INVITED REVIEW
Ahead of print publication  

Testosterone treatment and cardiovascular events in prescription database studies


 Department of Psychiatry and Behavioral Sciences, University of Washington School of Medicine, VA Puget Sound Health Care System, 1660 South Columbian Way Seattle, WA 98108, USA

Date of Submission09-Dec-2016
Date of Acceptance15-May-2017
Date of Web Publication11-Aug-2017

Correspondence Address:
Molly M Shores,
Department of Psychiatry and Behavioral Sciences, University of Washington School of Medicine, VA Puget Sound Health Care System, 1660 South Columbian Way Seattle, WA 98108
USA
Login to access the Email id

Source of Support: None, Conflict of Interest: None

  Abstract 

Over the past decade, there has been a substantial increase in the number of men who are treated with testosterone. Despite this increase in the use of testosterone, the risks of adverse cardiovascular events are unclear as meta-analyses have reported conflicting findings and no clinical studies have been large enough or long enough to adequately assess for cardiovascular risks. The goal of this paper is to review large prescription database studies of testosterone treatment and adverse cardiovascular events and mortality with the aim of providing some guidance for clinicians and researchers in this controversial area.

Keywords: cardiovascular diseases; cohort studies; myocardial infarction; stroke; testosterone/adverse effects; thrombosis


Article in PDF

How to cite this URL:
Shores MM. Testosterone treatment and cardiovascular events in prescription database studies. Asian J Androl [Epub ahead of print] [cited 2017 Aug 17]. Available from: http://www.ajandrology.com/preprintarticle.asp?id=212903


  Introduction Top


Testosterone (T) was discovered in the 1930s and for decades it was primarily prescribed for the treatment of hypogonadism due to genetic abnormalities or disorders of the pituitary-testicular axis. However, an increased interest in T as a potential agent to enhance well-being and possibly ameliorate symptoms of aging occurred after several studies reported that low serum T concentrations were common in aging and were associated with obesity, insulin resistance, diabetes, sexual dysfunction, and mortality.[1],[2],[3],[4],[5],[6],[7],[8],[9] In conjunction with an increased interest in T and the approval of more convenient T formulations (patches and gels), the use of T increased dramatically, particularly in the USA.[10] This substantial increase in T prescriptions led to concerns that T was being prescribed inappropriately, particularly after a study reported that 26% of men did not have a serum T concentration testing before being prescribed T.[11] Additional concerns occurred after two cohort studies,[12],[13] and one meta-analysis[14] found that T was associated with an increased risk for cardiovascular (CV) events. These led to an FDA Safety Communication in 2015 that the cardiovascular effects of T treatment were unclear[15] and that prescribers should review this potential risk with men before initiating T treatment.

This paper will review retrospective, observational studies that used prescription databases to examine the association of T treatment with cardiovascular events, including myocardial infarction, stroke, thrombosis, and mortality. The aim of this review is to critically assess and summarize the results of each study to provide some guidance for clinicians and researchers in this controversial area.


  Materials and Methods Top


A literature search was done on the association of T treatment and CV events and mortality from 1948 to October 2016 using Ovid MEDLINE database (http://ovid.com/site/catalog/databases/901.jsp). The search terms included testosterone, cardiovascular, myocardial infarction, stroke, mortality, thrombosis, cohort studies, retrospective studies, and database. Additional references were obtained from reviews and meta-analyses of T treatment. Eligible studies were either retrospective cohort or case-control studies that utilized prescription or insurance claims databases and were published in English. We excluded meta-analyses, clinical trials or studies, case reports, and chart reviews.[16],[17],[18]


  Results Top


The search strategy yielded 14 papers that were published between 2012 and 2016. [Table 1] and Supplementary Table 1 [Additional file 1] provide additional details.
Table 1: Summary of database studies


Click here to view


Shores et al.[19] examined the association between T treatment and mortality using a regional Veterans Health Administration (VHA) database of men with a low T concentration and no history of prostate cancer or prior T treatment. T treatment was modeled in a time-varying manner. Specifically, men were classified as untreated until the date a T prescription was filled and then were classified as treated on the date the prescription was filled. A propensity score was used to adjust for the observational study design. The propensity score is the probability of being treated based on specified covariates. The covariates used to estimate the propensity scores included baseline T concentration, osteoporosis, sexual dysfunction, sleep apnea, elevated hematocrit, body mass index (BMI), medical morbidity, hospitalizations, diabetes, and sexual dysfunction. One thousand and thirty-one men (mean age, 62 years) were eligible and 38.6% (398) of them were T treated. In adjusted analyses, T treatment was associated with a decreased risk for mortality, with a hazard ratio (HR) of 0.64 (95% confidence interval [CI]: 0.44-0.95). There were no interactions with age, diabetes, or coronary heart disease and no differences in analyses that excluded men who died within the 1st year.

Prior to this study, several studies had reported that low serum T concentrations were associated with increased mortality, but had not examined whether T treatment was associated with mortality. This study was one of the first to examine the association of T treatment with mortality in men. Weaknesses of the study were the small size, limited information on T dose, T duration, and follow-up T concentrations and that covariates were adjusted only at baseline and were not updated. Strengths of the study were that it restricted the cohort to men with low T concentrations, which increased the probability that treated and untreated men were androgen deficient.

