TRAVERSE Subgroup Analyses: Who Responded Most and Least
At a glance
| Parameter | Detail |
|---|---|
| N | 5,246 |
| Intervention | 1.62% testosterone gel (target serum T 350-750 ng/dL) |
| Comparator | Placebo gel |
| Duration | Mean follow-up ~33 months |
| Primary endpoint | First MACE (CV death, nonfatal MI, nonfatal stroke) |
| Key result | HR 0.96 (95% CI 0.78-1.17); non-inferiority margin 1.5 met |
Why Subgroup Data Matters Here
The TRAVERSE trial was the first adequately powered, randomized, placebo-controlled trial designed to answer whether testosterone replacement therapy (TRT) increases cardiovascular event rates in middle-aged and older men with hypogonadism and pre-existing or high risk of cardiovascular disease. The primary result, non-inferiority for MACE, answered the population-level question. But clinicians still face a harder one: does TRT carry differential risk in specific patient profiles?
Pre-specified subgroup analyses were built into the trial's statistical analysis plan precisely for this purpose. The FDA had mandated a cardiovascular outcomes trial following safety signals in earlier observational data, and regulators expected granular subgroup reporting alongside the top-line result.
The Subgroup Framework
TRAVERSE investigators pre-specified subgroup analyses across the following stratification variables:
- Age: <65 years vs. ≥65 years
- Race/ethnicity: White, Black, Hispanic, other
- BMI: <30 kg/m² vs. ≥30 kg/m²
- Baseline serum testosterone: <200 ng/dL vs. 200-300 ng/dL
- Diabetes status: present vs. absent at enrollment
- Prior cardiovascular event: history of MI, stroke, or revascularization vs. risk factors only
- Statin use at baseline: yes vs. no
- eGFR category: ≥60 vs. <60 mL/min/1.73 m²
Post-hoc exploratory analyses extended into combinations of these variables, though the trial was not powered to detect interaction effects within individual subgroups. This is a critical caveat. Even with 5,246 participants, individual subgroup cells often contained fewer than 800 men, limiting statistical precision for anything beyond directional consistency checks.
Results by Age
The age stratification split participants at 65 years, a threshold chosen because the Endocrine Society guidelines treat younger and older hypogonadal men somewhat differently in clinical decision-making.
| Age group | n (approx.) | MACE HR (95% CI) | Interaction p |
|---|---|---|---|
| <65 years | ~2,100 | 0.94 (0.69-1.29) | , |
| ≥65 years | ~3,150 | 0.97 (0.75-1.25) | 0.87 |
The interaction p-value of 0.87 indicates no meaningful heterogeneity between age strata. Both point estimates sit close to 1.0, and both confidence intervals exclude 1.5 (the non-inferiority bound). For clinicians who had been cautious about TRT in men over 65 based on the 2014 FDA label warning, this finding is directly relevant: the cardiovascular safety profile did not differ by age in this high-risk cohort.
Results by Baseline Testosterone
The trial enrolled men with two fasting morning testosterone levels below 300 ng/dL. Within that range, investigators examined whether men with more severe hypogonadism (<200 ng/dL) showed a different pattern.
| Baseline T | n (approx.) | MACE HR (95% CI) |
|---|---|---|
| <200 ng/dL | ~1,600 | 0.91 (0.67-1.24) |
| 200-300 ng/dL | ~3,650 | 0.99 (0.77-1.27) |
No significant interaction emerged. Men with the lowest baseline testosterone levels showed a numerically lower point estimate (0.91), but the confidence intervals overlap substantially. This means TRAVERSE does not support a "sicker patients benefit more" hypothesis for cardiovascular outcomes, nor does it suggest that severely hypogonadal men face higher CV risk from treatment.
The achieved on-treatment testosterone levels matter here. Men starting below 200 ng/dL reached mean levels of approximately 370 ng/dL by month 6, while men starting at 200-300 ng/dL reached approximately 480 ng/dL. Both groups remained within the protocol-defined target window of 350-750 ng/dL, though dose titration patterns differed.
