Relation of Height and Body Mass Index to Renal Cell Carcinoma in Two Million Norwegian Men and Women

Transcription

1 American Journal of Epidemiology Copyright 2004 by the Johns Hopkins Bloomberg School of Public Health All rights reserved Vol. 160, No. 12 Printed in U.S.A. DOI: /aje/kwh345 Relation of Height and Body Mass Index to Renal Cell Carcinoma in Two Million Norwegian Men and Women Tone Bjørge 1, Steinar Tretli 2, and Anders Engeland 3 1 Department of Pathology, Norwegian Radium Hospital, Oslo, Norway. 2 Cancer Registry of Norway, Institute of Population-based Cancer Research, Oslo, Norway. 3 Division of Epidemiology, Norwegian Institute of Public Health, Oslo, Norway. Received for publication July 3, 2004; accepted for publication September 16, A positive association between body mass index (BMI) and renal cell carcinoma (RCC) has been observed. The association between height and RCC has been less clear. The authors explored these relations in a very large Norwegian cohort. Height and weight were measured in two million Norwegian men and women aged years during During follow-up, 6,453 cases of RCC were registered in the national cancer database. Measurements were also performed in 227,000 adolescents aged years, and 154 cases of RCC were registered. Relative risks for RCC were estimated using Cox proportional hazards regression. The risk of RCC increased with increasing BMI among both adults and adolescents. Among adults, the relative risk associated with a one-unit increase in BMI was 1.05 (95% confidence interval (CI): 1.04, 1.06) in both sexes. The relative risk associated with a 10-cm increase in height was 1.19 (95% CI: 1.13, 1.26) in men and 1.17 (95% CI: 1.09, 1.26) in women. In a subgroup analysis, the relation between BMI and RCC was most pronounced in men and women who were never smokers, and the relation between height and RCC was confined to ever smokers. The authors conclude that elevated BMIs are associated with RCC risk in both males and females across a wide age range. body height; body mass index; carcinoma, renal cell; kidney neoplasms Abbreviations: BMI, body mass index; CI, confidence interval; RCC, renal cell carcinoma. Overweight or obesity has been linked to increased risk of total mortality (1) and to increased risk of death from several site-specific cancers (2), including kidney cancer. It is estimated that in the year 2000, kidney cancer was responsible for 1.9 percent of new cancer cases and 1.5 percent of cancer deaths worldwide (3). The incidence of kidney cancer is highest in North America and Europe (4). In 2000, the ageadjusted (world standard population) incidence of kidney cancer, excluding the renal pelvis, was 8 per 100,000 personyears and 4 per 100,000 person-years in men and women, respectively, and constituted 2.5 percent and 1.6 percent of the (age-adjusted) cancer incidence in Norway (5). Most kidney cancers ( 80 percent) are renal cell carcinomas (RCCs) (6). Cigarette smoking is an established causal factor for the development of RCC (7). Obesity has also been suggested to be a risk factor for RCC, and a number of studies, mostly case-control studies, have explored this relation (7, 8). The relation between obesity and RCC has been most pronounced in women (7, 9, 10). However, a recent review found a positive relation between body mass index (BMI) and RCC that was equally strong for both sexes (8). High blood pressure or use of medication for hypertension may also play a role, although the mechanism is unknown (7). Age and smoking have been considered to be the most important confounding factors in the relation between obesity and RCC (8). Few studies have explored the association between height and risk of RCC, and the results have been inconsistent (11). Our aim in the present study was to explore the associations between height and BMI and the risk of RCC in a very large Norwegian cohort of both men and women with a long follow-up period. Correspondence to Dr. Anders Engeland, Division of Epidemiology, Norwegian Institute of Public Health, P.O. Box 4404 Nydalen, N-0403 Oslo, Norway ( anders.engeland@fhi.no). 1168

2 Relation of Height and BMI to RCC 1169 MATERIALS AND METHODS We have previously described the study population and the methods used in the present study (12 14). Here we give a brief description only. Study population In a number of health surveys carried out during the period , height and weight were measured in a standardized way by a trained staff among 2,001,719 persons (963,709 men and 1,038,010 women) aged years (the adult cohort ). In a tuberculosis screening program conducted in , height and weight were measured in 227,224 young people (115,270 boys and 111,954 girls) aged years (the adolescent cohort ). Some persons were included in both cohorts. Consequently, the total number of study subjects was 2,089,182. Deaths, emigrations, and cases of kidney cancer (International Classification of Diseases, Seventh Revision, code 180) observed in