Vigen et al.[13] used a national VHA database to examine the risk for composite CV events (myocardial infarction [MI], ischemic stroke, and all-cause mortality) in 8709 older veterans (mean age, 63 years) with low T concentrations who had a coronary angiogram. More than 80% of men had prevalent coronary artery disease (CAD) by angiography. The researchers modeled T as a time-varying covariate and adjusted for confounders using stabilized inverse probability of treatment weights that were updated whenever a new event occurred. In adjusted analyses, T treatment was associated with a 29% increased risk for composite CV events (HR: 1.29, 95% CI: 1.04-1.58). However, in unadjusted analyses, T-treated men had a much lower event rate (10%) than untreated men (21%).

This study generated a great deal of publicity, as it was the largest study at that time to report an association of T treatment with an increased risk of CV events. The authors were criticized for not discussing the unadjusted absolute event rates or that the statistical adjustments completely reversed the results of the unadjusted absolute event rates. Other weaknesses were that cohort entry time was not defined and no adjustments were made for baseline T concentrations, which were much lower in the treated men (175 vs 205 ng dl−1 ). This may have influenced the results because low T concentrations are associated with increased mortality.[4],[5],[9],[20] Another limitation was that there was a long lag time from study entry to T treatment (531 days) and the authors excluded men who had MIs during that time. This exclusion removed men with MI events from the untreated group and biased the results in favor of nontreatment. Finally, this was an unusual cohort of elderly male veterans with very high medical morbidity, so these results may not generalize to typical men who use T and tend to be healthier in their 50s.[21] Strengths of the study were the large size, baseline low T concentrations, and that confounders were adjusted for over time. In conclusion, this paper was one of the first studies to show that T treatment was associated with an increased risk for composite CV outcomes, but the paper was heavily criticized due to methodological flaws.[22]

Finkle et al.[12] used an insurance claims database to examine MI risk in middle-aged men (mean age, 54 years) by comparing the MI rate in the 3 months after T treatment to the MI rate in the 12 months before treatment. All the 55 593 men in the study were treated with intramuscular (IM) T. The overall risk for MI was greater after treatment than before treatment (HR: 1.36, 95% CI: 1.03-1.81), with an increased risk for MI in men aged 65 years or older (HR: 2.19, 95% CI: 1.27-3.77). In men <65 years old, those with a history of heart disease had greater risk (HR: 2.90, 95% CI: 1.49-5.62). They also compared the MI rate in T-treated men to men treated with a phosphodiesterase type-5 inhibitor (PDE5I) and found that T treatment was associated with an increased risk with a rate ratio of 1.90 (95% CI: 1.04-3.49) compared to men treated with a PDE5I. A major limitation was that the results would be biased if clinicians were hesitant to prescribe T after an MI. If so, this would bias the results toward increased risk with T treatment. Finally, PDE5I-treated men are not an appropriate comparison group for several reasons. First, many clinicians do not prescribe a PDE5I for men with known or symptomatic CAD (due to the possibility of myocardial ischemia from increased cardiac demand associated with sexual intercourse). Secondarily, men treated with a PDE5I might be less likely to have an MI because PDE5Is have beneficial cardiovascular effects.[23] Thus, the lower MI rate in PDE5I-treated men might have been due to less prevalent CAD and due to cardioprotective effects of PDE5Is rather than to greater risks with T treatment.

The strengths of the study were that it was a very large study and comparing MI rates in the same man before and after treatment eliminated between-subject confounding. In conclusion, Finkle et al.[12] reported an increased MI rate post-T treatment compared to pretreatment.

Baillargeon et al.[24] conducted a cohort study using Medicare data to examine MI risk in 6355 men who started IM T and were matched to 19 065 controls. Exclusions included T treatment in the 12 months prior to IM T and end-stage renal disease. An MI prognostic score was used to stratify MI risk. In adjusted analyses, IM T was not associated with overall risk for MI (HR: 0.84, 95% CI: 0.69-1.02). However, in men with a high prognostic score for MI, T treatment was associated with a decreased risk for MI (HR: 0.69, 95% CI: 0.53-0.92). In sensitivity analyses that used different criteria for cohort entry and follow-up times, there was no increased risk for MI with IM T.

Limitations included lack of baseline or follow-up serum T concentrations, so it was not clear whether the men met minimal criteria for T treatment and whether therapeutic T concentrations were reached following T treatment. In addition, men were treated only with IM T and treatment was of short duration (49% with two or fewer T-injections). Hence, it is unclear whether these results generalize to men who use transdermal T or use T for a longer duration. A final limitation was that the study did not exclude men who had a history of a prior MI, which is considered a weakness in scales that rate the quality of observational studies.[25],[26] A higher quality study is one that excludes patients who already had the studied outcome as this design allows for a clearer ascertainment of risk related to medication by decreasing the risk related to baseline disease. The strengths of this study were that it examined MI risk by risk stratification for MI. In conclusion, this study found that elderly men treated with IM T for a short time had no increased risk for MI, but in men at greatest risk for MI, T treatment was associated with a decreased risk for MI.

