Renal Cell Cancer and Body Size at Different Ages: An Italian Multicenter Case-Control Study

Transcription

1 American Journal of Epidemiology ª The Author Published by the Johns Hopkins Bloomberg School of Public Health. All rights reserved. For permissions, please Vol. 166, No. 5 DOI: /aje/kwm108 Advance Access publication June 24, 2007 Original Contribution Renal Cell Cancer and Body Size at Different Ages: An Italian Multicenter Case-Control Study Luigino Dal Maso 1, Antonella Zucchetto 1, Alessandra Tavani 2, Maurizio Montella 3, Valerio Ramazzotti 4, Renato Talamini 1, Vincenzo Canzonieri 5, Antonio Garbeglio 6, Eva Negri 2, Annamaria Tonini 7, Carlo La Vecchia 2,8, and Silvia Franceschi 9 1 S.O.C. Epidemiologia e Biostatistica, Centro di Riferimento Oncologico, Aviano (PN), Italy. 2 Istituto di Ricerche Farmacologiche Mario Negri, Milano, Italy. 3 Servizio di Epidemiologia, Istituto Tumori Fondazione Pascale, Napoli, Italy. 4 Servizio Integrato di Epidemiologia e Sistemi Informativi, Istituto Nazionale Tumori Regina Elena, Roma, Italy. 5 S.O.C. Anatomia Patologica, Centro di Riferimento Oncologico, Aviano (PN), Italy. 6 Unità Operativa di Urologia, Azienda Ospedaliera Santa Maria degli Angeli, Pordenone, Italy. 7 Consiglio Nazionale delle Ricerche, Istituto di Fisiologia Clinica, Milano, Italy. 8 Istituto di Statistica Medica e Biometria, Università degli Studi di Milano, Milano, Italy. 9 International Agency for Research on Cancer, Lyon, France. Received for publication November 2, 2006; accepted for publication March 7, An increased risk of renal cell cancer (RCC) has been reported in overweight persons. The authors aimed to clarify which anthropometric measures are associated with risk of RCC and whether risk may vary according to selected variables. Between 1992 and 2004, they carried out an Italian multicenter case-control study including 767 (494 men, 273 women) incident of RCC and 1,534 hospital, frequency-matched to. To estimate odds ratios and 95% confidence intervals, they used conditional logistic regression matched on study center, sex, and age and adjusted for period of interview, years of education, smoking habits, and family history of kidney cancer. Using body-size measurements taken 1 year prior to diagnosis/interview, the authors found an odds ratio of 1.3 (95% confidence interval (CI): 1.0, 1.7) among obese persons (body mass index (BMI; weight (kg)/height (m) 2 ) 30) versus normal-weight persons (BMI <25) and an odds ratio of 1.5 (95% CI: 1.1, 2.0) among persons in the highest tertile of waist-to-hip ratio. Direct associations emerged for BMI 30 (vs. <25) at ages 30 years (odds ratio ¼ 1.5, 95% CI: 1.0, 2.3) and 50 years (odds ratio ¼ 1.5, 95% CI: 1.1, 2.0). The direct association with waist-to-hip ratio was stronger among women than among men. RCC risks among overweight and obese persons were apparently higher in never smokers, persons with the clear-cell histologic type, and persons with a Fuhrman nuclear grade of G3 G4. body mass index; body size; carcinoma, renal cell; case-control studies; waist-hip ratio Abbreviations: BMI, body mass index; CI, confidence interval; OR, odds ratio; RCC, renal cell carcinoma; WHR, waist-to-hip ratio. Renal cell carcinoma (RCC) accounts for approximately 2 percent of all cancers in adults in many Western countries, and its incidence has been rising over the last few decades (1). Although several potential risk factors have been recognized, including tobacco smoking (2, 3), hypertension (4), and family history of kidney cancer (5, 6), the etiology of RCC is still largely undefined. It has been suggested that obesity, generally measured as body mass index (BMI; weight (kg)/height (m) 2 ), is related to RCC (7), and obesity may be responsible, at Correspondence to Dr. Luigino Dal Maso, Epidemiology and Biostatistics Unit, Aviano Cancer Institute, Aviano Via Franco Gallini 2, Aviano (PN), Italy ( epidemiology@cro.it). 582

