Body Size and Renal Cell Cancer Incidence in a Large US Cohort Study

Abstract

Renal cell cancer (RCC) incidence has increased in the United States over the past three decades. The authors analyzed the association between body mass index (BMI) and invasive RCC in the National Institutes of Health (NIH)–AARP Diet and Health Study, a large, prospective cohort aged 50–71 years at baseline initiated in 1995–1996, with follow-up through December 2003. Detailed analyses were conducted in a subcohort responding to a second questionnaire, including BMI at younger ages (18, 35, and 50 years); weight change across three consecutive age intervals; waist, hip, and waist-to-hip ratio; and height at age 18 years. Incident RCC was diagnosed in 1,022 men and 344 women. RCC was positively and strongly related to BMI at study baseline. Among subjects analyzed in the subcohort, RCC associations were strongest for baseline BMI and BMI recalled at age 50 years and were successively attenuated for BMI recalled at ages 35 and 18 years. Weight gain in early (18–35 years of age) and mid- (35–50 years of age) adulthood was strongly associated with RCC, whereas weight gain after midlife (age 50 years to baseline) was unrelated. Waist-to hip ratio was positively associated with RCC in women and with height at age 18 years in both men and women.
Introduction

The incidence of kidney cancer, which combines renal parenchyma (renal cell) and renal pelvis cancers, is increasing in the United States, with all of the increase due to renal cell cancer.[1] High body mass has been consistently identified as a risk factor,[2] which is of concern because a large proportion of the US adult population is overweight and obese.[3]

Recent body mass is positively related to renal cell cancer in both men and women.[4] A limited number of studies have investigated whether the body mass relation differs according to the age period during which body mass is considered—that is, whether body mass at any stage of adulthood is particularly associated with renal cell cancer—with conflicting results.[5-15] Limited evidence suggests that weight gain is related to renal cell cancer risk,[11,12,15,16] but, again, whether timing of weight gain matters is not understood. Most analyses report no association between height and renal cell cancer,[7-9,12,15,17] although intriguing recent evidence suggests a relation between waist-to-hip ratio, a measure of abdominal adiposity, and renal cell cancer.[13,15-17]

We analyzed the relations of renal cell cancer incidence with body size, including body mass index (BMI) at younger ages, weight change, waist and hip sizes, and height at age 18 years in the National Institutes of Health (NIH)–AARP Diet and Health Study, a large cohort of men and women living in six states and two metropolitan areas across the United States. The availability of several measures of body size and the large number of incident cases of renal cell cancer that have occurred enabled us to consider these associations in detail.Study Population

Design and maintenance of the NIH–AARP Diet and Health Study cohort have been described previously.[18] Briefly, in 1995–1996, a self-administered baseline questionnaire was mailed to 3.5 million AARP members, 50–71 years of age, requesting information on demographic and anthropometric characteristics, dietary intake, and health-related behaviors. A total of 566,402 persons satisfactorily completed the baseline questionnaire. Exclusions were as follows: 719 with a diagnosis of renal cell cancer before study entry, 477 cases for whom only mortality information on kidney cancer was available (excluded because this study specifically evaluated renal cell cancer incidence), 15,760 whose questionnaires were completed by surrogates, and 13,400 for whom information on height or weight was missing. We further excluded 4,637 AARP members whose energy intake estimates were implausible and 2,637 reporting extreme values for weight or height. The baseline analytical cohort, which was used to analyze the relation of baseline BMI and height to renal cell cancer, included 528,772 participants (312,500 men and 214,906 women). In 1996–1997, a second (subcohort) questionnaire was mailed to persons responding to the first one, and responses were received from 334,908 persons. The second questionnaire collected more detailed information on body weight history and baseline waist and hip size. After exclusions similar to those for the baseline questionnaire, the remaining 320,618 participants (185,758 men and 134,860 women) constituted the analytical subcohort.