Sharma et al.[27] used a VHA database to examine the association between T treatment and CV events and mortality in 83 010 men. Men in the cohort had at least two low T concentrations prior to T treatment and were excluded if they had prior T treatment, MI, or ischemic stroke. T treatment status was characterized as T-treated normal T concentrations after treatment (normalized T), T-treated low T concentrations after treatment, or not T treated. In adjusted analyses, men with normalized T concentrations after T treatment had a lower risk for mortality (HR: 0.44, 95% CI: 0.42-0.46), stroke (HR: 0.64, 95% CI: 0.43-0.96), and MI (HR: 0.76, 95% CI: 0.63-0.93) than untreated men. T-treated men who had low T concentrations after treatment had similar risks compared to untreated men.

Limitations of this study were that they did not exclude men with prostate cancer, who may be less likely to be T treated and may have greater mortality, which could bias the results in favor of T treatment. Strengths of the study design were that men with a history of MI or stroke were excluded. Other strengths were that men were required to have two low T concentrations before T treatment, which was consistent with clinical guidelines and made it more likely that the men were hypogonadal. Another strength was the use of an analytic approach with a propensity score analysis and matching. A final strength was that outcomes were assessed in relation to follow-up T concentrations. The study found that T risks differed by follow-up T concentrations with men who achieved physiologic T concentrations having decreased risks, while men with persistently low T concentrations had risks that were similar to untreated men. This finding of differential risks associated with follow-up T concentrations was a noteworthy finding as it indicated the importance of achieving physiologic T concentrations with T treatment. This finding might explain conflicting results of prior studies because most studies did not evaluate T concentrations after treatment. In summary, in this well-designed study, Sharma et al.[27] found that T treatment with normalized T concentrations following treatment was associated with a 56% decreased risk for mortality and 24%-36% decreased risk for CV events compared to men with persistently low T concentrations.

Layton et al.[28] examined the risk for hospitalization, thrombosis, death, or a composite CV outcome (MI, angina, stroke) by different T formulations (IM, patch, or gel), using three databases: an insurance database (n = 515 132), Medicare database (n = 22 376), and an United Kingdom (UK) database (n = 6607). The IM T formulations consisted primarily of testosterone cypionate or enanthate, which are shorter acting than long-acting testosterone undecanoate which is typically administered every 3 months. Men were excluded from the study if they had a history of MI, stroke, or thrombosis or if they used T in the 6 months prior to the T prescription. T concentrations were obtained in 52%-70% of the cohort, but there was limited information on the results and documented low T concentrations were available in only 1.5% of the insurance cohort, 20% of the UK cohort, and none of the Medicare cohort. Men were followed up for 1 year after T treatment, and risks with IM T and T patch were compared to risks with T gel. In pooled analyses, that combined the results of the three databases, IM T compared to T gel was associated with a 21%-30% increased risk for MI, stroke, and composite CV events and a 34% increased risk for mortality, but no increased risk for thrombosis. The T patch had similar risks compared to T gel. There were no significant differences in analyses that excluded men with cancer or that used a shorter follow-up time.

Limitations were that there was no comparison of T treatment to a control group of untreated men, so the study does not address whether T treatment increases risks compared to untreated men. Other limitations were that T concentrations were known in only a small percentage of men, and there was no information on outpatient thrombotic events, which may have affected the results. Strengths of this study were that they excluded men with a past history of MI, stroke, or thrombosis. In conclusion, the Layton study found that IM T compared to T gel was associated with increased risks for MI, stroke, composite CV events, and all-cause mortality, but not for thrombosis.[28]

Tan et al.[29] evaluated the rate of MI and stroke in T-treated men who were followed at "Low T" centers. The cohort consisted of 19 968 men, aged 20-86 years, who were diagnosed with hypogonadism before being prescribed T and were monitored with a standardized protocol. The rates of MI and stroke in the T-treated men were compared to rates in two different community registries. Men in the community registries were older than T-treated men, but no adjustments were made for any differences between men in the registries and T-treated men. The authors reported that the incidence rate ratios (IRR) for MI (IRR: 0.14, 95% CI: 0.08-0.18) and stroke (IRR 0.11, 95% CI: 0.02-0.13) were markedly lower in T-treated men compared to men in community registries.

A major weakness of the study was that the T-treated men were not compared to a similar population of younger men, followed in a specialty clinic, but instead were compared to older, community dwelling men. Furthermore, limited information was provided on differences in age and medical morbidity between groups and no adjustments were made for differences between groups. Finally, the huge reduction in risk (86%-89%) associated with T treatment is not believable and is likely related to differences between groups. The strengths of this study were that the T-treated men were diagnosed with hypogonadism prior to T treatment and were treated with a standardized protocol. In conclusion, this study found that men followed at T centers had a large reduction in MI and stroke rate compared to men in community registries, but the study was seriously flawed because the comparison groups were drawn from different communities and no adjustments were made for differences between groups.

Etminan et al.[30] conducted a large case-control study to assess T treatment and nonfatal MI using data from an insurance claims database. MI cases (30 066) were matched to 120 264 controls by age, cohort entry date, and length of follow-up. T exposure was classified as current, past, or nonexposed. The study found no overall risk with current or past T, or by T formulation. However, men who initiated T in the 3 months prior to MI had a 41% increased risk for MI (HR: 1.41, 95% CI: 1.06-1.87), and this risk was greater in men with CAD (HR: 1.78, 95% CI: 0.93-3.40).

