Prevalence and prognosis of a low serum testosterone in men with type 2 diabetes: the Fremantle Diabetes Study Phase II

BackgroundBecause published studies have usually involved imprecise assays and selected patients with limited additional data and follow-up, the consequences of a low serum testosterone in diabetes are unclear. This study assessed the prevalence, associates and prognosis of a low testosterone in community-dwelling men with type 2 diabetes. DesignLongitudinal observational study. Patients788 men (mean SD age: 658 113 years) followed for 40 11 years. MeasurementsSerum testosterone, SHBG, erectile dysfunction (ED; Sexual Health Inventory for Men score <22), anaemia (haemoglobin <130 g/l), all-cause mortality. ResultsThe mean +/- SD total serum testosterone by liquid chromatography/mass spectrometry was 131 +/- 59 nmol/l (306% <10 nmol/l). Most men with a total testosterone <10 nmol/l (670%) had a normal/low serum LH. Serum testosterone was independently associated with anaemia (P < 0001), but not ED (P = 080), in logistic regression models. The optimal cut-point (Youden Index) for anaemia was 98 nmol/l (sensitivity 536%, specificity 754%). During the follow-up, 102 men (129%) died. There was a U-shaped relationship between total serum testosterone quintiles and death (P = 0003, log rank test). The middle quintile (>111 to 137 nmol/l) had the lowest risk and there was a 78% increased risk for highest (>169 nmol/l) vs lowest (86 nmol/l) quintile in Cox proportional hazards modelling (P = 0036). Free serum testosterone and SHBG quintiles were not associated with death. ConclusionsThese data provide some support for the general conventional serum testosterone <10 nmol/l cut-point in identifying an increased risk of anaemia and the subsequent death in men with type 2 diabetes, but indicate that high-normal levels are also an adverse prognostic indicator.