2 Body Mass Index and Renal Cell Cancer Risk 583 TABLE 1. Distribution of 767 of renal cell carcinoma and 1,534 frequency-matched according to sex, age, and selected covariates, Italy, Variable Cases Controls No. % No. % Sex Male Female Age (years) < Education (years) < Smoking habits* Never smoker Current smoker <20 cigarettes/day cigarettes/day Ex-smokery Family history of kidney cancerz No , Yes Histologic type Clear-cell Papillary Chromophobic Unclassified Unknown Tumor extension (TNM classification) T T T T Unknown Fuhrman nuclear grade G G G G Unknown * Numbers of do not sum to the total because of missing values. y Ex-smokers were subjects who had not smoked for at least 4 years. z In first-degree relatives. TNM, tumor-node-metastasis. TABLE 2. Distribution of 767 of renal cell cancer and 1,534 frequency-matched according to body-size measurements taken 1 year before diagnosis/interview and corresponding odds ratios, Italy, Variable * * Odds ratioy 95% confidence interval Height (tertiles)z , , 1.18 v 2 for trend (1 df) 0.41 p for trend 0.52 Weight (tertiles){ , , 1.46 v 2 for trend (1 df) 2.28 p for trend 0.13 Body mass index# < < , , 1.69 v 2 for trend (1 df) 1.97 p for trend 0.16 (per five-unit increase) , 1.33 Waist-to-hip ratio (tertiles)** , , 2.00 v 2 for trend (1 df) 7.97 p for trend <0.01 (per 0.05-unit increase) , 1.28 * Numbers may not sum to totals because of missing values. y Estimates from conditional logistic regression models, matched on study center, sex, and age and adjusted for calendar period of interview, years of education, smoking habits, and family history of kidney cancer (first-degree relatives). z Tertile cutpoints for height were 170 cm and 175 cm in males and 160 cm and 165 cm in females. Reference category. { Tertile cutpoints for weight were 74 kg and 82 kg in males and 62 kg and 71 kg in females. # Weight (kg)/height (m) 2. ** Tertile cutpoints for waist-to-hip ratio as measured at diagnosis/ interview were and in males and and in females. least in part, for the rising incidence rates (8). Anthropometric measures other than BMI, including waist-tohip ratio (WHR) and variation in BMI throughout the

3 584 Dal Maso et al. TABLE 3. Distribution of 767 of renal cell carcinoma and 1,534 frequency-matched according to lifetime body-size measurements and corresponding odds ratios, Italy, Variable * * Odds ratioy 95% confidence interval Perceived body size at age 12 years Thinner than peers z Same as peers , 1.40 Heavier than peers , 1.37 v 2 for trend (1 df) 0.70 p for trend 0.40 BMI,{ at age 30 years# < ,003 1z 25 < , , 2.25 v 2 for trend (1 df) 4.41 p for trend 0.04 (per five-unit increase) , 1.47 BMI at age 50 years** < z 25 < , , 2.03 v 2 for trend (1 df) 5.81 p for trend 0.02 (per five-unit increase) , 1.34 Highest BMI during lifetime < z 25 < , , 1.64 v 2 for trend (1 df) 3.38 p for trend 0.07 (per five-unit increase) , 1.32 Table continues life course, have been less frequently considered (9 12). Only scanty information is available on the variation of RCC risk in obese patients according to tumor histologic type, stage or grade, and other potential risk factors. To obtain further information on the link between RCC and lifetime anthropometric measures and whether such a potential relation may vary according to smoking habits, histologic type, and clinical presentation, we carried out a large case-control study in Italy. TABLE 3. Continued Variable * MATERIALS AND METHODS of * Odds ratioy 95% confidence interval Lowest BMI during lifetime < z 20 < , , 2.46 v 2 for trend (1 df) 9.37 p for trend <0.01 (per five-unit increase) , 1.62 Increase in BMI from age 30 years# < z 2 < , , 1.12 v 2 for trend (1 df) 0.86 p for trend 0.35 Increase in BMI from lowest BMI < z 3 < , , 1.21 v 2 for trend (1 df) 0.09 p for trend 0.77 * Numbers may not sum to totals because of missing values. y Estimates from conditional logistic regression models, matched on study center, sex, and age and adjusted for calendar period of interview, years of education, smoking habits, and family history of kidney cancer (first-degree relatives). z Reference category. BMI, body mass index. { Weight (kg)/height (m) 2. # Subjects aged less than 30 years were excluded. ** Subjects aged less than 50 years were excluded. Between 1992 and 2004, we conducted a hospital-based case-control study on RCC in four areas of Italy: the provinces of Pordenone and Gorizia (northeastern Italy), the greater Milan area (northwestern Italy), the province of Latina (central Italy), and the urban area of Naples (southern Italy). Cases were 767 subjects (494 men and 273 women; median age, 62 years (range, years)) admitted to major teaching or general hospitals in one of the above areas with incident cancer of the renal parenchyma (International Classification of Diseases, Ninth Revision, code 189.0). Histologic information included cancer cell type, according to the 2004 World Health Organization classification (13, 14), tumor extension, according to the tumor-node-metastasis classification (15), and Fuhrman nuclear grade (16). Patients reporting transplantation or chronic renal failure were excluded.