A weakness of this study was that men with prior MI were not excluded. Inclusion of men with prior MI makes it more difficult to ascertain whether risk for MI was associated with T treatment or to the underlying CV disease. Strengths were the large number of cases, that most cases were matched to three controls, and that the study examined whether there were greater risks associated with T initiation. In conclusion, this large case-control study found no overall risk for MI with T, but an increased risk within 3 months of T initiation, which was greater in men with a history of CAD.

Wallis et al.[31] examined the association of T treatment with mortality and composite CV events (MI, stroke, thrombosis) and prostate cancer in a matched cohort study of men aged 66 or older using eight Canadian databases. T-treated men (10 311) were matched to 28 029 controls based on age, region, diabetes, medical comorbidity, and index year. Testosterone treatment was modeled in a dichotomous manner (any T or no T) and as a cumulative dose by days of dispensed T. In adjusted analyses, T treatment was associated with an increased risk for composite CV events (HR: 1.10, 95% CI: 1.03-1.20), but decreased risk for mortality (HR: 0.88, 95% CI: 0.84-0.93). In analyses by treatment duration, risks for CV events and mortality were greater with initiation and then decreased with a longer treatment time.

The weaknesses of the study were that there were no adjustments for risk factors that affect thrombosis (e.g., surgery, cancer, hospitalization, or medications) that could have influenced the estimates of composite CV risk. The strengths of the study were that it was a large, matched, population-based study that excluded men with baseline CVD, had a follow-up time of over 5 years, and examined risks associated with the duration of T treatment. In summary, in a matched cohort study, Wallis et al.[31] found that there was a modest increased risk for composite CV events and decreased risk for mortality with greatest risks for composite CV events and mortality during the first 4 months of treatment.

Anderson et al.[32] conducted a study of 4736 men in Utah to examine the association of T treatment and composite CV events (MI, stroke, total mortality) and specific CV events. Inclusion criteria included age ≥50 years, a baseline low T concentration, and a follow-up T concentration. Men were excluded if they had a history of cancer. T exposure was classified by follow-up T concentrations (low, normal, or high) at 1 and 3 years. All the men with normal or high follow-up T concentrations were T treated and 18% of men with follow-up low T concentrations were T treated. At 3-year follow-up, men with a normal follow-up T concentration had decreased composite CV events (HR: 0.74, 95% CI: 0.56-0.98) and decreased mortality (HR: 0.65, 95% CI: 0.47-0.90) compared to men with low T concentrations. T treatment was not associated with specific outcomes of MI or stroke. Results were similar in analyses stratified by age and history of CVD.

The weaknesses of the study were that there were no exclusions for past MI or stroke, no information about T treatment, or when the follow-up T concentration was obtained in relation to T treatment. In addition, this was a relatively small study, with few events (only 18 MIs and 12 strokes in men with low T concentrations), which limits the power of the study. The strengths of the study were that all men had baseline low T and follow-up T concentrations. In summary, in men who had normalization of T concentrations after T treatment, there was a 25%-35% decreased relative risk for composite CV events and mortality compared to men with persistently low T concentrations but no significant association with MI and stroke as separate outcomes.

Baillargeon et al.[33] conducted a case-control study of T treatment and thrombosis using an insurance database by examining the use of T 15 days prior to a thrombotic event. They matched 7643 cases of thrombosis to 22 929 controls based on age, region, hypogonadism, prothrombotic disorders, and thrombosis date (index date). They excluded men who had risk factors for thrombosis (hospitalization, anticoagulant use, or cancer) in the 30 days, 90 days, and 12 months prior to the index date. In adjusted analyses, there was no association of T treatment or T formulations with thrombosis and no interaction with prothrombotic disease. These results were unchanged in analyses using different time periods for T use prior to the thrombotic event (30 and 60 days). In separate analyses, there was an increased risk for thrombosis with megestrol acetate, corticosteroids, and nonsteroidal anti-inflammatory drugs (NSAIDs) and a decreased risk with antiplatelet drugs.

Limitations of the study were the small number of T-treated men (158 in the cases) and that no adjustments were made for smoking. Strengths of the study were the large size and that it matched cases with up to three controls. In addition, the study confirmed that medications that are known to be associated with thrombosis were associated with thrombosis in this cohort, which lent validity to the results. In summary, this study found no increased risk for thrombosis with T treatment and no interaction of T with prothrombotic disease. The finding of no interaction with prothrombotic disease is of particular interest given prior small studies that had reported an association of T treatment with thrombosis in patients with baseline prothrombotic disease.[34],[35]

Sharma et al.[36] examined the risk of T treatment for thrombosis in a VHA database of 71 407 men who had repeatedly low T concentrations. T treatment was characterized as follows: not T treated, T treated with a normal follow-up T concentration, and T treatment with a low follow-up T concentration. They excluded men who had only one low T concentration, history of thrombosis, hypercoaguable disorders, cancer, or were treated with T or warfarin. The groups were matched on age, BMI, diabetes, congestive heart failure, and chronic kidney disease. A propensity score analysis was done to adjust for differences between the groups. In adjusted analyses, there were no differences in thrombosis between T-treated normalized T and untreated men (HR: 1.10, 95% CI: 0.78-1.54) and no differences between T-treated low T and untreated men (HR: 1.14, 95% CI: 0.78-1.65).

