
248
0.6 (0.4; 0.8)
25 B BMI \ 30
19
27
0.8 (0.4; 2.0)
33
47
0.6 (0.3; 1.0)
BMI C 30
14
16
0.3 (0.1; 0.8)
36
34
0.4 (0.2; 0.6)
Total
101
195
518
779
30 B BMI \ 32.5 BMI C 32.5 Total
\0.001
ptrend
0.015
Squamous cell carcinoma BMI \ 18.5
6
5
4
3.2 (0.4; 24.6)
229
168
1.0 reference
18.5 B BMI \ 25
63
31
1.0 reference
25 B BMI \ 30
217
108
0.6 (0.4; 0.9)
25 B BMI \ 30
19
8
3.1 (0.6; 16.2)
33
24
0.7 (0.3; 1.4)
BMI C 30
14
5
0.5 (0.1; 2.8)
36 518
18 324
0.4 (0.2; 0.9)
Total
104
48
30 B BMI \ 32.5 BMI C 32.5 Total ptrend
2.0 (0.3; 12.6)
BMI \ 18.5
3
18.5 B BMI \ 25
0.002
0.50
Adenocarcinoma BMI \ 18.5
BMI \ 18.5
3
11
3.8 (0.9; 15.5)
5
13
1.5 (0.4; 5.7)
18.5 B BMI \ 25
229
195
1.0 reference
18.5 B BMI \ 25
63
77
1.0 reference
25 B BMI \ 30
217
100
0.5 (0.4; 0.7)
25 B BMI \ 30
19
12
0.5 (0.2; 1.5)
33
18
0.5 (0.2; 1.0)
BMI C 30
14
11
0.3 (0.1; 0.9)
36
8
0.2 (0.1; 0.4)
Total
101
113
518
332
30 B BMI \ 32.5 BMI C 32.5 Total
\0.001
ptrend
0.012
Large cell carcinoma BMI \ 18.5
BMI \ 18.5
3
1
4.1 (0.3; 51.2)
18.5 B BMI \ 25
224
42
1.0 reference
25 B BMI \ 30
214
24
0.6 (0.3; 1.1)
33
2
0.2 (0.0; 1.2)
BMI C 30
36 510
4 73
0.3 (0.1; 1.1)
Total
30 B BMI \ 32.5 BMI C 32.5 Total ptrend
0.006
5
2
0.3 (0.0; 4.7)
18.5 B BMI \ 25
49
16
1.0 reference
25 B BMI \ 30
18
4
0.8 (0.1; 5.6)
–
–
–
72
22 0.03
a
Adjusted for age (B50, 50–60, 60–70, [70), area of residence (Calvados, Doubs-Belfort, He´rault, Ise`re, Loire-Atlantique, Manche, Bas-Rhin, Haut-Rhin, Somme, Vende´e), tobacco smoking consumption: CSI (comprehensive smoking index, continuous variable), educational level (general secondary education, vocational secondary education, primary education), occupational exposure (yes/no), previous chronic bronchitis (yes/no), parental history of lung cancer (yes/no)
appropriate to consider our main result as an inverse association between BMI and the risk of lung cancer rather than considering solely an increased risk of lung cancer for
underweight people. Furthermore, and contrary to general thought, we showed in controls weak but positive and significant correlations between CSI and BMI among current
123
1124
(r = 0.11) and former smokers (r = 011). These positive correlations are in agreement with a previous finding reported by Chiolero et al. [35], showing that heavy smokers tend to have a higher BMI than do light smokers, attributed to a clustering of risky behaviors. Because COPD has been associated with decreased BMI [50, 51] and could be associated with lung cancer [3–9], we adjusted all of our analyses for chronic bronchitis. We also repeated the analyses after exclusion of people who reported suffering from chronic bronchitis, and our results remained unchanged (data not shown). This dose-dependent, inverse association was previously reported in case–control [21] and cohort studies [18, 22, 23, 26–28, 32]; similarly, the association between weight loss during adulthood and lung cancer risk has been previously reported, albeit less frequently [21–23]. We categorized BMI according to the WHO classification, consistent with the studies of Andreotti et al. [18] and Kollarova et al. [29], whereas most previous studies [21–28, 31, 32, 34] employed percentiles of the study population. WHO BMI categories clearly isolate under- or overweight people, unlike percentiles. Our use of WHO BMI categories could contribute to the strength of the observed associations. Smoking exposure results in the accumulation of reactive oxygen species (ROS) that can exceed the antioxidant capability of the cell, causing oxidative stress. Urinary levels of 8-hydroxydeoxyguanosine (8-OHdG), a reliable marker of oxidative stress, are increased with tobacco smoking [55–62], notably among lung cancer patients [63]. Interestingly, a significant inverse association, taking into account level of tobacco smoking, has been found between BMI and urinary 8-OHdG [56, 58, 61], particularly among smokers [57, 58]. Carcinogen–DNA adducts are thought to represent the biologically effective dose of a mutagenic carcinogen. Similar to the 8-OHdG results, several studies have found an inverse relation between benzo(a)pyrene (BaP) adduct level and BMI among smokers [64–66], a result that was not observed among non-smokers [67]. These works support the idea that increased body fat impacts adduct levels, probably by affecting the distribution of the carcinogen. BaP, one powerful carcinogenic agent of tobacco smoking, is highly lipid soluble and can deposit in fat stores, removing it from circulation [64]. In conclusion, this work confirms the inverse relation between lung cancer risk and BMI in current smokers and shows, in addition, that weight change during adulthood may be an important marker of risk. In this respect, it would be very interesting to examine the role of lifetime weight variation and the possible release of carcinogens into the circulation when people lose weight. From a public health point of view, if BMI constitutes an individual risk factor for lung cancer, we might consider integrating it into a future risk score.
123
Cancer Causes Control (2012) 23:1113–1126 Acknowledgments The authors thank American Journal Experts for their editorial assistance. This work was supported by the French agency of health security (ANSES); the Fondation de France; the French National Research Agency (ANR); the National Institute of Cancer (INCA); the Foundation for Medical Research (FRM); The French Institute for Public Health Surveillance (InVS); The Health Ministry (DGS); the Organization for the Research on Cancer (ARC); and the French Ministry of work, solidarity and public function (DGT). Conflict of interest
None.
References 1. WHO (2008) GLOBOCAN. Cancer incidence and mortality worldwide in 2008. http://globocan.iarc.fr/ 2. Devesa SS, Bray F, Vizcaino AP, Parkin DM (2005) International lung cancer trends by histologic type: male:female differences diminishing and adenocarcinoma rates rising. Int J Cancer 117(2):294–299 3. Alberg AJ, Ford JG, Samet JM (2007) Epidemiology of lung cancer: ACCP evidence-based clinical practice guidelines. Chest 132(3 Suppl):29S–55S 4. Koshiol J, Rotunno M, Consonni D, Pesatori AC, De Matteis S, Goldstein AM et al (2009) Chronic obstructive pulmonary disease and altered risk of lung cancer in a population-based casecontrol study. PLoS One 4(10):e7380 5. Schwartz AG, Cote ML, Wenzlaff AS, Van Dyke A, Chen W, Ruckdeschel JC et al (2009) Chronic obstructive lung diseases and risk of non-small cell lung cancer in women. J Thorac Oncol 4(3):291–299 6. Wasswa-Kintu S, Gan WQ, Man SF, Pare PD, Sin DD (2005) Relationship between reduced forced expiratory volume in one second and the risk of lung cancer: a systematic review and metaanalysis. Thorax 60(7):570–575 7. Alavanja MC, Brownson RC, Boice JD Jr, Hock E (1992) Preexisting lung disease and lung cancer among nonsmoking women. Am J Epidemiol 136(6):623–632 8. Rodriguez-Roisin R, Soriano JB (2008) Chronic obstructive pulmonary disease with lung cancer and/or cardiovascular disease. Proc Am Thorac Soc 5(8):842–847 9. Tyczynski JE, Bray F, Parkin DM (2003) Lung cancer in Europe in 2000: epidemiology, prevention, and early detection. Lancet