4 Body Mass Index and Renal Cell Cancer Risk 585 A) B) Odds ratio Odds ratio BMI 1 year before diagnosis/interview WHR C) Odds ratio BMI at age 30 years Control subjects were 1,534 patients (988 men and 546 women; median age, 62 years (range, years)) admitted to the same network of hospitals as. Causes of admission included acute nonneoplastic conditions not associated with hormone-related disorders and urinary or genital tract diseases (32 percent had nontraumatic orthopedic disorders, 26 percent had traumatic injuries, 14 percent had surgical conditions, and 28 percent had miscellaneous other illnesses, such as eye, ear, nose, throat, dental, or skin disorders). Controls were frequency-matched to (2:1) according to study center, sex, and quinquennium of age. Centrally trained staff interviewed the eligible subjects during their hospital stay. Fewer than 5 percent of the approached and refused the interview. The response rates were similar across hospitals and geographic areas. All interviews were conducted using a structured questionnaire that collected information on age, education and other socioeconomic factors, physical activity, smoking habits, alcohol intake, diet, and history of cancer in first-degree relatives; the questionnaire also included a problemoriented medical history. In a detailed section of the questionnaire, study subjects were asked to report their height, D) Odds ratio BMI at age 50 years FIGURE 1. Odds ratios for renal cell carcinoma according to body mass index (BMI; weight (kg)/height (m) 2 ) at different times and waist-to-hip ratio (WHR), as estimated by regression spline models, Italy, Curves are shown for the spline models that were the best-fitting according to Akaike s Information Criterion. The regression equations included terms for study center, sex, age, calendar period of interview, years of education, smoking habits, and family history of kidney cancer (first-degree relatives). The filled circle in part C shows the location of a knot. Dashed lines represent 95% confidence intervals. A) BMI 1 year before diagnosis/interview (2.4% of subjects were out of displayed range; p for linear trend ¼ 0.18); B ) WHR (4.4% of subjects were out of displayed range; p for linear trend < 0.01); C ) BMI at age 30 years (4.4% of subjects were out of displayed range; p for linear trend < 0.01); D ) BMI at age 50 years (2.1% of subjects were out of displayed range; p for linear trend < 0.01). weight 1 year before cancer diagnosis or interview (in ), weight at different ages, lifelong highest and lowest weights, and perceived body size at age 12 years (i.e., thinner than, the same as, or heavier than one s peers). BMI was computed as weight (kg) divided by height squared (m 2 ). The interviewer measured the circumference of the participant s waist (2 cm above the umbilicus) and hips (maximal protrusion); hence, waist-to-hip ratio (WHR) was also computed. In 38 percent of RCC patients and 34 percent of control subjects, the waist or hips could not be measured for technical reasons. However, median BMI in subjects with a measured WHR was not significantly different from that in subjects with missing WHR 26.2 and 25.9, respectively. Corresponding values were 26.4 and 26.2 among and 26.1 and 25.8 among. Standard BMI categories (<25, 25 <30, and 30) were used to allow comparison with most previous studies. Conditional logistic regression was applied to calculate odds ratios and the corresponding 95 percent confidence intervals. In all analyses, were matched to on study center, sex, and age, and results were adjusted for calendar period of interview, years of education, smoking habits, and family history of kidney cancer in first-degree relatives (17).

