Long Term Cardiovascular Safety of Testosterone Therapy: A Review of the TRAVERSE Study

Abstract

TRAVERSE (TheRapy for Assessment of long-term Vascular events and Efficacy ResponSE in hypogonadal men) is multicentre randomized, double-blind, placebo-controlled, noninferiority trial of testosterone therapy, enrolling 5,246 men 45 to 80 years of age who had pre-existing or a high risk of cardiovascular disease and who reported symptoms of hypogonadism. Subjects required two fasting testosterone levels of less than 10.4 nmol/L. Patients were randomly assigned to receive daily transdermal 1.62% testosterone gel (dose adjusted to maintain testosterone levels between 12 nmol/L and 26 nmol/L) or placebo gel for a mean 27.1 months. The primary cardiovascular safety end point was the first occurrence of any component of a composite of death from cardiovascular causes, non-fatal myocardial infarction, or non-fatal stroke, assessed in a time-to-event analysis. TRAVERSE found no increase in major adverse cardiac events or prostate related events, including prostate cancer, effectively addressing the concerns raised by the United States Food and Drug Administration.

INTRODUCTION

TRAVERSE [1] is multicentre randomized, double-blind, placebo-controlled, noninferiority trial of testosterone therapy, enrolling 5,246 men 45 to 80 years of age who had pre-existing or a high risk of cardiovascular disease and who reported symptoms of hypogonadism. Subjects required two fasting testosterone levels of less than 10.4 nmol/L. Patients were randomly assigned to receive daily transdermal 1.62% testosterone gel (dose adjusted to maintain testosterone levels between 12 nmol/L and 26 nmol/L) or placebo gel for a mean 27.1 months. The study was commissioned in the US at the request of the FDA [2], following the publication of the testosterone in older men [3], and Vigen [4] studies in the US, which led to a black box warning on all testosterone products. In contrast the European medicines agency [5], reviewed the same evidence base and found no cause for concern. The primary cardiovascular safety end point was the first occurrence of any component of a composite of death from cardiovascular causes, non-fatal myocardial infarction, or non-fatal stroke, assessed in a time-to-event analysis. TRAVERSE found no increase in major adverse cardiac events [1] or prostate related events [6]. Further papers from TRAVERSE addressing bone [7], anaemia [8], depression [9], somatic symptoms and sexual function [10] have been published, with some confusing conclusions. The key findings from each paper are discussed below.

PRIMARY ENDPOINTS

1. Major adverse coronary events (MACE)

In men with hypogonadism and pre-existing or a high risk of cardiovascular disease, testosterone replacement therapy (TRT) was noninferior to placebo with respect to the incidence of major adverse cardiac events (the Primary Study Endpoint) (Fig. 1, 2).

Fig. 1. TRAVERSE study. On-treatment sensitivityanalyses, primary MACE safety endpoints. MACE: major adverse coronary events, HR: hazard ratio, CI: confidence interval.

Fig. 2. Primary cardiovascular (CV) safety endpoint. HR: hazard ratio, CI: confidence interval, MACE: major adverse cardiovascular events, MI: myocardial infarction.

There was a small non-significant reduction in mortality with 16 fewer deaths in the testosterone group, which needs to be taken into account when some of the reported issues are discussed later in this paper (Fig. 3).

Fig. 3. TRAVERSE Trial: results. AE: adverse event, CVD: cardiovascular disease, HR: hazard ratio, SAE: serious adverse event.

2. Prostate related events

Testosterone therapy was associated with low and similar incidences of high grade or any prostate cancer, acute urinary retention, and invasive surgical procedures for benign prostatic hyperplasia (BPH) compared with a placebo. There were 5 high grade prostate cancers in the TRT group vs. 3 in the placebo group.

TRT did not worsen lower urinary tract symptoms (LUTS). There was no increase in prostate cancer or prostate related events which strongly supports the view that routine digital rectal examination (DRE) is not required unless clinically indicated for other reasons in line with a recent meta-analysis [11]. This conclusion is clinically important as recent guidelines from the European Association of Urology (EAU) recommended DRE, before TRT, within 3 to 6 months and after 12 months, potentially 3 in the first year [12]. In contrast, the Society for Endocrinology [13] stated that DRE was not necessary, as “endocrinologists are not expert” and that it “may be potentially harmful.” Clearly these divergent views were a potential medico-legal dilemma as cases of prostate cancer were bound to occur coincidentally and it would be a logistic nightmare for patients to be routine managed by specialists with conflicting views.