Oncol 4(1):45–55 10. Brenner AV, Wang Z, Kleinerman RA, Wang L, Zhang S, Metayer C et al (2001) Previous pulmonary diseases and risk of lung cancer in Gansu Province, China. Int J Epidemiol 30(1):118–124 11. Pairon JC, Brochard P, Le Bourgeois JP, Ruffie´ P (eds) (2000) Les cancers professionnels. Editions Margaux Orange, Paris, France 12. WHO/FAO (2003) Diet, nutrition and the prevention of chronic disease. In: WHO technical reports series 916. Report of a Joint WHO/FAO Expert Consultation, Geneva 13. Calle EE, Kaaks R (2004) Overweight, obesity and cancer: epidemiological evidence and proposed mechanisms. Nat Rev Cancer 4(8):579–591 14. Ursin G, Longnecker MP, Haile RW, Greenland S (1995) A meta-analysis of body mass index and risk of premenopausal breast cancer. Epidemiology 6(2):137–141 15. Smith M, Zhou M, Whitlock G, Yang G, Offer A, Hui G et al (2008) Esophageal cancer and body mass index: results from a prospective study of 220,000 men in China and a meta-analysis of published studies. Int J Cancer 122(7):1604–1610
Cancer Causes Control (2012) 23:1113–1126 16. Kabat GC, Chang CJ, Wynder EL (1994) The role of tobacco, alcohol use, and body mass index in oral and pharyngeal cancer. Int J Epidemiol 23(6):1137–1144 17. Franceschi S, Dal Maso L, Levi F, Conti E, Talamini R, La Vecchia C (2001) Leanness as early marker of cancer of the oral cavity and pharynx. Ann Oncol 12(3):331–336 18. Andreotti G, Hou L, Beane Freeman LE, Mahajan R, Koutros S, Coble J et al (2010) Body mass index, agricultural pesticide use, and cancer incidence in the Agricultural Health Study cohort. Cancer Causes Control 21(11):1759–1775 19. Chyou PH, Nomura AM, Stemmermann GN (1994) A prospective study of weight, body mass index and other anthropometric measurements in relation to site-specific cancers. Int J Cancer 57(3):313–317 20. Drinkard CR, Sellers TA, Potter JD, Zheng W, Bostick RM, Nelson CL et al (1995) Association of body mass index and body fat distribution with risk of lung cancer in older women. Am J Epidemiol 142(6):600–607 21. Goodman MT, Wilkens LR (1993) Relation of body size and the risk of lung cancer. Nutr Cancer 20(2):179–186 22. Kabat GC, Kim M, Hunt JR, Chlebowski RT, Rohan TE (2008) Body mass index and waist circumference in relation to lung cancer risk in the women’s health initiative. Am J Epidemiol 168(2):158–169 23. Kabat GC, Miller AB, Rohan TE (2007) Body mass index and lung cancer risk in women. Epidemiology 18(5):607–612 24. Kabat GC, Wynder EL (1992) Body mass index and lung cancer risk. Am J Epidemiol 135(7):769–774 25. Kanashiki M, Sairenchi T, Saito Y, Ishikawa H, Satoh H, Sekizawa K (2005) Body mass index and lung cancer: a case-control study of subjects participating in a mass-screening program. Chest 128(3):1490–1496 26. Kark JD, Yaari S, Rasooly I, Goldbourt U (1995) Are lean smokers at increased risk of lung cancer? The Israel Civil Servant Cancer Study. Arch Intern Med 155(22):2409–2416 27. Knekt P, Heliovaara M, Rissanen A, Aromaa A, Seppanen R, Teppo L et al (1991) Leanness and lung-cancer risk. Int J Cancer 49(2):208–213 28. Koh WP, Yuan JM, Wang R, Lee HP, Yu MC (2010) Body mass index and smoking-related lung cancer risk in the Singapore Chinese Health Study. Br J Cancer 102(3):610–614 29. Kollarova H, Machova L, Horakova D, Cizek L, Janoutova G, Janout V (2008) Is obesity a preventive factor for lung cancer? Neoplasma 55(1):71–73 30. Kubik A, Zatloukal P, Boyle P, Robertson C, Gandini S, Tomasek L et al (2001) A case-control study of lung cancer among Czech women. Lung Cancer 31(2–3):111–122 