5 586 Dal Maso et al. TABLE 4. Odds ratio for renal cell carcinoma according to tertile of waist-to-hip ratio in strata of selected variables, Italy, * Variable Additional adjustment for energy or alcohol intake, physical activity, or history of treated hypertension did not materially modify the risk estimates. Selected analyses were repeated separately according to strata of sex, age at diagnosis/interview, smoking habits, histologic type, tumor extension, and Furman nuclear grade. Data on histologic type, extension, and grade were not systematically collected at all participating centers; consequently, the proportion of missing values was high. Stratified analyses were conducted on BMI at age 30 years to avoid exclusion of subjects (16 percent) below age 50 years. Use of BMI at age 30 years permitted evaluation of the impact of long-duration overweight on RCC risk. To avoid potentially arbitrary categorizations, we also investigated the dose-response relation between BMI and risk of RCC using logistic regression splines (18, 19) with the appropriate calculation of pointwise confidence intervals. We selected the optimal number of segments by minimizing Akaike s Information Criterion (20). Tertile of waist-to-hip ratioy 1z 2 3 OR,{ 95% CI RESULTS OR{ 95% CI p value (test for heterogeneity across strata) Sex 0.04 Male , , 1.68 Female , , 4.65 Age (years) at diagnosis/ interview 0.52 < , , , , 1.97 Smoking habits 0.74 Never smoker , , 3.04 Ever smoker , , 1.98 Histologic type 0.33 Clear-cell , , 1.99 Other specified subtype , , 2.85 Unknown , , 2.59 Tumor extension (TNM classification) 0.07 T , , 1.78 T2 T , , 4.22 Unknown , , 1.70 Fuhrman nuclear grade 0.83 G1 G , , 2.23 G3 G , , 2.21 Unknown , , 2.53 * The analysis included 767 of renal cell carcinoma and 1,534 frequency-matched. Numbers of and may not sum to these totals because of missing values. y Tertile cutpoints for waist-to-hip ratio were and in males and and in females. z Reference category (odds ratio ¼ 1). OR, odds ratio; CI, confidence interval; TNM, tumor-node-metastasis. { Estimates from conditional logistic regression models, matched on study center, sex, and age and adjusted for calendar period of interview, years of education, smoking habits, and family history of kidney cancer (first-degree relatives). Table 1 shows the distribution of and according to sex, age, and other selected covariates, including histologic type, extension, and Fuhrman nuclear grade. Compared with, had a higher level of education, were more frequently heavy current smokers, and more frequently reported a family history of kidney cancer in firstdegree relatives. Median BMIs 1 year before diagnosis/interview were 26.4 (range, ) for and 26.1 (range, ) for ; corresponding figures for WHR were 0.95 (range, ) for and 0.94 (range, ) for. Table 2 shows the distribution of RCC and and the corresponding odds ratios, according to various body size measures 1 year before diagnosis/interview. No association emerged between height and RCC. An increased risk of RCC emerged for persons in the highest tertile of

6 Body Mass Index and Renal Cell Cancer Risk 587 TABLE 5. Odds ratio for renal cell carcinoma according to body mass index at age 30 years in strata of selected variables, Italy, * body weight (odds ratio (OR) ¼ 1.18, 95 percent confidence interval (CI): 0.95, 1.46) as compared with the lowest and for obese (BMI 30) persons (OR ¼ 1.29, 95 percent CI: 0.99, 1.69) as compared with persons with a BMI <25. The association between BMI 1 year before diagnosis/interview and RCC risk was approximately linear, with an odds ratio of 1.19 (95 percent CI: 1.06, 1.33) for a five-unit increase in BMI (table 2). WHR was also directly associated with RCC risk (for the highest tertile vs. the lowest, OR ¼ 1.50, 95 percent CI: 1.12, 2.00) (table 2). A consistent and significant association emerged between BMI during adulthood and RCC (table 3). The odds ratios were 1.46 (95 percent CI: 0.95, 2.25) among persons who were obese (BMI 30) at age 30 years and 1.48 (95 percent CI: 1.07, 2.03) among persons who were obese at age 50 years, as compared with a BMI <25. The odds ratio was Body mass indexy at age 30 yearsz <25 25 <30 30 OR{,# 95% CI{ OR# 95% CI p value (test for heterogeneity across strata) Sex 0.91 Male , , 2.42 Female , , 3.18 Age (years) at diagnosis/ interview 0.95 < , , , , 2.47 Smoking habits 0.60 Never smoker , , 3.51 Ever smoker , , 2.17 Histologic type 0.08 Clear-cell 259 1, , , 3.11 Other specified subtype 105 1, , , 1.74 Unknown 128 1, , , 