Limitations of the study were that there were no adjustments for surgery, hospitalization, or trauma, which could be associated with thrombosis. Strengths of the study were that cohort entry required two low T concentrations, which is consistent with recommended clinical guidelines. Other strengths were that they examined whether T treatment resulted in normalization of T concentrations. This is important to assess as men with low T after treatment may have different risks than men with normalized T after treatment. In summary, this study found that T treatment was not associated with an increased risk for thrombosis in older men with baseline low T concentrations.

Martinez et al.[37] used a clinical database from the UK to examine risk for thrombosis in a large case-control study of 19 215 thrombosis cases matched to 909 530 controls. The mean age was 65 years and thrombosis was assessed using a validated algorithm with a sensitivity and specificity >92%. Testosterone exposure was examined as current treatment, recent treatment, or not exposed. Cases were matched with up to fifty controls based on date of thrombosis, age, pathologic hypogonadism, and risk factors for thrombosis (surgery, hospitalization, trauma, or cancer). Controls had at least 2 years of baseline data and no history of thrombosis. Analyses adjusted for demographics, BMI, smoking, drinking, prothrombotic disease, polycythemia, chronic obstructive pulmonary disease (COPD), diabetes, heart failure, MI, peripheral vascular disease, stroke, medications associated with thrombosis (megestrol acetate, NSAIDs, corticosteroids, antiplatelet medication), and symptoms that may have led to a T prescription (e.g., osteoporosis, sexual dysfunction, fatigue, depression, hot flushes). Current or recent T treatment was not associated with thrombosis, but T initiation was associated with a 63% increased risk for thrombosis (HR: 1.63, 95% CI: 1.12-2.37).

Potential weaknesses of the study were that T treatment was uncommon with only 69 T-treated men in the cases and 1251 in the controls. Strengths of the study were the use of a validated algorithm to detect thrombosis, exclusion of men with a prior history of thrombosis, and that risk was examined by the timing of T treatment. In conclusion, in this large, well-designed, case-control study, T was associated with an increased risk for thrombosis within the first 6 months of treatment but not associated with risk after that.

Li et al.[38] used data from an insurance claims database to examine the association of T treatment and thrombosis in men using two different study designs - a retrospective cohort design and a nested case-control design. Risk was examined by any T treatment, T formulation (gel, IM, patch, or other), and age (>65 or ≤65 years). Inclusion criteria were enrollment in a health plan and hypogonadism, which was defined as either T treatment or a diagnosis of hypogonadism. Exclusion criteria were a history of thrombosis in the year prior to cohort entry. The primary outcome was incident idiopathic thrombosis, which was defined as a thrombotic event that was not associated with risk factors for thrombosis.

For the retrospective cohort design, T-treated men were matched to untreated hypogonadal men based on their propensity score.[19] The index date was the date of the first T prescription for treated men and a randomly assigned date for untreated men. In adjusted analyses, there was no association of T treatment, T formulation, or age with idiopathic thrombosis. These results were unchanged in a sensitivity analysis in which the outcome was all thrombotic events.

For the nested case-control design, cases (men with thrombotic events) were selected from the original cohort and were matched with up to four controls based on age and index date. Testosterone treatment was modeled by T formulation and by current or past T treatment (thrombosis occurred more than 90 days after the end of the T prescription). T formulation and treatment (current or past) were not associated with the primary outcome of idiopathic thrombosis. In contrast, in sensitivity analyses with an outcome of all thrombotic events, IM T and past T treatment were associated with thrombosis, while current T treatment was not associated with thrombosis. The authors felt it was unlikely that past T treatment was actually associated with thrombosis and that these were likely spurious results related to residual confounding.

Several weaknesses and methodologic issues limit this paper. First, the data came from an insurance claims database and there was no information on serum T or prostate-specific antigen (PSA) concentrations, which limits the ability to assess whether men were appropriately treated with T. Another weakness was a low concordance rate of only 70% for the primary outcome of idiopathic thrombosis. Other weaknesses were an incomplete description of the adjusted analyses in the cohort study and in the case-control study that controls were matched to cases only on age and index date and were not matched on other potentially relevant covariates. In addition, in the case-control study, the investigators did not adjust for multiple comparisons (28 comparisons), which may have led to false-positive results. A potential strength of the study was the use of two different study designs. However, the case-control analysis was flawed due to limited matching of the controls to the cases and due to a lack of adjustment for multiple comparisons. In summary, this study found no association between T treatment and thrombosis with the primary outcome of idiopathic thrombosis in analyses using two different study designs. In contrast, in sensitivity analyses, IM T and past T treatment were associated with thrombosis; however, the above methodologic weaknesses limit the validity of the results of this sensitivity analysis.


  Summary Top


In 14 retrospective prescription database studies, most studies found that T treatment was not associated with increased risks for thrombosis, MI, stroke, composite CV events, or mortality.

Summary of results by specific outcomes

MI


In several large studies of MI that included 153 269 T-treated men, most studies found no increased risk for MI with T treatment. The largest of these was a well-designed study that included approximately 65 000 T-treated men and reported a 24% decreased risk for MI with T treatment.[27] Another study found no overall risk for MI, but detected a 41% increased risk for MI with T initiation.[30] A final study reported a 36% increased MI rate post-T treatment compared to pre-T treatment.[12] However, those results could be biased if clinicians are less likely to prescribe T treatment following an MI.