Recent studies [14] have suggested that patients’ dislike of DRE is a potential barrier to seeking help. The findings of TRAVERSE should help to clarify that screening for prostate cancer is a completely different clinical issue and that, in reality, men on TRT are likely to be at an advantage by seeing physicians regularly and being able to discuss a range of men’s health issues, including prostate-specific antigen testing.

SECONDARY ENDPOINTS

1. New onset diabetes

TRAVERSE showed a reduction in diabetes progression but the authors appear to have made a misinterpretation of their data on incident diabetes, in the NEJM paper [1] they report that 189 men on testosterone therapy (TTh) and 213 on placebo developed incident diabetes which they list as 7.3% and 8.2% and non-significant. Fig. 3 also points out that circa 70% in both cohorts had diabetes at baseline. As clearly patients with baseline diabetes cannot develop incident diabetes, these percentages need to be based on the circa 30% (1,500) without baseline diabetes. Checking the figures closely, we see that 607 men on testosterone vs. 568 on placebo had prediabetes at baseline and it is highly likely that these were the ones who progressed to incident diabetes. It is therefore 189/607 (31%) on TTh vs. 213/558 (40%) on placebo, which equates to a 22.5% reduction (p=0.029) in new onset diabetes. These findings are in line with earlier studies suggesting that 5% to 10% per annum progress from pre- to overt diabetes.

The authors conceded that 25% of TTh patients failed to reach the normal range and the mean increase was only 9.3 to 12.9 nmol/L, suggesting that patients were undertreated. In the T trials the median increase with testosterone gel was from 7.63 to 17.35 and in the testosterone for diabetes mellitus (T4DM) trial 13.4 to 16.8 nmol/L suggesting that treatment goals were achieved in the 2 earlier studies (Fig. 4). Despite under-treatment, the older TRAVERSE patients with greater comorbidities achieved a 22.5% reduction in diabetes progression.

Fig. 4. Mean baseline and post-treatment testosterone levels (nmol/L) in pivotal testosterone studies. T4DM: testosterone for diabetes mellitus, TU: testosterone undecanoate.

The authors presented data at The Endocrine society meeting 2024 with a p-value of 0.027 (Fig. 5) which as altered to 0.06 in the diabetes paper, presumably with the latter taken from the 4 year end point of HbA1c (when compliance rates would have been at the lowest), not the protocol defined end point of “newly diagnosed diabetes.” This is important as the anaemia arm of the study showed a clear increase in hematocrit known to impact diagnosis based on HbA1c [15].

Fig. 5. TRAVERSE study. Effects of testosterone replacement therapy (TRT) vs. placebo on hemoglobin A1c (HbA1c) in men with prediabetes (A) and diabetes at baseline (B).

Likewise, the graphs based on fasting blood glucose control over 4 years for both baseline diabetes and prediabetes are misleading as the study, although recording baseline medications, did not record changes in medications, or lifestyle intervention over 4 years. The prescribing of medication for prediabetes is listed as zero in both cohorts at baseline but it is difficult to believe that patients subjected to “normal primary care” did not receive any medication, especially as there has been enthusiasm for using metformin in such cases. It is also unlikely that the 70% of men with diabetes at baseline did not have multiple changes in diabetes medication over 4 years. A positive effect of TRT might have resulted in reduction in diabetes medication (or need to increase medication with placebo) rather than changes in HbA1c or blood glucose. In particular the summary of product characteristics (SPCs) of testosterone products [16] warns “doses of diabetes medications may require adjustment when TRT is prescribed.” In fact, regulatory rules require reduction in diabetes medication (or clinical or biochemical hypoglycaemia) to be reported as “adverse events,” even though the decision to stop or reduce diabetes medication would, in normal circumstances, be considered clinically beneficial. Because no medication changes or lifestyle interchanges were recorded over the 4 year period, the conclusion:

“In middle-aged and older men with hypogonadism and prediabetes in the TRAVERSE Diabetes Study, the incidence of progression from prediabetes to diabetes did not differ significantly between testosterone- and placebo-treated men. TRT alone did not improve glycaemic control in men with hypogonadism and prediabetes or diabetes” is misleading.