31. Kubik AK, Zatloukal P, Tomasek L, Pauk N, Havel L, Krepela E et al (2004) Dietary habits and lung cancer risk among nonsmoking women. Eur J Cancer Prev 13(6):471–480 32. Olson JE, Yang P, Schmitz K, Vierkant RA, Cerhan JR, Sellers TA (2002) Differential association of body mass index and fat distribution with three major histologic types of lung cancer: evidence from a cohort of older women. Am J Epidemiol 156(7):606–615 33. Tulinius H, Sigfusson N, Sigvaldason H, Bjarnadottir K, Tryggvadottir L (1997) Risk factors for malignant diseases: a cohort study on a population of 22,946 Icelanders. Cancer Epidemiol Biomarkers Prev 6(11):863–873 34. Rauscher GH, Mayne ST, Janerich DT (2000) Relation between body mass index and lung cancer risk in men and women never and former smokers. Am J Epidemiol 152(6):506–513 35. Chiolero A, Faeh D, Paccaud F, Cornuz J (2008) Consequences of smoking for body weight, body fat distribution, and insulin resistance. Am J Clin Nutr 87(4):801–809
1125 36. Hamilton W, Peters TJ, Round A, Sharp D (2005) What are the clinical features of lung cancer before the diagnosis is made? A population based case-control study. Thorax 60(12):1059–1065 37. Khalid U, Spiro A, Baldwin C, Sharma B, McGough C, Norman AR et al (2007) Symptoms and weight loss in patients with gastrointestinal and lung cancer at presentation. Support Care Cancer 15(1):39–46 38. Buccheri G, Ferrigno D (2004) Lung cancer: clinical presentation and specialist referral time. Eur Respir J 24(6):898–904 39. Corner J, Hopkinson J, Fitzsimmons D, Barclay S, Muers M (2005) Is late diagnosis of lung cancer inevitable? Interview study of patients’ recollections of symptoms before diagnosis. Thorax 60(4):314–319 40. Luce D, Stucker I (2011) Investigation of occupational and environmental causes of respiratory cancers (ICARE): a multicenter, population-based case-control study in France. BMC Public Health 11:928 41. WHO (World Health Organization) (2000) International classification of diseases for oncology, 3rd edn. WHO, Geneva 42. Rothman KJ, Greenland S (2008) Modern epidemiology, 3rd edn. Lippincott Williams and Wilkins, Philadelphia 43. WHO-IARC (2007) IARC monographs on the evaluation of carcinogenic risks to humans. Human papillomaviruses, vol 90. WHO-IARC, Lyon, France 44. WHO-IARC (2012) IARC monographs on the evaluation of carcinogenic risks to humans. Chemical agents and related occupations, vol 100. WHO-IARC, Lyon, France 45. Consonni D, De Matteis S, Lubin JH, Wacholder S, Tucker M, Pesatori AC et al (2010) Lung cancer and occupation in a population-based case-control study. Am J Epidemiol 171(3):323–333 46. Ahrens W, Merletti F (1998) A standard tool for the analysis of occupational lung cancer in epidemiologic studies. Int J Occup Environ Health 4(4):236–240 47. WHO (2000) Obesity: preventing and managing the global epidemic. Report of a WHO consultation. WHO technical report series 894. World Health Organization, Geneva 48. WHO Global database on body mass index. http://apps. who.int/bmi/index.jsp?introPage=intro_3.html 49. Leffondre K, Abrahamowicz M, Xiao Y, Siemiatycki J (2006) Modelling smoking history using a comprehensive smoking index: application to lung cancer. Stat Med 25(24):4132–4146 50. Lamprecht B, McBurnie MA, Vollmer WM, Gudmundsson G, Welte T, Nizankowska-Mogilnicka E et al (2011) COPD in never-smokers: results from the population-based burden of obstructive lung disease study. Chest 139(4):752–763 51. Mannino DM, Buist AS (2007) Global burden of COPD: risk factors, prevalence, and future trends. Lancet 370(9589):765–773 