these cohorts were identified by linkage to the Death Registry at Statistics Norway (15) and to the Cancer Registry of Norway (5). The two registries are population-based and cover the entire Norwegian population. A unique 11-digit identification number assigned to all persons living in Norway after 1960 simplified the linkage process. In the present study, only histologically verified diagnoses of RCC were included. Persons who received a diagnosis of RCC prior to the measurement of height and weight were excluded (409 persons in the adult cohort). In the analyses, the subjects were followed from the date of measurement to the date of RCC diagnosis, emigration, age 100 years, or death or until December 31, Altogether, 2,001,310 persons (963,485 men and 1,037,825 women) were eligible for study in the adult cohort. A very small number of these subjects (43 men and 37 women) were lost to follow-up. In the adolescent cohort, only three boys out of 227,224 boys and girls were lost to follow-up. Some of the health surveys carried out during included questions about smoking habits (self-administered questionnaire). A subanalysis was performed including only those who had answered questions on smoking habits. The study subjects were divided into never, former, and current smokers. Information on smoking habits was available for 647,093 persons (322,746 men and 324,347 women) included in the adult cohort, after exclusions. The subanalysis was performed with the same exclusion criteria and follow-up procedure as the main analysis. Statistical methods Cox proportional hazards regression models (16) with time since height and weight measurement used as the time variable were fitted to obtain relative risk estimates for RCC. In the analyses, categorized variables for age at measurement, year of birth, BMI (weight (kg)/height (m) 2 ), and height were included. In the adult cohort, BMI was categorized using the World Health Organization categorization (17): <18.5, underweight; , normal; , pre-obese/overweight; and 30.0, obese. In the adolescent cohort, BMI was categorized using growth percentiles (low, <25th; medium, 25th 74th; high, 75th 84th; very high, 85th) from a US reference population, following the guidelines of the Centers for Disease Control and Prevention and the National Center for Health Statistics (18, 19). We performed tests for trend in the risk of RCC by BMI and height by conducting analyses including BMI and height, respectively, as continuous variables. To explore whether BMI had a differential impact on the incidence of RCC for different age groups, we performed analyses stratified on both age at measurement and attained age. These analyses did not add information to the main analysis. All of the above-mentioned analyses were performed with the statistical software package SPSS (20). The results are presented as relative risks and 95 percent confidence intervals. The hazard function for RCC by BMI in the Cox model was estimated using penalized spline functions in S-Plus (21), with 4 degrees of freedom. RESULTS Adult cohort The 2,001,230 persons (963,442 men and 1,037,788 women) included in the present study were followed for an average of 23 years (range, 0 40 years), which constituted 45.7 million person-years of follow-up (table 1). The mean age at measurement was 44 years. During follow-up, 6,453 cases of RCC were diagnosed among the study subjects. The mean age at diagnosis was 66 years, and measurements were made an average of 18 years prior to diagnosis. More than twice-as-large proportions of both underweight and obesity were observed in women as in men. The proportions of underweight were 0.7 percent and 2.0 percent in men and women, respectively, while the proportions of obesity were 6 percent and 13 percent. An increase in RCC risk was seen in both sexes with increasing BMI and height (table 2). Adjusting the BMI results for height and vice versa did not change the results. The relative risk associated with a one-unit increase in BMI was 1.05 (95 percent confidence interval (CI): 1.04, 1.06) in both sexes. For a person with a height of 176 cm, a one-unit increase in BMI is 3 kg. Splitting the upper BMI category gave us relative risks of 1.89 (95 percent CI: 1.22, 2.94) and 2.76 (95 percent CI: 0.89, 8.56) among men with BMIs of and 40, respectively, compared with men of normal weight. The corresponding figures for women were 2.13 (95 percent CI: 1.75, 2.60) and 2.59 (95 percent CI: 1.81, 3.70). The proportions of persons with BMIs of and 40 were 0.5 percent and 0.1 percent, respectively, in men and 2.4 percent and 0.6 percent, respectively, in women. The mean BMIs in these categories were 37 and 43, respectively, in both sexes. In stratified analyses separating persons born before 1930 from persons born in 1930 or later, we observed a somewhat higher increase in the relative risk of RCC by increasing BMI in persons born from 1930 onwards than in persons born before The relative risk associated with a 10-cm increase in height was 1.19 (95 percent CI: 1.13, 1.26) in men and 1.17 (95