Stroke

In three studies of stroke that included 88 852 T-treated men, T treatment was not associated with an increased risk for stroke. The largest of these was a well-designed study that included approximately 65 000 T-treated men and reported a 36% decreased risk for stroke with T treatment.[27] Another study also reported that T was associated with decreased stroke risk, but that study had serious methodological flaws, which limits the validity of its findings.[29] Finally, in a smaller study, with a low number of strokes, there was no association of T treatment with stroke risk.[32]

Mortality

In four studies, including 69 282 men, T treatment was associated with a decreased risk for all-cause mortality ranging from 22% to 66%.[19],[27],[31],[32] No study found an overall increased mortality risk, but one study found a transient 11% increased mortality risk with T initiation.[31]

Composite CV outcomes

Three studies, including 15 613 T-treated men, reported conflicting results on the association of T treatment with composite CV outcomes. One study reported that T was associated with a decreased risk for composite CV events.[32] In that study, much of the decreased risk for composite CV events was due to a decreased risk for mortality. Another study reported a 10% increased risk for composite CV events, which varied by T timing, with greater risk associated with T initiation and shorter treatment.[31] Finally, a third study reported an increased risk for composite CVD of 29% in a cohort of men with high medical morbidity and CAD.[13] However, that study was heavily criticized due to methodologic flaws.[22]

Thrombosis

In four studies of thrombosis, which included over 62 000 T-treated men, T was not associated with an overall risk for thrombosis.[33],[36],[37],[38] Surprisingly, two studies also found that men with baseline risk factors for thrombosis had no increased risk for thrombosis with T treatment.[33],[37] However, T initiation was associated with a 63% increased risk for thrombosis in one study.[37]

Limitations and strengths of the studies

There are several weaknesses to the prescription database studies. Many of the studies did not exclude men with a prior history of MI, stroke, or thrombosis and this could have influenced results, as men with a history of prevalent CVD are at an increased risk for recurrent CV events. Other limitations were that there was generally limited information on T dosage and duration, which is necessary to assess whether there is an increasing risk associated with cumulative dose of T. Another major weakness was that many studies had no information on baseline T concentrations and likely consisted of a heterogeneous group of men with both low and normal T concentrations. Such heterogeneous groups could lead to differences in outcomes, as low T concentrations are associated with CV outcomes and mortality.[6],[7],[8],[39],[40] Another limitation is that the outcomes in prescription database studies are based on medical records and coding that are not adjudicated and may be incorrect. Finally, all observational studies have the limitation that patients were not randomized to treatment, which leads to confounding. Even though attempts were made to minimize confounding through matching and adjustments, there may still be residual confounding due to unmeasured covariates.

The strengths of the prescription database studies were that they were able to examine a large number of T-treated men for several years for risk of CV events and mortality and whether there were differential risks related to age, T formulation, or T treatment timing. In general, most studies found no association of T risks with age or T formulations, although one large study found that IM T had increased risks compared to T gel. In contrast, all studies that examined risks with T initiation detected an increased risk for MI, thrombosis, composite CV events, and mortality with initiation.[30],[31],[37] Clinicians should be aware that there may be increased risks for CV events with T initiation, particularly within the first 3-6 months of T treatment.


  Conclusions Top


In large database studies that included over 215 000 T-treated men, most studies did not find that T treatment was associated with increased risks for mortality or for MI, stroke, or thrombosis. However, given the limitations of observational data, the cardiovascular risks of T treatment will remain unclear until a large, prospective, randomized, double-blind, placebo-controlled trial is conducted that is specifically designed to examine the risk for CV events with T treatment. Until then, clinicians should follow established clinical guidelines[41] for T replacement and review potential risks of treatment with patients, recognizing that risks might be greatest in the first several months after initiation of testosterone therapy.


  Author Contributions Top


The author was solely responsible for researching, writing, and creating the tables for this paper. The author read and approved the final manuscript.


  Competing Interests Top


The author declares no competing interests.


  Acknowledgments Top


This work was supported by NIH grant (No. 5R01AG042934-02).

Supplementary information is linked to the online version of the paper on the Asian Journal of Andrology website.

 
  References Top

1.
Shabsigh R, Arver S, Channer KS, Eardley I, Fabbri A, et al. The triad of erectile dysfunction, hypogonadism and the metabolic syndrome. Int J Clin Pract 2008; 62: 791-8.  Back to cited text no. 1
    
2.
Yeap BB, Chubb SA, Hyde Z, Jamrozik K, Hankey GJ, et al. Lower serum testosterone is independently associated with insulin resistance in non-diabetic older men: the Health In Men Study. Eur J Endocrinol 2009; 161: 591-8.  Back to cited text no. 2
    
3.
Isidori AM, Giannetta E, Pozza C, Bonifacio V, Isidori A. Androgens, cardiovascular disease and osteoporosis. J Endocrinol Invest 2005; 28: 73-9.  Back to cited text no. 3
    
4.
Shores MM, Matsumoto AM, Sloan KL, Kivlahan DR. Low serum testosterone and mortality in male veterans. Arch Intern Med 2006; 166: 1660-5.  Back to cited text no. 4
    