The study was approved by the Special Studies Institutional Review Board of the US National Cancer Institute. All study participants provided written informed consent.
Case Ascertainment

Incident renal cell cancer cases were identified through state cancer registries, linked by name, address, sex, date of birth, and, if available, Social Security number. Renal cell carcinoma was defined as International Classification of Disease for Oncology code C260 and incorporating histology codes (8010, 8032, 8140, 8211, 8246, 8260, 8310, 8312, 8320) consistent with renal cell carcinoma. Participants were followed until diagnosis of first renal cell cancer, date they moved out of the registry area, death, or date of last follow-up on December 31, 2003.
Assessment of Height, Weight, and Other Exposures

Self-reported height and weight were collected from the baseline questionnaire, and height at age 18 years; weight at ages 18, 35, and 50 years; and waist and hip circumferences were collected from the subcohort questionnaire. Persons whose body size measures were missing or extreme were excluded on a per-analysis basis to conserve sample size. Extreme values were defined as more than three interquartile ranges below the 25th percentile or more than three interquartile ranges above the 75th percentile. As an example, this procedure removed subjects with the lowest 0.07 percent and highest 0.08 percent baseline weights. Baseline BMI was calculated from baseline height and weight as weight (kilograms) divided by height (meters) squared. BMI at ages 18, 35, and 50 years was calculated from the subcohort questionnaire by using height at age 18 years and weight at ages 18, 35, and 50 years. BMI was divided into seven categories for analysis in the full cohort: <18.5, 18.5–<22.5 (referent), 22.5–<25, 25–<27.5, 27.5–<30, 30–<35, and ≥35 kg/m2; in the subcohort, the highest category was ≥30 kg/m2. These categories correspond to or are nested within the World Health Organization classifications for underweight (<18.5 kg/m2), normal weight (18.5–<25 kg/m2), overweight (25–<30 kg/m2), and obesity (≥30 kg/m2). Because participants tended to report much lower weights at ages 18 and 35 years, an alternative categorization was created dividing the normal BMI range into three finer categories: 18.5–<21 (referent), 21–<23, and 23–<25 kg/m2).

Weight change within 4 kg was defined as stable for analyses of 18–35 years, 35–50 years, and 50 years to baseline age intervals. A wider range of weight gain of –4 kg to 10 kg was considered stable for the longer, 18 years to baseline age interval to achieve a sufficiently large referent group. Height at age 18 years, waist and hip sizes, and waist-to-hip ratio were analyzed as quintiles.

Cigarette smoking was categorized as never smoked, formerly smoked (quit smoking 1–9 years ago or quit smoking ≥10 years ago), and currently smoked at baseline or quit for <1 year, and, among smokers, by dose (1–10, 11–20, 21–30, or >30 cigarettes/day), with a separate category assigned to missing values. A three-level physical activity index was created based on two questions addressing sports and non-sports-related activity. Percentage of energy consumed as protein was categorized to quartiles and other anthropometric variables (height, weight) to quintiles. Self-reported physician-diagnosed history of diabetes and hypertension were dichotomous.
Data Analysis

Age-adjusted and multivariate relative risks were estimated by using Cox regression analysis with age as the underlying time metric.[19] All statistical tests were two sided, with α = 0.05 considered statistically significant. Trends were evaluated as grouped linear variables by using BMI category medians. For weight change analyses, the trend reported is for weight stability or gain, excluding persons who lost ≥4 kg.

Multivariate models were adjusted for age, smoking, physical activity, percentage of energy consumed as protein, and history of diabetes. Analyses combining men and women were adjusted for gender. Analyses based on the subcohort were adjusted for these variables and, in addition, history of hypertension. Numerous potential confounders were tested but were found to not affect risk estimates appreciably: dietary variables of fat, carbohydrate, fruit and vegetable, and red meat intake (all adjusted for calories) and total calories, and, for women, reproductive variables including number of children, age at first livebirth, menopausal hormone replacement therapy, and multivitamin use. Weight change analyses were also adjusted for BMI at the beginning of the weight change interval (initial BMI) and for height at age 18 years.

Effect modification was evaluated by both tests of interaction and analysis of heterogeneity (stratification according to categories of the third variable, using a single referent group). Potential effect modifiers included age, smoking, protein intake, history of diabetes, history of hypertension, and BMI at a younger age. Statistical interactions involving naturally continuous variables (e.g., BMI, weight change) were modeled as continuous, linear cross-product terms. Statistical significance was evaluated by using the likelihood ratio test, comparing full and reduced models, with p = 0.05 considered statistically significant.