In contrast the T4DM study [17] involved 1,007 men with prediabetes 79.5% and diabetes 20.5% and a baseline testosterone <14 nmol/L treated with 12 weekly testosterone decanoate (or placebo) achieved trough level increases from 13.4 to 16.8 nmol/L on active therapy (Fig. 4) and a significant 41% reduction in incident diabetes. As the post treatment levels in TRAVERSE failed to reach the pre-treatment levels in T4DM, it is highly likely that the older TRAVERSE patients with greater comorbidities were undertreated in terms of diabetes prevention. Although these points are clearly discussed in the original paper by Snyder et al [18], all subsequent papers from TRAVERSE do not discuss issues of under-treatment or compliance which are fundamentally important to the conclusions from those studies.

2. Sexual function

There was a significant increase in sexual desire and sexual frequency but not erectile function, which is not surprising in a population of mean age 65 years, 70% with diabetes plus multiple cardiovascular comorbidities. Erectile dysfunction (ED) in these patients was most likely related to endothelial dysfunction or diabetic neuropathy rather than hormonal factors. In common with other recent RCTs, only 4% were taking ED medications. In other studies, such as BLAST, conducted in a hypogonadal population with T2DM, rates of PDE5 inhibitor use was closer to 30%, and improvement in ED was significantly higher.

In the T trials [18] with lower rates of diabetes and better compliance, a significant improvement in erectile function score of 2.64 points was found versus placebo, Couples in the TRT group engaged in sexual activity 4 times per week described by authors as —modest response” with similar results in T4DM (2.11 points vs. placebo), where once again, very low rates of PDE5 inhibitor use was reported [16].

These findings support the 2023 BSSM Guidelines [19] recommendation for routine daily use of PDE5 inhibitors in hypogonadal men with ED. This raises the question as to why PDE5 inhibitor use was so low in men with diabetes in all these studies and suggests that future research might involve pre-treating men with a daily PDE5 inhibitor before conducting placebo controlled studies of TRT. This is especially relevant as recent observational studies have reported a significant dose related reduction in MACE and mortality with PDE5 inhibitors [20].

The high baseline IPSS scores in T4DM, suggestive of BPH/LUTS would also support the argument for greater use of daily PDE5 inhibitors, especially when one observational study has reported 40% reduction in MACE with LUTS treated with daily tadalafil versus alpha blocker [21].

It could be argued that both TRAVERSE and the T trial suggest that increasing sexual desire through TRT and not treating the severe ED present with diabetes might result in frustration for the couple and worsen the outcome.

3. Depression

Depression was found in 50% of men with hypogonadism and was mild to moderate (defined as a PHQ [patient health questionnaire] score >4) but 10% were severe. It must be remembered that these were men with significant cardiovascular risk and high rates of sexual dysfunction at baseline and 70% had type 2 diabetes. Modest improvement in depression and aging male symptom scores were found in men with mild to moderate cases. The significant improvements in non-specific symptoms such as mood and energy are more likely to result from improvement in hypogonadism than depression per se.

The authors suggested that screening for depression in hypogonadism and treating first with TRT is more likely to improve sexual symptoms in contrast to anti-depressants which frequently have adverse effects [9].

Surely the 50% association of depression and hypogonadism, also suggests that men presenting with depression (often to psychiatrists) should also have their testosterone checked, but the authors stop short of making this conclusion when surely in the majority of cases, these are the same patients just presenting by a different pathway. The authors also point out that the study was conducted at the time of the COVID crisis and that this affected compliance and retention, although no COVID related adverse events or deaths were reported in the 5,204 patients recruited.

Several observational studies have shown that regular dosing with PDE5 inhibitors, either alone or in conjunction with TRT improve depression and cognitive function [22] but TRAVERSE was not powered to detect this because only around 8% were taking any form of PDE5 inhibitor.

Men with major depression may need antidepressant therapy or referral for psychiatric support, but the majority of men with mild to moderate depressive symptoms will be treated by primary care physicians who need to be aware of the associated medical comorbidities in these patients. For these patients a modest improvement in depression combined with other benefits in terms of sexual function, diabetes and anaemia may equate to significant overall clinical benefit.

4. Anaemia

Testosterone therapy corrected anaemia (detected in 815 of 5,204 or 15.7%) and energy levels in a significantly greater proportion of testosterone-treated than placebo-treated men at 6 months (143 of 349 [41.0%] vs. 103 of 375 [27.5%]). Improvement in anaemia was associated with moderate but significant improvement in energy levels and minor improvement in cognitive function.