3 1170 Bjørge et al. TABLE 1. Numbers of observed cases of renal cell carcinoma (RCC), person-years of follow-up, and overall rates of RCC among persons aged years at study entry, Norway, Variable No. of cases Men (n = 963,442) Women (n = 1,037,788) Person-years of follow-up RCC rate* No. of cases Person-years of follow-up Time (years) since height and weight measurement ,669, ,064, ,109, ,546, ,478, ,982, ,939, ,522, ,505, ,125, ,061, ,674, ,234, ,748, Age (years) at height and weight measurement ,714, ,432, ,511, ,120, ,319 6,287, ,761, ,011 3,427, ,408, ,696, ,413, , , Year of birth < , , ,772, ,548, ,027 3,485, ,494, ,183 4,900, ,447, ,208, ,707, ,489, ,944, ,673, ,836,470 2 Body mass index < , , ,908 12,254, ,061 14,004, ,638 7,634, ,262, , ,863, Height (cm) < ,105 7 <160/ , ,912, ,269, ,469 14,200, / 170 2,187 11,474, ,189, ,015 6,112, RCC rate* Total 3,821 20,998, ,632 24,664, * Number of cases per 100,000 person-years. Weight (kg)/height (m) 2. Split categories pertain to men and women, respectively. percent CI: 1.09, 1.26) in women. The estimated relative risks increased monotonically by increasing height across the height categories. The relative risk in tall ( cm) versus short ( cm) men was 1.86 (95 percent CI: 1.18, 2.93). The relative risk in tall ( cm) versus short ( cm) women was 2.21 (95 percent CI: 1.49, 3.29). To exclude the possibility that, at the time of BMI measurement, weight was influenced by the presence of

4 Relation of Height and BMI to RCC 1171 TABLE 2. Relative risk of renal cell carcinoma in Cox regression analyses among persons aged years at study entry, Norway, * Men (3,821 cases) Women (2,632 cases) Variable RR 95% CI RR 95% CI Body mass index < , , , , , , 2.06 p for trend <0.001 <0.001 Height (cm)# < , 0.77 <160/ , , , / , , 1.19 p for trend <0.001 <0.001 * Age at height and weight measurement (six categories) and birth cohort (seven categories) were included in the model in addition to either body mass index or height. RR, relative risk; CI, confidence interval. Weight (kg)/height (m) 2. Referent. Body mass index and height, respectively, were included in the trend test as continuous variables. # Split categories pertain to men and women, respectively. undiagnosed RCC, we analyzed the data after omitting the first 5 years of follow-up. The number of cases was reduced by 600, but similar results were found. That is, the largest change was that the relative risk in obese women changed from 1.85 (95 percent CI: 1.66, 2.06) to 1.92 (95 percent CI: 1.71, 2.14). The association between BMI and RCC was also explored in detail using penalized spline functions (figure 1). The incidence of RCC increased steadily with BMI in both sexes. In the subcohort with known smoking habits, we performed an analysis similar to that shown in table 2, including 7,816,818 person-years (mean follow-up of 12 years). The mean age at measurement in the subcohort was 42 years. Compared with the entire cohort, the persons in the subcohort were from younger birth cohorts. About 50 percent of the entire cohort and about 90 percent of the subcohort was born in 1930 or later. In the subcohort, a total of 530 histologically verified cases of RCC were diagnosed in men and 251 in women. Mean ages at diagnosis were 58 years and 59 years in men and women, respectively. Similar results were obtained in analysis including this subcohort only (table 3) as in the main analysis. The number of cases in the lowest categories of BMI and height were very low (one and three cases in the lowest category of BMI in men and women, respectively, and two cases and one case in the lowest height category). Including the three-category smoking variable did not change the estimates in the subcohort. However, stratified analyses in never, former, and current smokers revealed a somewhat more pronounced increase in the risk of RCC with increasing BMI in never smokers than in former or current smokers (table 3). The functions for risk of RCC by BMI among never smokers in the subcohort are shown in figure 2. The increasing risk of RCC with increasing height seemed to be confined to former or current smokers. Adolescent cohort The 227,221 adolescents (115,267 boys and 111,954 girls) included in the present study were followed for an average of 33 years (range, 0 40 years), constituting 7.5 million personyears of follow-up. The mean age at measurement was 17 years. During follow-up, 154 RCC cases were diagnosed among the study subjects. The mean age at diagnosis was 45 years, and measurements were made, on average, 28 years prior to diagnosis. The proportions of low BMI were 24 percent and 16 percent in boys and girls, respectively, while the proportions of high and very high BMI were 8 percent and 5 percent, respectively, in boys and 11 percent and 8 percent in girls. An increase in risk was seen with increasing BMI in both sexes, especially in boys (table 4).

