Body weight and risk of soft-tissue sarcoma

The relation between body mass (BMI) and soft-tissue sarcoma (STS) risk was evaluated in a case–control study from Northern Italy based on 217 incident STS and 1297 hospital controls. The risk of STS rose with BMI, with multivariate odds ratios of 3.49 (95% confidence interval (CI) 1.06–11.55) among men and 3.26 (95% CI 1.27–8.35) among women with a BMI >30 kg m–2 compared to those with BMI ≤ 20 kg m–2. © 1999 Cancer Research Campaign


Data analysis
Body mass index (BMI) was computed according to Quetelet's index (weight height -2 , kg m -2 ), which is essentially an indicator of weight unrelated to height (Benn, 1971). Odds ratios (OR) with their corresponding 95% confidence intervals (CI) were derived using unconditional multiple logistic regression analysis, fitted by the method of maximum likelihood (Breslow and Day, 1980). The regression models included terms for study centre, year of recruitment, age (8 levels plus a continuous term) and education. Table 1 gives the distribution of STS cases and controls by sex according to age and BMI, and the OR of STS according to BMI. The risk of STS rose with BMI, and the OR was 3.28 in subjects with a BMI > 30 compared to those with BMI ≤ 20. The OR was 3.49 for men and 3.26 for women. The risk estimates did not change after further adjustment for employment in agriculture and radiation therapy: the OR became 3.41 and 3.53 respectively for men and women with BMI > 30 compared to those with BMI ≤ 20.

RESULTS
A separate analysis for broad categories of anatomical sites of STS showed no differences in the strength of association by site: the OR for BMI > 30 compared to ≤ 20 for men and women combined was 3.02 (95% CI 1.21-7.52) for STS in the upper or lower limbs (ICD 171.2 and 171.3), 3.75 (95% CI 0.94-14.96) for STS in the abdomen or pelvis (ICD 171.5 and 171.6) and 2.88 (95% CI 0.66-12.58) for those arising in other or unspecified sites.
Height was not associated with risk of STS; the OR for the highest quartile compared to the lowest one was 1.14 (95% CI 0.59-2.22) in men and 0.50 (95% CI 0.25-0.99) in women.

DISCUSSION
The results of this study suggest a direct association between risk of STS and BMI in men and women and no relation with height, in agreement with the findings of Zahm et al (1989). Overweight and obesity are major public health issues and are associated with several chronic diseases and shorter life expectancy (Manson et al, 1987), probably through multiple biological mechanisms. Obese subjects were at elevated risk of leukaemias and lymphomas in the American Cancer Society cohort study conducted between 1959 and 1972 (Lew and Garfinkel, 1979). It is biologically plausible that overweight and obese people may be at higher risk of STS as, besides radiation, selected chemicals and some drugs and infectious agents (Zahm et al, 1996;Zahm and Fraumeni, 1997), herbicides and pesticides, particularly those containing 2,3,7,8-tetrachlorodibenzo-p-dioxin, chlorophenols and dibenzofurans, have been associated to STS risk (Hoar et al, 1986;Kang et al, 1987;Woods et al, 1987;Reif et al, 1989;Saracci et al, 1991;Bertazzi et al, 1993;Kogevinas et al, 1997;Hoppin et al, 1998), and most of these chemicals accumulate in human adipose tissue (Patterson et al, 1996;Kang et al, 1991;Orban et al, 1994). Very lean subjects may be selectively at low risk because of inherent characteristics of their adipose tissue, and of its specific role in STS pathogenesis. Moreover, immunological and hormonal imbalances in obese subjects have been reported, including changes in leptin (Friedman, 1997) and insulin growth factors, that regulate metabolism and play an important role in proliferation and apoptosis of several cell types (Daughaday, 1990;Tritos and Mantzoros, 1998), and that have been related to the risk of several neoplasms (Mantzoros et al, 1997;Petridou et al, 1999).

Body weight and risk of soft-tissue sarcoma
A possible source of bias in this study is the misclassification of body mass (Millar, 1986;Nieto-Garcia et al, 1990), as selfreported measures tend to overestimate height and underestimate weight (Palta et al, 1982;Stewart, 1982;Millar, 1986;Nieto-Garcia et al, 1990). However, such information bias is likely to be similar for cases and controls, and cannot explain the strong association observed. Hospital controls may differ from the general population in several respects (Breslow and Day, 1980). However, overweight and obesity are related to an increased prevalence of several chronic diseases (Negri et al, 1988;Pagano et al, 1997), so the use of hospital controls should, if anything, underestimate the true association (Breslow and Day, 1980). A major difficulty in this and other epidemiological studies on STS is the relatively low incidence and wide heterogeneity of this neoplasm, as aetiological differences by subsite and cell type have been suggested (Zahm et al, 1996;Zahm and Fraumeni, 1997).
The prevalence of overweight and obesity has been rising over the last few years in Italy (Pagano et al, 1997) and other Western countries (Galuska et al, 1996). This might partly explain the increased trends in incidence and mortality from STS (La Vecchia et al, 1992;Ross et al, 1993).
Thus, these data suggest that BMI is directly related to STS risk, and offer another reason for public health interventions to reduce overweight and obesity. Estimates from multiple logistic regression equations, including terms for study centre, year of recruitment, age and education. The 'All' category was further allowed for sex. b The sum does not add up to the total because of some missing values. c Reference category.