This is in line with the T4DM [16] and the T Trial [17] where anaemia was found in 15% to 20% and correction of anaemia occurred with TRT even in patients with an alternative clinical diagnosis and treatment. TRAVERSE was the first study to demonstrate that functional symptom improvement is closely related to correction of anaemia. It is worth noting that anaemia is an established cardiovascular risk factor [23].

It is important to consider that hypogonadal men rarely present with anaemia, but more often with sexual dysfunction and vague somatic symptoms that are more likely to be treated with PDE5 inhibitors or anti-depressants, which will miss this potential benefit from detecting and treating mild anaemia with TRT.

Anaemia is also known to increase HbA1c and correcting iron deficiency anaemia has been shown to lower HbA1c. It is recommended that anaemia is corrected before confirming a new diagnosis of diabetes [24]. A sub-analysis of TRAVERSE would be interesting to establish whether the 15.7% of patients with anaemia at baseline showed greater reduction in HbA1c with treatment as this might have important clinical indications, especially as the T trials showed TTh improved anaemia even in men with a previously diagnosed cause.

5. Hematocrit

Only six participants, whose hematocrit level exceeded 54% at the lowest (20.25 mg) testosterone dose, had their study medication discontinued in TRAVERSE. Time-dependent Cox hazards models did not show an association between change in hematocrit and the risk of MACE (hazard ratio [HR], 0.97; 95% confidence interval [95% CI], 0.92–1.02) or venous thrombotic embolism (HR, 0.94; 95% CI, 0.84–1.05). There was no association between hematocrit and risk of cardiovascular events. These findings are in contrast with the T4DM [18] which found hematocrit levels of over 54% in around 20% of patients on testosterone undecanoate, although only 8% withdrew because of a raised second level. This is probably related to the prolonged higher testosterone levels with the depot injection in T4DM [18]. The authors of T4DM, the T trial and TRAVERSE stress that these findings apply only to the protocols of these studies conducted within the SPC of these products and that the findings might not apply to short-acting self-administered injection regimes, especially where dosing is controlled by the patient and liable to be adjusted for enhanced physical performance.

6. Atrial fibrillation

A surprising finding was an increase in diagnosis of atrial fibrillation pointing to 91 (TTh) vs. 63 (placebo) cases in the 5,204 patients (p=0.02) and non-fatal arrythmias (p=0.001). This had not been reported in other RCTs. There were 16 fewer deaths in patients on TRT, meaning that paradoxically this could have resulted contributed to this small increase in non-fatal events. Several studies have suggested that sudden increase in physical activity, as was seen in the 6-minute walking test in the T trial, might increase the risk of AF [25] in the short term, while long-term exercise is considered beneficial. Sharma et al [26] reported rates of AF in 76,639 veterans over 6 years, divided into group 1 (normalized T), group 2 (non-normalized), and group 3 (low T untreated) and found the incidence of AF was as follows: group 1 (normalized TRT), 497 per 100,000 personyears; group 2 (non-normalized TRT), 626 per 100,000 person-years; group 3 (no TRT), 697 per 100,000 person-years (p=0.001; group 1 vs. 3). This study suggested a significant reduction in AF in patients treated to normal levels for a prolonged period.

This unusual finding of reported AF in TRAVERSE needs to be considered in the context of a secondary outcome measure. Electrocardiograms were not conducted at baseline nor as part of the trial protocol, just as reported medical events. Findings from the testosterone in older men (TOM) trial [3] suggested that such adverse events might be explained by early response leading to increase in mobility and walking capacity in a previous frail group of men with comorbidities.

7. Acute kidney injury

The diagnosis acute kidney injury (AKI) is based on acute rise in serum creatinine in urine output (oliguria) but 418 (TTh) vs. 393 (placebo) had stage 3 kidney disease at baseline (a difference of 25) and yet the authors listed 60 vs. 40 cases of AKI (a difference of 20) as being significant (p=0.04). Strangely 16 extra deaths from cardiovascular disease (classified by the authors as non-significant), occurred in the placebo group. As it was highly unlikely that the patients who died did so without a rise in creatinine and with normal urine flow, the event of death potentially underestimated to true rate of AKI in the placebo group. Sharma et al reported 38,706 men with chronic kidney disease (CKD) and low testosterone untreated and 9,755 treated to normal levels over 6 years and found that TTh delayed progression to end stage CKD by 284 days (p=0.0001) and delayed death by 328 days (p=0.05) [27].

