• body mass index;
  • high grade prostate cancer;
  • predictive accuracy


Many investigators suggested that obesity predisposes to adverse prostate cancer characteristics and outcomes. We tested the effect of obesity on the rate of aggressive prostate cancer at either prostate biopsy or radical prostatectomy (RP). Clinical and pathological data were available for 1,814 men. Univariable and multivariable logistic regression models addressed the rate of high grade prostate cancer (HGPCa) at either biopsy or final pathology. Clinical stage, prostate-specific antigen (PSA), percentage of free PSA and prostate volume were the base predictors. All models were fitted with and without body mass index (BMI), which quantified obesity. BMI and its reciprocal (InvBMI) were coded as cubic splines to allow nonlinear effects. Predictive accuracy (PA) was quantified with area under curve estimates, which were subjected to 200 bootstrap resamples to reduce overfit bias. Gains in PA related to the inclusion of BMI were compared using the Mantel–Haenszel test. HGPCa at biopsy was detected in 562 (31%) and HGPCa at RP pathology was present in 931 (51.3%) men. In either univariable or multivariable models predicting HGPCa at biopsy, BMI or InvBMI failed to respectively reach statistical significance or add to multivariable PA (BMI gain = 0%, p = 1.0; InvBMI gain = −0.2%, p = 0.9). Conversely, in models predicting HGPCa at RP, BMI and InvBMI represented independent predictors but failed to increase PA (BMI gain = 0.7%, p = 0.6; InvBMI gain = 0.5, p = 0.7%). Obesity does not predispose to moreaggressive prostate cancer at biopsy. Similarly, obesity doesnot change the ability to identify those who may harbor HGPCa at RP. © 2007 Wiley-Liss, Inc.