5.
Laughlin GA, Barrett-Connor E, Bergstrom J. Low serum testosterone and mortality in older men. J Clin Endocrinol Metab 2008; 93: 68-75.  Back to cited text no. 5
    
6.
Corona G, Monami M, Boddi V, Cameron-Smith M, Fisher AD, et al. Low testosterone is associated with an increased risk of MACE lethality in subjects with erectile dysfunction. J Sex Med 2010; 7: 1557-64.  Back to cited text no. 6
    
7.
Malkin CJ, Pugh PJ, Morris PD, Asif S, Jones TH, et al. Low serum testosterone and increased mortality in men with coronary heart disease. Heart 2010; 96: 1821-5.  Back to cited text no. 7
    
8.
Hyde Z, Norman PE, Flicker L, Hankey GJ, Almeida OP, et al. Low free testosterone predicts mortality from cardiovascular disease but not other causes: the Health in Men Study. J Clin Endocrinol Metab 2012; 97: 179-89.  Back to cited text no. 8
    
9.
Khaw KT, Dowsett M, Folkerd E, Bingham S, Wareham N, et al. Endogenous testosterone and mortality due to all causes, cardiovascular disease, and cancer in men: European prospective investigation into cancer in Norfolk (EPIC-Norfolk) Prospective Population Study. Circulation 2007; 116: 2694-701.  Back to cited text no. 9
    
10.
Handelsman DJ. Global trends in testosterone prescribing, 2000-2011: expanding the spectrum of prescription drug misuse. Med J Aust 2013; 199: 548-51.  Back to cited text no. 10
    
11.
Baillargeon J, Urban RJ, Kuo YF, Holmes HM, Raji MA, et al. Screening and monitoring in men prescribed testosterone therapy in the U.S., 2001-2010. Public Health Rep 2015; 130: 143-52.  Back to cited text no. 11
    
12.
Finkle WD, Greenland S, Ridgeway GK, Adams JL, Frasco MA, et al. Increased risk of non-fatal myocardial infarction following testosterone therapy prescription in men. PLoS One 2014; 9: e85805.  Back to cited text no. 12
    
13.
Vigen R, O′Donnell CI, Baron AE, Grunwald GK, Maddox TM, et al. Association of testosterone therapy with mortality, myocardial infarction, and stroke in men with low testosterone levels. JAMA 2013; 310: 1829-36.  Back to cited text no. 13
    
14.
Xu L, Freeman G, Cowling BJ, Schooling CM. Testosterone therapy and cardiovascular events among men: a systematic review and meta-analysis of placebo-controlled randomized trials. BMC Med 2013; 11: 108.  Back to cited text no. 14
    
15.
Seftel AD. Re: testosterone products: drug safety communication - FDA cautions about using testosterone products for low testosterone due to aging; requires labeling change to inform of possible increased risk of heart attack and stroke. J Urol 2015; 194: 759-60.  Back to cited text no. 15
    
16.
Ramasamy R, Scovell J, Mederos M, Ren R, Jain L, et al. Association between testosterone supplementation therapy and thrombotic events in elderly men. Urology 2015; 86: 283-5.  Back to cited text no. 16
    
17.
Eisenberg ML, Li S, Herder D, Lamb DJ, Lipshultz LI. Testosterone therapy and mortality risk. Int J Impot Res 2015; 27: 46-8.   Back to cited text no. 17
    
18.
Muraleedharan V, Marsh H, Kapoor D, Channer KS, Jones TH. Testosterone deficiency is associated with increased risk of mortality and testosterone replacement improves survival in men with type 2 diabetes. Eur J Endocrinol 2013; 169: 725-33.  Back to cited text no. 18
    
19.
Shores MM, Smith NL, Forsberg CW, Anawalt BD, Matsumoto AM. Testosterone treatment and mortality in men with low testosterone levels. J Clin Endocrinol Metab 2012; 97: 2050-8.  Back to cited text no. 19
    
20.
Maggio M, Lauretani F, Ceda GP, Bandinelli S, Ling SM, et al. Relationship between low levels of anabolic hormones and 6-year mortality in older men: the aging in the Chianti Area (InCHIANTI) study. Arch Intern Med 2007; 167: 2249-54.  Back to cited text no. 20
    
21.
Hall SA, Ranganathan G, Tinsley LJ, Lund JL, Kupelian V, et al. Population-based patterns of prescription androgen use, 1976-2008. Pharmacoepidemiol Drug Saf 2014; 23: 498-506.  Back to cited text no. 21
    
22.
Morgentaler A, Lunenfeld B. Testosterone and cardiovascular risk: world′s experts take unprecedented action to correct misinformation. Aging Male 2014; 17: 63-5.  Back to cited text no. 22
    
23.
Kloner RA. Cardiovascular effects of the 3 phosphodiesterase-5 inhibitors approved for the treatment of erectile dysfunction. Circulation 2004; 110: 3149-55.  Back to cited text no. 23
    
24.
Baillargeon J, Urban RJ, Kuo YF, Ottenbacher KJ, Raji MA, et al. Risk of myocardial infarction in older men receiving testosterone therapy. Ann Pharmacother 2014; 48: 1138-44.  Back to cited text no. 24
    
