by David Y. Josephson, MD, FACS
For years, internists, endocrinologist and urologists alike have touted the benefits of testosterone replacement in men due to its protective affects on the cardiovascular system. Clinical trials had previously consistently shown improvements in a number of cardiac outcomes and reduction in adverse cardiovascular events (i.e. stroke, heart attack or death) in men with both clinical and laboratory evidence of hypogonadism who were treated with exogenous testosterone. However, a recent small randomized study Testosterone in Older Men with Mobility Limitations (TOM) trial was prematurely stopped after a small subset of men were noted to have adverse cardiac events after a 3 year period of testosterone replacement.
More recently however, a large retrospective observational study published in JAMA has created a lot of buzz and controversy after concluding that a subset of men in the Veterans Administration system who used testosterone therapy were at increased risk of adverse cardiovascular outcomes. The study cohort included over 23,000 men over the age of 60 in the VA system from 2005 to 2011 who underwent coronary angiography to work up heart disease and who also had a testosterone level checked. Of the twenty three thousand men, 14,000 were excluded: if their T levels were above 300 (not consistent with hypogonadism); if they had previously received T therapy; they had missing coronary anatomy data; or if their baseline Prostate Specific Antigen (PSA) was above 4 ng/ml.
When analyzing the raw data and as one would expect, of the 7400 men in the group not receiving T, 21% had adverse cardiovascular events compared to 10% in the 1220 patients in the Testosterone treatment group. However, because this was not a randomized comparison the study authors used a very complex statistical methodology incorporating inversed probability of treatment weighting to adjust for confounders to come up with adjusted rates of events that were quite the contrary to the raw percentage differences. By adjusting for over 30 variables and applying stabilized weights for each patient, the absolute rate of CV events were 20% in the no testosterone therapy group vs. 26% in the testosterone treated group.
Now, before we jump to conclude that T therapy is detrimental to the health of men with potential pre-existing heart disease the limitations of the study need to be exploited. First of all, this is a retrospective observational study and unmeasured cofounders and hidden bias will certainly exist. Second, as the authors point out, testosterone levels were recorded without note of the time of day when the lab was drawn. We all know that circadian effects of Testosterone will fluctuate and certainly underestimate the testosterone levels if labs are drawn after the early morning hours. In addition, patients who had started T therapy prior to getting cardiac angiography, those who had received T therapy after being diagnosed with an MI or those who had their hypogonadism diagnosed as part of labs outside the VA were excluded. This subset of patients accounted for about 4500 exclusions alone. Patients were considered to be part of the treatment group if they received a shot of intramuscular T or they received a prescription for a patch/gel without taking into account subsequent hormone levels of if the medication was actually taken. Compliance to taking medication and levels to confirm adequate treatment are not part of this analysis. The most concerning limitation however is the fact that outcomes were measured by ICD-9 codes and not actually confirmed by a formal chart review.
As a urologist who takes great pride in my patient outcomes, I am very skeptical of any study utilizing ICD-9 data for result analysis as there is usually a great component of subjectivity involved in coding. The authors do a very nice job exploring potential pathways of how Testosterone can increase CV risk through increases in platelet thromboxane A, plaque formation, increases in DHT induced smooth muscle proliferation and cytokine release. Despite these potential mechanisms and the findings of the article I am still hesitant to stop offering hormone replacement in my hypogonadism patients until a validation study is performed. While the article does raise some concerns about the links between cardiovascular events and T therapy, I am sure a currently accruing study the T Trial will ultimately provide better insight on the relationship.