The recent study from Vigen R et al. aimed at assessing the association between testosterone therapy and all-cause mortality, myocardial infarction, or stroke among 8709 male patients who underwent coronary angiography between 2005 and 2011 and had testosterone levels checked. Included in this retrospective analysis were those with testosterone levels suggestive of hypogonadism (<300 ng/dL). The authors divided hypogonadic patients into two groups, i.e. patients who received testosterone replacement therapy and patients who did not receive testosterone therapy. From the yearly analysis of the event rates the authors concluded that testosterone increases the cumulative risk of death, myocardial infarction or stroke by 29% after 3 years of treatment. The implications of the study are –if taken at face value- are important for patient management and of concern for both patients and treating physicians.
Reference:
Rebecca Vigen et al. Association of Testosterone Therapy With Mortality, Myocardial Infarction, and Stroke in Men With Low Testosterone Levels. JAMA. 2013;310(17):1829-1836. doi:10.1001/jama. 2013.280386
These results seem to be related more to methodological flaws rather than to a real detrimental effect of testosterone. The study is retrospective and based on ICD-9 codes and not on chart reviews. Importantly, the authors performed an arbitrary selection of patients were receiving or not receiving testosterone therapy. The authors excluded 4,468 patients with hypogonadism regardless whether they were receiving testosterone (2798 patients) or not.
In contrast with the authors conclusions, during a mean follow up of 2.3 years total mortality and occurrence of myocardial infarction in the study were lower in men receiving testosterone compared to men not receiving testosterone therapy (5.5% vs 9.1% and 1.9% vs 5.6%).  Also of importance, the results seem to be greatly influenced by the number of patients with a follow up longer than 2 years. The number of patients not receiving testosterone therapy in the Kaplan Meier Survival curves drops drastically after year 2 when the survival curve for patients not receiving testosterone therapy becomes unusually shallow.
Only from this time point onwards (year 3) the difference between treated and untreated patients became evident. This is of importance since in patients who were adherent to testosterone replacement therapy the mean number of days from the first to the last prescription of testosterone was around 1 year (376 days).
Another relevant issue is the fact that the authors have allocated a sizeable proportion of patients (17.6%) to the testosterone therapy group despite the fact that those had filled only 1 prescription of testosterone during the whole observation period while there is no information regarding the number of patients with less than 6 prescriptions. This imputation method is methodologically incorrect since intention to treat is to be used only in efficacy studies while for safety signals it is important to assess treatment adherence or at least a direct causality between long-term treatment and the observed effect.
Apart from the methodological issues of the study there is no biological plausibility to support the claimed effect reported in the paper. The supposed increased risk associated with testosterone therapy use was not related to differences in risk factors as both groups had similar blood pressure and LDL levels.  Several studies have demonstrated that testosterone replacement improves intermediate outcomes, myocardial ischemia and exercise capacity in patients at risk and in those with proven cardiovascular disease. Furthermore, previous studies have also reported a 39% reduction in mortality among patients receiving testosterone therapy. In conclusion, results presented in the report do not categorically support a causal relationship between testosterone replacement therapy and increased cardiovascular risk.