5 1172 Bjørge et al. FIGURE 1. Log relative risk of renal cell carcinoma from penalized spline functions (4 df) among persons aged years at study entry, according to body mass index, Norway, Results were adjusted for birth cohort and age at height and weight measurement. Dashed lines, 95% confidence interval. DISCUSSION In this study, a Norwegian cohort of more than two million persons was followed prospectively with regard to RCC risk for an average of 23 years. More than 6,500 carcinomas were observed, and the risk of RCC increased with both BMI and height in both sexes. The risk of RCC increased with both BMI in adulthood and BMI in adolescence. Restricting the analysis to never smokers in a subcohort changed the results somewhat. Never smokers had the largest increase in risk of RCC with BMI. The positive relation between height and RCC was confined to ever smokers. BMI is a simple and commonly used measure of body mass relative to height. However, BMI does not distinguish between weight associated with muscles and weight associated with fat (17). Although BMI is a crude indicator for percentage of body fat, there is generally very good correlation between BMI and the percentage of body fat in large populations (22). The categorization used in the present study (17) is recommended by the World Health Organization. The categorization is independent of age and is the same for both sexes. Nevertheless, the percentage of body fat mass is different for persons of different ages and different between the sexes. Hence, one should interpret results from studies including persons within a large age range, like the present one, with caution. We conducted separate analyses for men and women and for intervals of attained age and age

6 Relation of Height and BMI to RCC 1173 TABLE 3. Relative risk of renal cell carcinoma in Cox regression analyses in a subcohort of participants with known smoking habits among persons aged years at study entry, Norway, * Whole subcohort Never smokers Former smokers Current smokers RR 95% CI RR 95% CI RR 95% CI RR 95% CI Men (530 cases) (106 cases) (149 cases) (275 cases) Body mass index < , , , , , , , , , , , , 2.36 p for trend < Height (cm) < , , , , , , , , , , , , 1.92 p for trend Women (251 cases) (108 cases) (34 cases) (109 cases) Body mass index < , , , , , , , , , , , , 2.16 p for trend Height (cm) < , , , , , , , , , , , , 2.01 p for trend * Age at height and weight measurement and birth cohort were included in the model in addition to either body mass index or height. Analyses included the whole subcohort, never smokers only, former smokers only, and current smokers only, respectively. RR, relative risk; CI, confidence interval. Weight (kg)/height (m) 2. Referent. Body mass index and height, respectively, were included in the trend test as continuous variables. at measurement. These analyses did not indicate that the crude impression from the main analyses, which were adjusted for age at measurement, was wrong. In addition, total mortality in the present cohort showed the same overall pattern by BMI across age groups (23). In the review by Bergström et al. (8), the largest study was a population-based case-control study including 1,700 RCC cases (24). The majority of the studies included were casecontrol studies that relied on retrospective self-reporting of weight and height. One of the largest cohort studies on this topic (25) included 759 RCCs diagnosed in 360,000 men. In the present study, we could rely on standardized weight and height measurements taken in two cohorts, one with two million men and women measured at ages years and one with 227,000 persons measured in adolescence. The measurements were performed a long time before diagnosis. More than 6,500 RCC cases were diagnosed in these cohorts during follow-up. In a subcohort with known smoking habits, 768 RCC cases occurred. Subsequently, the diagnoses in our study were ascertained through a populationbased cancer registry, and the measurements and diagnoses were linked using a personal identification number. Bergström et al. (8) calculated a summary relative risk of RCC per one-unit increase in BMI of 1.07 (95 percent CI: 1.05, 1.09). They noted that studies controlling for smoking revealed a slightly higher relative risk than studies that did not control for smoking. In the present study, not controlling for smoking, we found a relative risk of 1.05 (95 percent CI: 1.04, 1.06) per one-unit increase in BMI in both sexes. Estimation of fluctuations in the hazard of RCC by BMI using

