Abstract
BACKGROUND: Obesity and diabetes are associated inversely with low-grade prostate cancer risk and affect steroid hormone synthesis but whether they modify each other's impact on prostate cancer risk remains unknown. METHODS: We examined the independent associations of diabetes, body mass index (BMI), 'a body shape index' (ABSI), hip index (HI), circulating testosterone, sex hormone binding globulin (SHBG) (per one standard deviation increase) and oestradiol ≥175 pmol/L with total prostate cancer risk using multivariable Cox proportional hazards models for UK Biobank men. We evaluated multiplicative interactions (p(MI) ) and additive interactions (relative excess risk from interaction (p(RERI) ), attributable proportion (p(AR) ), synergy index (p(SI) )) with obese (BMI ≥30 kg/m(2) ) and diabetes. RESULTS: During a mean follow-up of 10.3 years, 9417 incident prostate cancers were diagnosed in 195,813 men. Diabetes and BMI were associated more strongly inversely with prostate cancer risk when occurring together (p(MI) = 0.0003, p(RERI) = 0.032, p(AP) = 0.020, p(SI) = 0.002). ABSI was associated positively in obese men (HR = 1.081; 95% CI = 1.030-1.135) and men with diabetes (HR = 1.114; 95% CI = 1.021-1.216). The inverse associations with obesity and diabetes were attenuated for high-ABSI ≥79.8 (p(MI) = 0.022, p(RERI) = 0.008, p(AP) = 0.005, p(SI) <0.0001 obesity; p(MI) = 0.017, p(RERI) = 0.047, p(AP) = 0.025, p(SI) = 0.0005 diabetes). HI was associated inversely in men overall (HR = 0.967; 95% CI = 0.947-0.988). Free testosterone (FT) was associated most strongly positively in normal weight men (HR = 1.098; 95% CI = 1.045-1.153) and men with diabetes (HR = 1.189; 95% CI = 1.081-1.308). Oestradiol was associated inversely in obese men (HR = 0.805; 95% CI = 0.682-0.951). The inverse association with obesity was stronger for high-FT ≥243 pmol/L (p(RERI) = 0.040, p(AP) = 0.031, p(SI) = 0.002) and high-oestradiol (p(RERI) = 0.030, p(AP) = 0.012, p(SI) <0.0001). The inverse association with diabetes was attenuated for high-FT (p(MI) = 0.008, p(RERI) = 0.015, p(AP) = 0.009, p(SI) = 0.0006). SHBG was associated inversely in men overall (HR = 0.918; 95% CI = 0.895-0.941), more strongly for high-HI ≥49.1 (p(MI) = 0.024). CONCLUSIONS: Obesity and diabetes showed synergistic inverse associations with prostate cancer risk, likely involving testosterone reduction for diabetes and oestrogen generation for obesity, which were attenuated for high-ABSI. HI and SHBG showed synergistic inverse associations with prostate cancer risk.