2.99 Tumor extension (TNM{ classification) 0.86 T1 97 1, , , 2.93 T2 T , , , 2.76 Unknown 233 1, , , 2.67 Fuhrman nuclear grade 0.35 G1 G , , , 2.03 G3 G , , , 3.68 Unknown 241 1, , , 2.57 * The analysis included 767 of renal cell carcinoma and 1,534 frequency-matched. Numbers of and may not sum to these totals because of missing values. y Weight (kg)/height (m) 2. z Subjects aged less than 30 years were excluded. Reference category (odds ratio ¼ 1). { OR, odds ratio; CI, confidence interval; TNM, tumor-node-metastasis. # Estimates from conditional logistic regression models, matched on study center, sex, and age and adjusted for calendar period of interview, years of education, smoking habits, and family history of kidney cancer (first-degree relatives) (95 percent CI: 1.24, 2.46) when the lowest BMI during adult life was 25, as compared with <20, whereas an increase in BMI from the lowest level or from age 30 years was not associated with RCC risk (table 3). The pattern of risk for BMI at different ages and WHR was not modified using other categorizations (i.e., quartiles or quintiles; data not shown in tables). The shape of the best-fitting regression splines (doseresponse curve) for selected body mass measures is shown in figure 1. The risk of RCC increased monotonically for BMI at all ages considered and for WHR. These direct relations were approximately linear, and no threshold effect emerged. When the association between WHR (highest tertile vs. lowest) and RCC risk was examined in strata of selected variables, higher odds ratios emerged in women (OR ¼ 2.66, 95 percent CI: 1.52, 4.65), in never smokers (OR ¼

7 588 Dal Maso et al. TABLE 6. Odds ratio for renal cell carcinoma according to tertile of waist-to-hip ratio and body mass index at age 30 years, Italy, * Body mass indexy at age 30 yearsz 1.83, 95 percent CI: 1.10, 3.04), and in subjects with a tumor extension of T2 T4 (OR ¼ 2.62, 95 percent CI: 1.63, 4.22), while no difference emerged by histologic type (table 4). The direct association between BMI (30 vs. <25) at age 30 years and RCC risk was apparently stronger in women (OR ¼ 1.57, 95 percent CI: 0.78, 3.18) and, more notably, in never smokers (OR ¼ 1.82, 95 percent CI: 0.94, 3.51) (table 5). Additional adjustment for reproductive factors in women did not modify risk estimates. Moreover, an association between BMI and RCC emerged only in subjects with the clear-cell subtype of RCC (OR ¼ 1.84, 95 percent CI: 1.09, 3.11) and the highest Fuhrman grades (OR ¼ 1.82, 95 percent CI: 0.90, 3.68), whereas no difference emerged by tumor extension (table 5). The combined effect of WHR and BMI at age 30 years on RCC risk is shown in table 6. An apparently independent (additive) effect emerged for the two factors, with an odds ratio of 1.69 (95 percent CI: 1.14, 2.52) for overweight persons in the highest tertile of WHR. In overweight women in the highest tertile of WHR, the odds ratio was 2.83 (95 percent CI: 1.31, 6.14). Mutual adjustment for BMI and WHR did not substantially change any of the above results. After adjustment for potential confounders, the fraction of RCC attributable (21) to overweight in young adulthood (BMI 25) in this Italian population was 6 percent (95 percent CI: 0, 12). DISCUSSION Tertile of waist-to-hip ratio OR{,# 95% CI{ OR# 95% CI Both sexes < ** , , , , , 2.52 Men < ** , , , , , 2.24 Women < ** , , , , , 6.14 * The analysis included 767 of renal cell carcinoma and 1,534 frequency-matched. Numbers of and may not sum to these totals because of missing values. y Weight (kg)/height (m) 2. z Subjects aged less than 30 years were excluded. Tertile cutpoints for waist-to-hip ratio were and in men and and in women. { OR, odds ratio; CI, confidence interval. # Estimates from conditional logistic regression models, matched on center, sex (as appropriate), and age and adjusted for calendar period of interview, years of education, smoking habits, and family history of kidney cancer (first-degree relatives). ** Reference category. OR# 95% CI In this study, we observed direct associations between several measures of body weight and RCC risk. The association with BMI was approximately linear, with no evidence of a threshold effect. Our findings on BMI are consistent with the overall epidemiologic evidence from case-control and prospective studies (10, 12, 22 29). Some investigators have reported that the association between BMI and RCC incidence or mortality was stronger in women (26, 30). However, in our study, this association did not vary substantially