7 1174 Bjørge et al. FIGURE 2. Log relative risk of renal cell carcinoma from penalized spline functions (4 df) among never smokers aged years at study entry (in a subcohort of persons with known smoking habits), according to body mass index, Norway, Results were adjusted for birth cohort and age at height and weight measurement. Dashed lines, 95% confidence interval. spline curves indicated that there was no threshold value in the association between BMI and RCC. The risk of RCC increased steadily with increasing BMI. In the subcohort with known smoking habits, adjustment for smoking habits did not change these relations. However, we observed that the relation between BMI and risk of RCC was stronger for never smokers than for ever smokers: Relative risks were 1.09 (95 percent CI: 1.03, 1.15) and 1.07 (95 percent CI: 1.03, 1.11) in male and female never smokers, respectively. Since smoking increases the risk of RCC and is inversely related to BMI, this effect was expected (8). A slightly stronger association between kidney cancer and BMI in never smokers than in former and current smokers was recently observed in a large population-based case-control study (26). The association between BMI and risk of RCC may be explained through several hormonal mechanisms (8, 9). Earlier studies found a stronger association between obesity and RCC risk in women than in men. Genderspecific fat distributions and hormone levels might explain some of the difference (22). However, the stronger association observed in women might be partly explained by the present findings of a stronger relation between BMI and RCC risk in never smokers than in ever smokers. Previously,

8 Relation of Height and BMI to RCC 1175 TABLE 4. Relative risk of renal cell carcinoma in Cox regression analyses among persons aged years at study entry, Norway, * Body mass index, No. of cases Boys (n = 115,267) Girls (n = 111,954) RR 95% CI No. of cases RR 95% CI Low , , 2.17 Medium High , , 3.34 Very high , , 3.85 * Age at measurement (two categories) and birth cohort (two categories) were included in the model in addition to body mass index. Weight (kg)/height (m) 2. Defined by the percentiles in a US reference population (19): low, <25th; medium, 25th 74th; high, 75th 84th; very high, 85th. RR, relative risk; CI, confidence interval. Referent. a larger proportion of women than of men were never smokers. On the contrary, Bergström et al. (8) suggested that the sex difference found in some studies might be explained by the differences in the distribution of BMIs among men and women within broad categories. Reported results on the association between height and RCC have been inconsistent (11). However, the studies reviewed by Gunnell et al. (11), mainly case-control studies, were relatively small. One (27) of the two cohort studies included in the review found an increasing risk by height. Tulinius et al. (27) found that the relative risk associated with a 1-cm increase in height was 1.05 (95 percent CI: 1.02, 1.08) in men. In a more recent, large cohort study of Swedish men, no association between height and RCC was observed (25). In the present study, we found a slightly lower increase in risk per 1-cm increase in height (for both sexes, relative risk = 1.02, 95 percent CI: 1.01, 1.02) than did Tulinius et al. when smoking habits were not taken into consideration. However, we found an increased risk by height in both sexes. Moreover, because of our large study size, we were able to examine the risk in categories of height. We found a monotonically increasing risk by increasing height in both sexes. The tallest men and women had about twice the risk of the shortest men and women. This relation may pertain to genes or early-life exposures (11). In a subanalysis, it seemed that the association between height and RCC was confined to ever smokers. If this is a true relation, this may explain the inconsistent findings on the relation between height and risk of RCC; the proportion of smokers differs between populations. However, this was not an expected finding, and it should be replicated by other investigators before firm conclusions are drawn. In this study, we observed a positive association between BMI in adolescence and risk of RCC. However, we did not have enough persons with measurements taken in both adolescence and adulthood to explore whether this association was due to BMI in adolescence or BMI in adulthood. Since high BMI in adolescence is predictive for high BMI in adulthood (28), possible effects of these two factors on the risk of RCC are difficult to separate. Persons with an already-high BMI at the age of 20 years who gained further weight between the ages of 20 and 50 years were previously reported to have an increased risk of RCC (29). In summary, in this large Norwegian cohort, the risk of RCC increased with BMI for both sexes, being most pronounced in never smokers. The positive association between BMI and risk of RCC was also observed in persons with measurements taken in adolescence. In addition, we found an increasing risk of RCC by increasing height that seemed to be confined to ever smokers. ACKNOWLEDGMENTS The authors are grateful to all those who, over a period of almost 40 years, collected the data used in the present study. These persons were affiliated with the former National Health Screening Service, the Nord-Trøndelag Health Survey (HUNT), the Hordaland Health Survey (HUSK), and the Tromsø Study. REFERENCES 1. Calle EE, Thun MJ, Petrelli JM, et al. Body-mass index and mortality in a prospective cohort of U.S. adults. N Engl J Med 1999;341: Calle EE, Rodriguez C, Walker-Thurmond K, et al. Overweight, obesity, and mortality from cancer in a prospectively studied cohort of U.S. adults. N Engl J Med 2003;348: Parkin DM, Bray FI, Devesa SS. Cancer burden in the year 2000: the global picture. Eur J Cancer 2001;37(suppl 8):S Parkin DM, Pisani P, Ferlay J. Estimates of the worldwide incidence of 25 major cancers in Int J Cancer 1999;80: Cancer Registry of Norway. Cancer in Norway Oslo, Norway: Cancer Registry of Norway, 2002.

