This study aimed to test the dose-response relationship between fasting blood glucose (FBG) levels and risk of prostate cancer.

A systematic search was done of PubMed and Scopus from their inception up to January 2017. Prospective and retrospective studies reporting risk estimates of prostate cancer for two or more categories of blood glucose levels were identified, and two independent authors extracted the information. Relative risk (RR) was calculated using random-effects models and pooled.

Ten prospective cohort studies, one nested case-control study, one case-cohort study and three case-control studies (total n=1,214,947) involving 12,494 cases of prostate cancer were reviewed. The pooled RR of prostate cancer for the highest vs. lowest category of FBG was 0.88 (95% CI: 0.78–0.98, I²=25.5%, n=15 studies). A 10mg/dL increment in FBG level was not associated with risk of prostate cancer (0.98, 95% CI: 0.96–1.00, I²=45.4%, n=11 studies). Subgroup analyses yielded a significant inverse association only in the subgroup of cohort studies. Non-linear dose-response meta-analysis showed a very slight decrement in risk with increasing FBG levels. Sensitivity analyses using cohort studies showed a steep decrease in risk along with an increase in FBG from baseline levels of ≈70mg/dL across prediabetes and diabetes ranges.

Higher FBG levels are associated with lower risk of prostate cancer in cohort studies, but not in case-control studies, findings that limit interpretation of our present results.