8. Bone fracture

Paradoxically, the fracture incidence was numerically higher among men who received testosterone, 91 (3.5%) than among those who received placebo 64 (2.4%), an excess of 27 fractures in 5,204 patients. This was a surprising finding as fractures were largely fall related and occurred too early to be related to T therapy. The JAMA editorial on TRAVERSE [28] suggested a mechanism likely to be related to early increased activity with TRT, as shown in the prompt increase in the 6 minute walking time in the T trial, in an otherwise vulnerable group. There was no substantial between-group difference in the incidence of typical osteoporotic fractures of the spine, hip, humerus and wrist.

The issue of fall related fractures with early mobilization in older frail adults was clarified in a meta-analysis by Van Gameren et al [29], where they describe an early increase in fractures with increased activity but a long-term reduction with the development of increased muscular strength.

TRAVERSE did not include assessment of bone mineral density (BMD) pre and post treatment over 1 to 2 years. BMD was found to significantly improve with TRT in other studies, especially the T trials [7].

DISCUSSION

The authors of the T trials [18] concluded in 2021:

Testosterone treatment of older men with low testosterone levels improves overall sexual activity, sexual desire, and erectile function; improves areal and volumetric bone density, as well as estimated bone strength in the spine and the hip; corrects unexplained anaemia of aging; increases skeletal muscle mass, strength and power, self-reported mobility, and some measures of physical function; and modestly improves depressive symptoms.

They concluded that “The long-term effects of testosterone on major cardiovascular events and prostate cancer risk remain unclear.” It would appear that TRAVERSE has successfully addressed both these concerns, as the primary end point of cardiovascular safety associated with TTh was achieved and there were no additional cases of prostate cancer.

Hopefully, as this study was conducted at the request of the FDA, these positive findings will lead to the lifting of the warning on the SPC of testosterone products in the US. The lack of association of raised hematocrit with cardiovascular risk is also reassuring. This suggests patients are being denied potentially effective treatment because of concerns over modest hematocrit levels above 48%. TRAVERSE, the T trials and T4DM consisted of patients with functional (FH) rather than classical hypogonadism (CH). In reality, there are no well-powered studies in CH and yet endocrinologists often state that the benefits of treatment in CH are well established and that FH should be managed by addressing the associated pathologies such as obesity and type 2 diabetes, even though such patients had been receiving care in well-funded and resourced health care systems. Recently Zitzmann et al [30] evaluated 650 patients, predominantly CH, and concluded that greater benefits were seen in men with FH.

Likewise, the TRAVERSE data on prostate safety are very reassuring that treating appropriately diagnosed hypogonadism with TRT is not associated with increased risk of prostate cancer or BPH exacerbation and that such men might actually benefit from being under routine medical care of urologists and primary care physicians with greater opportunity to discuss male-related health care issues.

The authors of TRAVERSE are to be congratulated in effectively achieving the primary endpoints of the study, cardiovascular and prostate safety. They have made valiant efforts to address several important secondary endpoints in the same study. A major problem, however, was the way that the first paper [1] addressed the complex methodology. Essentially this was a study of “undertreated” hypogonadal men, with poor compliance in comparison with the T trials and T4DM, but readers of the secondary papers needed to refer back to the original paper to understand this point. The reality is that specialists are likely to read only the secondary papers related to their own expertise, namely psychiatrists for depression, hematologists for anaemia, urologists for ED, etc and miss the point of the importance of a series of moderate improvements across multiple specialisms because they are focused only on one problem on that particular day.

It does seem, however, that every positive testosterone study always comes with caveats and in this case, it is the strange report of a small increase in fracture rate, atrial fibrillation, and AKI where other studies have shown no such links. These issues might be related to populations with multiple comorbidities experiencing increased vitality and therefore increased physical activity before therapeutic benefits on the cardiovascular systems are seen [31,32]. This might justify caution in frail patients in the early stages of treatment but should not prevent them from achieving long-term benefits.

CONCLUSIONS

TRAVERSE is the largest study to date of TTh, and involved 5,204 men treated for 4 years at a cost of over $50 million. It is usual to end reviews such as this with the statement “more studies are required” but surely the time has now come to say that TRT to achieve normal ranges of testosterone is safe in terms of cardiovascular and prostate related events and effective if prescribed according to guidelines. This should help physicians to manage men with symptomatic testosterone deficiency with greater confidence.