9 1176 Bjørge et al. 6. Bray F, Sankila R, Ferlay J, et al. Estimates of cancer incidence and mortality in Europe in Eur J Cancer 2002;38: McLaughlin JK, Lipworth L. Epidemiologic aspects of renal cell cancer. Semin Oncol 2000;27: Bergström A, Hsieh CC, Lindblad P, et al. Obesity and renal cell cancer a quantitative review. Br J Cancer 2001;85: Moyad MA. Obesity, interrelated mechanisms, and exposures and kidney cancer. Semin Urol Oncol 2001;19: Martel CL, Lara PN Jr. Renal cell carcinoma: current status and future directions. Crit Rev Oncol Hematol 2003;45: Gunnell D, Okasha M, Davey Smith G, et al. Height, leg length, and cancer risk: a systematic review. Epidemiol Rev 2001;23: Engeland A, Tretli S, Bjørge T. Height and body mass index in relation to esophageal cancer: 23-year follow-up of two million Norwegian men and women. Cancer Causes Control 2004;15: Engeland A, Tretli S, Bjørge T. Height, body mass index, and ovarian cancer: a follow-up of 1.1 million Norwegian women. J Natl Cancer Inst 2003;95: Engeland A, Bjørge T, Søgaard AJ, et al. Body mass index in adolescence in relation to total mortality: 32-year follow-up of 227,000 Norwegian boys and girls. Am J Epidemiol 2003;157: Statistics Norway. We live longer than ever before. Oslo, Norway: Statistics Norway, (World Wide Web URL: Cox DR, Oakes D. Analysis of survival data. London, United Kingdom: Chapman and Hall Ltd, Obesity: preventing and managing the global epidemic. Report of a WHO consultation. World Health Organ Tech Rep Ser 2000;894: National Center for Health Statistics CDC growth charts: United States. Hyattsville, MD: National Center for Health Statistics, (World Wide Web URL: growthcharts/). 19. Kuczmarski RJ, Ogden CL, Grummer-Strawn LM, et al. CDC growth charts: United States. Adv Data 2000;(314): SPSS, Inc. SPSS for Windows, release Chicago, IL: SPSS, Inc, Insightful Corporation. S-Plus 6.1 for Windows guide to statistics. Vol 2. Seattle, WA: Insightful Corporation, Vainio H, Bianchini F. Weight control and physical activity. (IARC Handbooks of Cancer Prevention, vol 6). Lyon, France: International Agency for Research on Cancer, Engeland A, Bjørge T, Selmer RM, et al. Height and body mass index in relation to total mortality. Epidemiology 2003;14: Mellemgaard A, Lindblad P, Schlehofer B, et al. International renal-cell cancer study. III. Role of weight, height, physical activity, and use of amphetamines. Int J Cancer 1995;60: Chow WH, Gridley G, Fraumeni JF Jr, et al. Obesity, hypertension, and the risk of kidney cancer in men. N Engl J Med 2000; 343: Pan SY, Johnson KC, Ugnat AM, et al. Association of obesity and cancer risk in Canada. Am J Epidemiol 2004;159: Tulinius H, Sigfússon N, Sigvaldason H, et al. Risk factors for malignant diseases: a cohort study on a population of 22,946 Icelanders. Cancer Epidemiol Biomarkers Prev 1997;6: Engeland A, Bjørge T, Tverdal A, et al. Obesity in adolescence and adulthood and the risk of adult mortality. Epidemiology 2004;15: Bergström A. Renal cell cancer: the role of physical activity and body size. (Doctoral thesis). Stockholm, Sweden: Karolinska Institutet, 2002.