Endogenous Testosterone Levels and Cardiovascular Risk: Meta-Analysis of Observational Studies

INTRODUCTION: The relationship between endogenous testosterone (T) levels and cardiovascular (CV) risk in men is conflicting. AIM: To verify whether endogenous T levels represent a possible risk factor for CV morbidity and mortality. METHODS:We conducted a random effect meta-analysis considering all the available data from prospective observational studies comparing subjects with baseline reduced endogenous T levels to those with higher T levels as derived from an extensive MEDLINE, Embase, and Cochrane search. The identification of relevant studies was performed independently by 2 of the authors (G.R. and G.G.C.), and conflicts resolved by the third investigator (M.M.). MAIN OUTCOME MEASURES: CV mortality and morbidity were investigated. RESULTS: After screening, 37 observational studies, published between 1988 and 2017 including 43,041 subjects with a mean age of 63.5 years and mean follow-up of 333 weeks, were considered. Low endogenous T at enrollment predicted overall and CV mortality, as well as CV morbidity, when both unadjusted and fully adjusted models were considered (odds ratio = 1.26 [1.17; 1.36], 1.54 [1.25; 1.89], and 1.17 [1.01; 1.36]; all P < .05 when overall mortality, CV mortality, and CV incidence and fully adjusted models were considered, respectively). The data were confirmed even when non-population-based studies were excluded from the analysis. Meta-regression analysis applied to the fully adjusted model showed that the risk of CV mortality was inversely related to mean age at enrollment (S = -0.014 [-0.017;-0.010] and I = 1.073 [0.806;1.339]; both P < .0001) and directly related to the prevalence of diabetes and to the proportion of active smokers. CLINICAL IMPLICATIONS:Low endogenous T levels in aging men can represent a possible CV risk factor. STRENGTHS & LIMITATIONS: The present data demonstrated, for the first time, that low T predicts not only CV mortality but also CV morbidity. Data derived from studies reporting information on CV mortality suggested major publication bias although they were confirmed applying Duval and Tweedie trim and fill method. However, observational studies should be considered with caution due to the lack of complete follow-ups and due to the poor management of missing data. CONCLUSION:The present meta-analysis shows that low T in aging men is a marker of CV risk. The possible benefits of T treatment in reducing this risk should be examined in longer-term, specifically designed trials.