between men and women, a finding that is in agreement with larger cohort studies (24, 25), casecontrol studies (11, 31), and a recent pooled analysis (22). The relation of WHR with RCC has less frequently been explored (26, 27, 32), but consistent, direct associations that are stronger in women have been reported. WHR differed substantially in the two sexes, and this difference should largely explain the weaker association that emerged between WHR and RCC in men. The association between obesity and RCC risk may be mediated by several hormonal mechanisms (22). Overweight and obesity are accompanied by elevated levels of fasting serum and free insulin-like growth factor I (33), which contributes to cell growth and proliferation (34). Obesity also increases levels of free endogenous estrogens (35), which may, in turn, increase renal cell proliferation. Moreover, since obese persons have been reported to have a higher glomerular filtration rate and renal plasma flow, independently of hypertension, these conditions may increase the risk of kidney damage (36, 37) and therefore make the kidney more susceptible to carcinogens. In addition, overweight is correlated with hypertension and could share a causal pathway with hypertension, a consistently reported risk factor for RCC (4, 38), and diabetes mellitus, a more controversial risk factor for RCC (38 41). Physical

8 Body Mass Index and Renal Cell Cancer Risk 589 activity is another factor potentially related to overweight and RCC. However, in this same population, we previously found a moderate inverse association between occupational physical activity and risk of RCC, while we found no association for leisure-time physical activity, which has a low prevalence and is engaged in at only moderate levels in Italy (42). In our study, the relation between BMI and RCC was stronger among never smokers than among ever smokers. This finding is consistent with a recent prospective study (24), while another study showed a less clear modification effect (25). A possible explanation is the direct association between RCC and smoking, which is negatively related to BMI (43). However, in our control patients, the association between smoking habits and BMI was not significant. Our study suggests, for the first time, that the direct association between BMI and RCC risk is stronger for persons with the clear-cell histologic subtype. This can be related to recent findings showing that aberrant influx and storage of glucose are associated with an increased RCC risk (44). The relative excess of intracellular glycogen gives the cells their clear appearance and reflects impairments of glucose metabolism, including the up-regulation of the GLUT1 gene in the neoplastic cells (44). The direct association between measures of body size appeared stronger in subjects with high (G3 G4) Fuhrman nuclear grades, particularly for patients with G4-grade RCC, whose odds ratios were 3.2 (95 percent CI: 1.1, 9.4) for BMI 30 at age 30 years versus BMI <25 and 1.9 (95 percent CI: 0.7, 5.6) for the highest tertile of WHR versus the lowest. G4 tumors are the result of de-differentiation of more differentiated tumors. It may be hypothesized that overweight persons have metabolic alterations that facilitate the neoplastic de-differentiation. The stronger association between RCC and WHR or, to a lesser extent, BMI in subjects with an advanced tumor extension (T2 T4) at diagnosis reflects possible diagnosis difficulties in obese persons. The fraction of RCC attributable to overweight/ obesity in this population (6 percent, 95 percent CI: 0, 12) was lower than the corresponding figure (21 percent, 95 percent CI: 11, 31) reported in a previous case-control study conducted in Canada (21, 25), based on 1,345 and an odds ratio of 2.7. However, the lower proportion of overweight/obesity in Italy (45) as compared with Canada also explains the difference. Problems of reliability in anthropometric measures cannot be excluded. However, there was no evidence that weight was differentially reported by and (46). Although it is known that study subjects tend to underestimate weight and overestimate height (47 49), there was no reason for differential recall bias in subjects reporting current weight and height. In our study, and were interviewed in similar hospital settings, and the general population was unaware of the possible link between anthropometric measures and cancer. In addition, Klipstein-Grobusch et al. (50) have shown that past body measures are generally well correlated with