Low circulating free testosterone is associated with reduced incidence of prostate cancer: A pooled analysis of individual participant data from 20 prospective studies
Session type: Proffered paper
Theme: Epidemiology and prevention
Experimental and clinical evidence implicates testosterone in the etiology of prostate cancer, but previous findings from prospective epidemiological studies do not support an association. The androgen saturation model may explain the disparity, hypothesising that the stimulatory effects of androgens increase up to androgen receptor saturation but that above this point any further increase in androgen concentration will not induce a further increase in prostate tissue growth. Because the saturation point is thought to be low, until now there have been insufficient prospective data available to test this theory. We aimed to test the androgen saturation model, hypothesising that men with very low circulating free testosterone concentrations will have a reduced risk of prostate cancer.
We analysed individual participant data from 20 prospective studies, with up to 6,933 prostate cancer cases and 12,088 controls in the Endogenous Hormones, Nutritional Biomarkers and Prostate Cancer Collaborative Group. Conditional logistic regression was used to estimate odds ratios for prostate cancer based on study-specific tenths of free testosterone.
Men in the lowest study-specific tenth of free testosterone had a lower risk of prostate cancer (OR=0.79, 95% CI 0.70-0.88; P<0.001) compared with men in the 2nd-10th tenths. There was no evidence of heterogeneity in the association by obesity, age or family history. However, there was heterogeneity by tumour grade (Phet= 0.002), with a decreased risk of low-grade prostate cancer (OR=0.76, 95% CI 0.67-0.88) and an increased risk of high-grade disease (OR=1.65, 95% CI 1.04-2.63) in men with very low levels of free testosterone.
These results support the androgen saturation model in relation to prostate cancer risk. Further large-scale studies which examine co-morbidities and data across a broad range of circulating hormone concentrations are needed to elucidate whether the observed associations are causal or attributable to bias.