Modelling body mass index and endometrial cancer risk in a pooled‐analysis of three case–control studies

Objective To quantify the relation between body mass index ( BMI ) and endometrial cancer risk, and to describe the shape of such a relation. Design Pooled analysis of three hospital‐based case–control studies. Setting Italy and S witzerland. Population A total of 1449 women with endometrial cancer and 3811 controls. Methods Multivariate odds ratios ( OR ) and 95% confidence intervals (95% CI ) were obtained from logistic regression models. The shape of the relation was determined using a class of flexible regression models. Main outcome measure The relation of BMI with endometrial cancer. Results Compared with women with BMI 18.5 to <25 kg/m 2 , the odds ratio was 5.73 (95% CI 4.28–7.68) for women with a BMI ≥35 kg/m 2 . The odds ratios were 1.10 (95% CI 1.09–1.12) and 1.63 (95% CI 1.52–1.75) respectively for an increment of BMI of 1 and 5 units. The relation was stronger in never‐users of oral contraceptives ( OR 3.35, 95% CI 2.78–4.03, for BMI ≥30 versus <25 kg/m 2 ) than in users ( OR 1.22, 95% CI 0.56–2.67), and in women with diabetes ( OR 8.10, 95% CI 4.10–16.01, for BMI ≥30 versus <25 kg/m 2 ) than in those without diabetes ( OR 2.95, 95% CI 2.44–3.56). The relation was best fitted by a cubic model, although after the exclusion of the 5% upper and lower tails, it was best fitted by a linear model. Conclusions The results of this study confirm a role of elevated BMI in the aetiology of endometrial cancer and suggest that the risk in obese women increases in a cubic nonlinear fashion. The relation was stronger in never‐users of oral contraceptives and in women with diabetes. Tweetable Abstract Risk of endometrial cancer increases with elevated body weight in a cubic nonlinear fashion.