corresponding measures, even in older people. Hospital-based case-control studies may be more susceptible to selection and information bias than cohort studies (17). Subjects with diseases potentially linked to diet and dietary modifications were excluded from the control group, and major confounding factors for RCC were considered in the multivariate analysis. In particular, careful allowance was made for education and smoking habits, and additional allowance for potentially confounding factors did not alter the risk estimates. Possible sources of selection bias should also have been limited, since and were drawn from the same catchment areas and participation was almost complete. In conclusion, this Italian case-control study provides further evidence of a direct association between different measures of body weight and risk of RCC. Consistency of the results at different ages supports the hypothesis of a causal role of overweight in renal carcinogenesis. Moreover, excess weight (BMI) at various ages predicts RCC risk in both sexes, while in women, abdominal obesity, as indicated by a high WHR, represents a stronger risk indicator. Further studies should confirm the differential effects of overweight by smoking habits, histologic type, and tumor grade. ACKNOWLEDGMENTS This work was supported by the Italian Association for Cancer Research and the Italian League Against Cancer, Milan, Italy. The authors thank L. Mei for editorial assistance, O. Volpato and Dr. M. Cozzi for study coordination, and G. Bessega, L. Zaina, Dr. M. Grimaldi, and Dr. O. Manganelli for their help in data collection. The authors are deeply thankful to Dr. D. Maruzzi for his support in identifying cancer and to Drs. P. Ascierto, G. Chiara, R. Di Lauro, L. Forner, A. Grandi, R. Magri, A. Mele, G. Tosolini, and E. Trevisanutto for providing hospital control patients. Conflict of interest: none declared. REFERENCES 1. McLaughlin JK, Lipworth L. Epidemiologic aspects of renal cell cancer. Semin Oncol 2000;27: IARC Working Group on the Evaluation of Carcinogenic Risks to Humans. Tobacco smoke and involuntary smoking. IARC Mongr Eval Carcinog Risks Hum 2004;83: Hunt JD, van der Hel OL, Garnett P, et al. Renal cell carcinoma in relation to cigarette smoking: meta-analysis of 24 studies. Int J Cancer 2005;114: Grossman E, Messerli FH, Boyko V, et al. Is there an association between hypertension and cancer mortality? Am J Med 2002;112: Tavani A, La Vecchia C. Epidemiology of renal-cell carcinoma. J Nephrol 1997;10: Gago-Dominguez M, Yuan JM, Castelaro JE, et al. Family history and risk of renal cell carcinoma. Cancer Epidemiol Biomarkers Prev 2001;10:

9 590 Dal Maso et al. 7. Vainio H, Bianchini F, eds. Weight control and physical activity. (IARC Handbooks of Cancer Prevention, vol 6). Lyon: IARC Press, Godley PA, Ataga KI. Renal cell carcinoma. Curr Opin Oncol 2000;12: Lindblad P, Wolk A, Bergström R, et al. The role of obesity and weight fluctuations in the etiology of renal cell cancer: a population-based case-control study. Cancer Epidemiol Biomarkers Prev 1994;3: van Dijk BA, Schouten LJ, Kiemeney LA, et al. Relation of height, body mass, energy intake, and physical activity to risk of renal cell carcinoma: results from the Netherlands Cohort Study. Am J Epidemiol 2004;160: 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: Samanic C, Zhow W-H, Gridley G, et al. Relation of body mass index to cancer risk in 362,552 Swedish men. Cancer Causes Control 2006;17: Eble JN, Sauter G, Epstein JI, et al, eds. World Health Organization Classification of Tumours. Pathology and genetics of tumours of the urinary system and male genital organs. Lyon, France: IARC Press, Lopez-Beltran A, Scarpelli M, Montironi R, et al WHO classification of the renal tumors of the adults. Eur Urol 2006; 49: Sobin LH, Wittekind CH, eds. TNM classification of malignant tumours. 