Introduction: Following reports of an increased risk of cardiovascular events in the Testosterone in Older Men with Mobility Limitations trial and the cohort study by Vigen et al (also reviewed in this issue of EndoScan), Finkle et al designed this trial to investigate the effects of testosterone supplementation on a single cardiovascular outcome—nonfatal myocardial infarction (MI)—and whether this effect was altered by age or the presence of preexisting cardiovascular disease.
Methods: The authors reviewed healthcare records of 55,593 men filled a first prescription for testosterone therapy between January 2008 and September 2010. Of this group, 48,539 were <65 years and 7,054 were ≥65 years. Data was unavailable on how much of the prescriptions were actually used. None of the men had a history of MI.
The authors measured the incidence of MI in the 90 days following the initial prescription compared with the rate in the one year prior to the initial prescription (post-/preprescription rate ratio [RR]). Comparators included the MI rates of the same men in the year preceding prescription, and MI rates of a cohort before and after filling a first prescription for phosphodiesterase type 5 inhibitors (PDE5I) sildenafil/tadalafil. Covariates included age, comorbidities, and prescription medications.
Results: In the overall group, the post-/preprescription RR for testosterone use was 1.36. In men ≥65 years, the RR for testosterone therapy was 2.19 and for PDE5I therapy was 1.15; the ratio of the rate ratios for testosterone use relative to PDE5I use (RRR) was 1.90. The RR for testosterone therapy (but not for PDE5I therapy) increased significantly with age (P=0.03).
In men <65 years, the RR was increased only in men with a history of cardiovascular disease with rates of 2.90 for testosterone therapy and 1.40 for PDE5I. The RRR was 2.07in this group of younger men with coronary artery disease, but was 0.90 in younger men without a history of cardiovascular disease.
Conclusion: The study showed a twofold increase in MI risk shortly after beginning testosterone therapy among men younger than 65 years who have a history of cardiovascular disease. In addition, a two-fold increase in MI risk was found shortly after treatment initiation in men older than 65 years.
Finkle WD, Greenland S, Ridgeway GK, et al. Increased risk of non-fatal myocardial infarction following testosterone therapy prescription in men. PLoS One. 2014 Jan 29;9(1):e85805.
Commentary by Tamara L Wexler MD, PhD
Finkle and colleagues used a large database to look at the frequency of acute non-fatal MI in the 90 days following initial testosterone prescription. The investigators found a relative risk of MI in men after testosterone therapy, increasing with age. For men younger than age 65 years, the risk was confined to those who carried a diagnosis of heart disease. The relative risk of MI was approximately two-fold in men ≥65 years within 90 days of initiating testosterone therapy.
Cohort sizes were large, with a predominance of men <65 years. In addition to looking at numerous covariates, including comorbidities and prescription medications, the study design shows a clever use of comparison groups, as a type of control – not only the pre-prescription period for the same testosterone therapy cohort, but a PDE5I cohort. For comparison between groups, the PDE5I cohort was weighted to match distributions of covariates in the testosterone therapy group.
The risk ratio of post- to pre-prescription testosterone therapy in men ≥65 years was 2.19. Comparing that RR to the weighted RR for PDE5I group, the RRR for men ≥65 years was 1.90; using the more standard unweighted Poisson regression, RRR was higher, at 2.27 for men ≥65 years. The risk of MI in men who did not refill their testosterone prescription was also examined in the 91 to 180 days post-prescription and found to be less than the initial 90 days; assuming that those patients had discontinued testosterone use, this finding suggests a reduction in relative risk after discontinuation of testosterone (although there is no way to determine compliance).
Retrospective cohort analyses have inherent limitations and should not be considered to have the same implications as randomized clinical trials. However, the very large size of the cohorts, and the use of various comparator groups, lends weight to the findings. While the authors took measures to control for biases, the study is observational, limiting what is known about the subjects. Testosterone (and estradiol) levels are not known, nor are the rationales for initiating testosterone treatment. This is relevant not only because testosterone level may itself impact morbidity, but because replacement may act differently in men with hypogonadism (such as from pituitary disease) and in men with declining levels due to aging (or men who do not meet criteria for hypogonadism).
The relative risk of MI may be particularly pronounced for older men—of note, the relative risk was even higher in men ≥75 years in this study. Increased effect in older men may be due to factors other than age, such as undiagnosed previous heart disease. The Veterans Affair study by Vigen et al (also reviewed in this issue of EndoScan) involved men undergoing coronary angiography, which may enrich for a population with suspected cardiovascular disease.
This study supports the need for randomized controlled trials, and for discussing potential adverse cardiovascular effects of testosterone therapy with patients, particularly among those with cardiovascular disease.