25.
Wells GA, Shea B, O′Donnell D, Peterson J, Welch V, et al. The Newcastle-Ottawa Scale (NOS) for Assessing the Quality of Nonrandomized Studies in Meta-analyses. Ottawa, ON: Ottawa Hospital Research Institute; 2011. Available from: http://www.ohri.ca/programs/clinical_epidemiology/oxford.asp. [Last accessed on 2017 Mar 24].  Back to cited text no. 25
    
26.
Higgins JP, Green S. Tools for assessing methodological quality or risk-of-bias in nonrandomized studies. In: Cochrane Handbook for Systematic Reviews of Interventions. Ver. 5.1.0. London: The Cochrane Collaboration; 2011. p. 412-9.  Back to cited text no. 26
    
27.
Sharma R, Oni OA, Gupta K, Chen G, Sharma M, et al. Normalization of testosterone level is associated with reduced incidence of myocardial infarction and mortality in men. Eur Heart J 2015; 36: 2706-15.  Back to cited text no. 27
    
28.
Layton JB, Meier CR, Sharpless JL, Sturmer T, Jick SS, et al. Comparative safety of testosterone dosage forms. JAMA Intern Med 2015; 175: 1187-96.  Back to cited text no. 28
    
29.
Tan RS, Cook KR, Reilly WG. Myocardial infarction and stroke risk in young healthy men treated with injectable testosterone. Int J Endocrinol 2015; 2015: 970750.  Back to cited text no. 29
    
30.
Etminan M, Skeldon SC, Goldenberg SL, Carleton B, Brophy JM. Testosterone therapy and risk of myocardial infarction: a pharmacoepidemiologic study. Pharmacotherapy 2015; 35: 72-8.  Back to cited text no. 30
    
31.
Wallis CJ, Lo K, Lee Y, Krakowsky Y, Garbens A, et al. Survival and cardiovascular events in men treated with testosterone replacement therapy: an intention-to-treat observational cohort study. Lancet Diabetes Endocrinol 2016; 4: 498-506.  Back to cited text no. 31
    
32.
Anderson JL, May HT, Lappe DL, Bair T, Le V, et al. Impact of testosterone replacement therapy on myocardial infarction, stroke, and death in men with low testosterone concentrations in an integrated health care system. Am J Cardiol 2016; 117: 794-9.  Back to cited text no. 32
    
33.
Baillargeon J, Urban RJ, Morgentaler A, Glueck CJ, Baillargeon G, et al. Risk of venous thromboembolism in men receiving testosterone therapy. Mayo Clin Proc 2015; 90: 1038-45.  Back to cited text no. 33
    
34.
Glueck CJ, Goldenberg N, Budhani S, Lotner D, Abuchaibe C, et al. Thrombotic events after starting exogenous testosterone in men with previously undiagnosed familial thrombophilia. Transl Res 2011; 158: 225-34.  Back to cited text no. 34
    
35.
Glueck CJ, Wang P. Testosterone therapy, thrombosis, thrombophilia, cardiovascular events. Metabolism 2014; 63: 989-94.  Back to cited text no. 35
    
36.
Sharma R, Oni OA, Chen G, Sharma M, Dawn B, et al. Association between testosterone replacement therapy and the incidence of DVT and pulmonary embolism: a retrospective cohort study of the veterans administration database. Chest 2016; 150: 563-71.  Back to cited text no. 36
    
37.
Martinez C, Suissa S, Rietbrock S, Katholing A, Freedman B, et al. Testosterone treatment and risk of venous thromboembolism: population based case-control study. BMJ 2016; 355: i5968.  Back to cited text no. 37
    
38.
Li H, Benoit K, Wang W, Motsko S. Association between use of exogenous testosterone therapy and risk of venous thrombotic events among exogenous testosterone treated and untreated men with hypogonadism. J Urol 2016; 195: 1065-72.  Back to cited text no. 38
    
39.
Corona G, Rastrelli G, Vignozzi L, Mannucci E, Maggi M. Testosterone, cardiovascular disease and the metabolic syndrome. Best Pract Res Clin Endocrinol Metab 2011; 25: 337-53.  Back to cited text no. 39
    
40.
Yeap BB, Hyde Z, Almeida OP, Norman PE, Chubb SA, et al. Lower testosterone levels predict incident stroke and transient ischemic attack in older men. J Clin Endocrinol Metab 2009; 94: 2353-9.  Back to cited text no. 40
    
41.
Bhasin S, Cunningham GR, Hayes FJ, Matsumoto AM, Snyder PJ, et al. Testosterone therapy in men with androgen deficiency syndromes: an Endocrine Society clinical practice guideline. J Clin Endocrinol Metab 2010; 95: 2536-59.  Back to cited text no. 41
    



 
 
    Tables

  [Table 1]



 

 
Top
 
 
  Search
 
 Search Pubmed for
 
    -  Shores MM
Access Statistics
Email Alert *
Add to My List *
* Registration required (free)  

 
  In this article
Abstract
Introduction
Materials and Me...
Results
Summary
Conclusions
Author Contributions
Competing Interests
Acknowledgments
References
Article Tables

 Article Access Statistics
    Viewed209    
    PDF Downloaded16    

Recommend this journal