5th ed. New York, NY: John Wiley and Sons, Inc, Fuhrman SA, Lasky LC, Limas C. Prognostic significance of morphologic parameters in renal cell carcinoma. Am J Surg Pathol 1982;6: Breslow NE, Day NE. Statistical methods in cancer research. Vol 1. The analysis of case-control studies. (IARC Scientific Publication no. 32). Lyon, France: International Agency for Research on Cancer, 1980: Rosenberg PS, Katki H, Swanson CA, et al. Quantifying epidemiologic risk factors using non-parametric regression: model selection remains the greatest challenge. Stat Med 2003;22: Greenland S. Dose-response and trend analysis in epidemiology: alternatives to categorical analysis. Epidemiology 1995; 6: Polesel J, Dal Maso L, Bagnardi V, et al. Estimating doseresponse relationship between ethanol and risk of cancer using regression spline models. Int J Cancer 2005;114: Benichou J, Chow WH, McLaughlin JK, et al. Population attributable risk of renal cell cancer in Minnesota. Am J Epidemiol 1998;148: Bergstrom A, Hsieh CC, Lindblad P, et al. Obesity and renal cell cancer a quantitative review. Br J Cancer 2001;85: Chiu BC-H, Gapstur SM, Chow W-H, et al. Body mass index, physical activity, and risk of renal cell carcinoma. Int J Obes 2006;30: Bjørge T, Tretli S, Engeland A. Relation of height and body mass index to renal cell carcinoma in two million Norwegian men and women. Am J Epidemiol 2004;160: Pan SY, Johnson KC, Ugnat A-M, et al. Association of obesity and cancer risk in Canada. Am J Epidemiol 2004;159: Pischon T, Lahmann PH, Boeing H, et al. Body size and risk of renal cell carcinoma in the European Prospective Investigation into Cancer and Nutrition (EPIC). Int J Cancer 2006;118: Nicodemus KK, Sweeney C, Folsom AR. Evaluation of dietary, medical and lifestyle risk factors for incident kidney cancer in postmenopausal women. Int J Cancer 2004;108: Lukanova A, Björ O, Kaaks R, et al. Body mass index and cancer: results from the Northern Sweden Health and Disease Cohort. Int J Cancer 2006;118: Lindgren AM, Nissinen AM, Tuomilehto JO, et al. Cancer pattern among hypertensive patients in North Karelia, Finland. J Hum Hypertens 2005;19: Calle E, Rodriguez C, Walker-Thurmond K, et al. Overweight, obesity and mortality from cancer in a prospectively studied cohort of US adults. N Engl J Med 2003;348: Yuan JM, Castelao JE, Gago-Dominguez M, et al. Hypertension, obesity and their medications in relation to renal cell carcinoma. Br J Cancer 1998;77: Prineas RJ, Folsom AR, Zhang ZM, et al. Nutrition and other risk factors for renal cell carcinoma in postmenopausal women. Epidemiology 1997;8: Frystyk J, Vestbo E, Skjaebaek C, et al. Free insulin-like growth factors in human obesity. Metabolism 1995;44: Kellerer M, von Eye Corleta H, Muhlhofer A, et al. Insulinand insulin-like growth-factor-i receptor tyrosine-kinase activities in human renal carcinoma. Int J Cancer 1995;62: Zumoff B. Hormonal abnormalities in obesity. Acta Med Scand 1998;723(suppl): Hall JE. Louis K. Dahl Memorial Lecture. Renal and cardiovascular mechanisms of hypertension in obesity. Hypertension 1994;23: Ribenstein J, du Cailar G, Mimran A. Combined effects of overweight and hypertension. Hypertension 1995;26: Zucchetto A, Dal Maso L, Tavani A, et al. History of treated hypertension and diabetes mellitus and risk of renal cell cancer. Ann Oncol 2007;18: La Vecchia C, Negri E, Franceschi S, et. al. A case-control study of diabetes mellitus and cancer risk. Br J Cancer 1994; 70: Wideroff L, Gridley G, Mellemkjaer L, et al. Cancer incidence in a population-based cohort of patients hospitalized with diabetes mellitus in Denmark. J Natl Cancer Inst 1997;89: Rousseau MC, Parent ME, Pollak MN, et al. Diabetes mellitus and cancer risk in a population-based case-control study among men from Montreal, Canada. Int J Cancer 2005;118: Tavani A, Zucchetto A, Dal Maso L, et al. Lifetime physical activity and the risk of renal cell cancer. Int J Cancer 2007; 120: Williamson DF, Madans J, Anda RF, et al. Smoking cessation and severity of weight gain in a national cohort. N Engl J Med 1991;324: Page T, Hodgkinson AD, Ollerenshaw M, et al. Glucose transporter polymorphisms are associated with clear-cell renal carcinoma. Cancer Genet Cytogenet 2005;163: Gallus S, Colombo P, Scarpino V, et al. Overweight and obesity in Italian adults 2004, and an overview of trends since Eur J Clin Nutr 2006;10: Casey VA, Dwyer JT, Berkey CS, et al. Long-term memory of body weight and past weight satisfaction: a longitudinal follow-up study. Am J Clin Nutr 1991;53:

10 Body Mass Index and Renal Cell Cancer Risk Niedhammer I, Bugel I, Bonenfant S, et al. Validity of selfreported weight and height in the French GAZEL cohort. Int J Obes Relat Metab Disord 2000;24: Spencer EA, Appleby PN, Cavey GK, et al. Validity of selfreported height and weight in 4808 EPIC Oxford participants. Public Health Nutr 2002;5: Avila-Funes JA, Gutierrez-Robledo LM, Ponce De Leon Rosales S. Validity of height and weight self-report in Mexican adults: results from the National Health and Aging Study. J Nutr Health Aging 2004;8: Klipstein-Grobusch K, Kroke K, Boeing H. Reproducibility of self-reported past body weight. Eur J Clin Nutr 1998;7:83 94.