52. Mandrekar SJ, Schild SE, Hillman SL, Allen KL, Marks RS, Mailliard JA et al (2006) A prognostic model for advanced stage nonsmall cell lung cancer. Pooled analysis of North Central Cancer Treatment Group trials. Cancer 107(4):781–792 53. Stommel M, Schoenborn CA (2009) Accuracy and usefulness of BMI measures based on self-reported weight and height: findings from the NHANES and NHIS 2001–2006. BMC Public Health 9:421 54. Unite´ de surveillance et d’e´pide´miologie nutritionnelle (Usen) (2007) E´tude nationale nutrition sante´ (ENNS 2006)—situation nutritionnelle en France en 2006 selon les indicateurs d’objectif et les repe`res du Programme national nutrition sante´ (PNNS). Institut de veille sanitaire, Universite´ de Paris 13, Conservatoire national des arts et me´tiers 55. Ichiba M, Yamada S, Ishii K, Gonda K, Murai R, Shimomura T et al (2007) Significance of urinary excretion of 8-hydroxy-20 deoxyguanosine in healthy subjects and liver disease patients. Hepatogastroenterology 54(78):1736–1740
123
1126 56. Kasai H, Iwamoto-Tanaka N, Miyamoto T, Kawanami K, Kawanami S, Kido R et al (2001) Life style and urinary 8-hydroxydeoxyguanosine, a marker of oxidative DNA damage: effects of exercise, working conditions, meat intake, body mass index, and smoking. Jpn J Cancer Res 92(1):9–15 57. Loft S, Vistisen K, Ewertz M, Tjonneland A, Overvad K, Poulsen HE (1992) Oxidative DNA damage estimated by 8-hydroxydeoxyguanosine excretion in humans: influence of smoking, gender and body mass index. Carcinogenesis 13(12):2241–2247 58. Mizoue T, Kasai H, Kubo T, Tokunaga S (2006) Leanness, smoking, and enhanced oxidative DNA damage. Cancer Epidemiol Biomarkers Prev 15(3):582–585 59. Pilger A, Germadnik D, Riedel K, Meger-Kossien I, Scherer G, Rudiger HW (2001) Longitudinal study of urinary 8-hydroxy-20 deoxyguanosine excretion in healthy adults. Free Radic Res 35(3):273–280 60. Pilger A, Rudiger HW (2006) 8-hydroxy-20 -deoxyguanosine as a marker of oxidative DNA damage related to occupational and environmental exposures. Int Arch Occup Environ Health 80(1):1–15 61. Sakano N, Wang DH, Takahashi N, Wang B, Sauriasari R, Kanbara S et al (2009) Oxidative stress biomarkers and lifestyles in Japanese healthy people. J Clin Biochem Nutr 44(2):185–195
123
Cancer Causes Control (2012) 23:1113–1126 62. Tamae K, Kawai K, Yamasaki S, Kawanami K, Ikeda M, Takahashi K et al (2009) Effect of age, smoking and other lifestyle factors on urinary 7-methylguanine and 8-hydroxydeoxyguanosine. Cancer Sci 100(4):715–721 63. Yano T, Shoji F, Baba H, Koga T, Shiraishi T, Orita H et al (2009) Significance of the urinary 8-OHdG level as an oxidative stress marker in lung cancer patients. Lung Cancer 63(1):111–114 64. Godschalk RW, Feldker DE, Borm PJ, Wouters EF, van Schooten FJ (2002) Body mass index modulates aromatic DNA adduct levels and their persistence in smokers. Cancer Epidemiol Biomarkers Prev 11(8):790–793 65. Palli D, Vineis P, Russo A, Berrino F, Krogh V, Masala G et al (2000) Diet, metabolic polymorphisms and DNA adducts: the EPIC-Italy cross-sectional study. Int J Cancer 87(3):444–451 66. Rundle A, Madsen A, Orjuela M, Mooney L, Tang D, Kim M et al (2007) The association between benzo[a]pyrene-DNA adducts and body mass index, calorie intake and physical activity. Biomarkers 12(2):123–132 67. Peluso M, Airoldi L, Munnia A, Colombi A, Veglia F, Autrup H et al (2008) Bulky DNA adducts, 4-aminobiphenyl-haemoglobin adducts and diet in the European prospective investigation into cancer and nutrition (EPIC) prospective study. Br J Nutr 100(3):489–495