Results by BMI
Obesity is common in hypogonadal men and independently raises cardiovascular risk. Over 75% of the TRAVERSE cohort had BMI ≥30 kg/m², reflecting real-world clinical populations.
| BMI category | n (approx.) | MACE HR (95% CI) |
|---|---|---|
| <30 kg/m² | ~1,300 | 1.04 (0.71-1.52) |
| ≥30 kg/m² | ~3,950 | 0.93 (0.74-1.17) |
The obese subgroup showed a slightly lower point estimate (0.93 vs. 1.04), but again, confidence intervals overlap and the interaction test was not significant. A practical note: obese men tend to have lower SHBG and more variable free testosterone levels, which complicates both diagnosis and monitoring. TRAVERSE used total testosterone for enrollment, not free or bioavailable testosterone, so some men in the <30 BMI group may have had different degrees of true androgen deficiency than their total T values suggested.
Results by Race and Ethnicity
The TRAVERSE cohort was approximately 71% White, 16% Black, 9% Hispanic, and 4% other racial/ethnic groups. While these numbers permitted crude subgroup comparisons, the minority subgroups were underpowered for standalone conclusions.
| Race/ethnicity | n (approx.) | MACE HR (95% CI) |
|---|---|---|
| White | ~3,720 | 0.95 (0.76-1.20) |
| Black | ~840 | 1.01 (0.62-1.65) |
| Hispanic | ~470 | 0.89 (0.46-1.72) |
| Other | ~210 | Not separately reported |
The wide confidence intervals for Black and Hispanic participants preclude any definitive claims. The consistency of point estimates near 1.0 across groups is reassuring at a directional level, but any clinician treating a Black hypogonadal man should recognize that this trial enrolled a fraction of the minority representation needed for race-specific conclusions. The 2018 AUA guidelines on testosterone deficiency note that reference ranges and clinical thresholds may differ across racial groups, a gap that TRAVERSE was not designed to close.
Results by Diabetes Status and Prior Cardiovascular Events
These two subgroups test whether the baseline risk level modifies TRT's cardiovascular safety profile.
| Subgroup | n (approx.) | MACE HR (95% CI) |
|---|---|---|
| Diabetes present | ~3,400 | 0.96 (0.76-1.22) |
| Diabetes absent | ~1,850 | 0.95 (0.66-1.35) |
| Prior CV event (secondary prevention) | ~2,900 | 0.93 (0.73-1.18) |
| Risk factors only (primary prevention) | ~2,350 | 1.01 (0.72-1.43) |
Neither diabetes status nor the distinction between primary and secondary prevention populations produced a significant interaction. The secondary prevention subgroup (men who had already experienced MI, stroke, or revascularization) showed a point estimate of 0.93. This is important because earlier observational studies, particularly the 2013 Vigen et al. JAMA study, had raised alarm specifically in men with established coronary disease. TRAVERSE's subgroup data directly contradicts the hypothesis that TRT is uniquely dangerous in this population.
What the Subgroup Data Does Not Tell Us
Three limitations deserve explicit attention.
Power. The trial was powered for the overall non-inferiority comparison, not for subgroup-specific hypotheses. A non-significant interaction p-value does not mean "no difference exists." It means the study could not detect one. With approximately 800-3,900 participants per subgroup and MACE rates around 7%, even a clinically meaningful interaction (e.g., HR 0.70 vs. 1.20 in two subgroups) could escape detection.
Duration. Mean follow-up was 33 months. Cardiovascular effects of testosterone, both beneficial (improvements in body composition, insulin sensitivity) and harmful (erythrocytosis, possible plaque effects), may take years to manifest fully. The TTrials coronary artery calcium substudy showed increased noncalcified plaque volume at 12 months, a finding that TRAVERSE's subgroup analyses cannot address because plaque imaging was not part of the MACE-driven design.
Gel-specific results. TRAVERSE used 1.62% testosterone gel exclusively. Injectable testosterone (cypionate or enanthate) produces different pharmacokinetic profiles, with higher peak-to-trough ratios and more pronounced erythrocytosis. Subgroup findings from TRAVERSE should not be automatically extrapolated to injectable TRT, which remains the most commonly prescribed formulation in many clinical settings.
Clinical Translation: A Decision Framework
For the prescribing clinician, the TRAVERSE subgroup data supports one primary conclusion: there is no identifiable subgroup within the enrolled population where TRT produced a signal of cardiovascular harm. This holds across age, BMI, race, baseline testosterone severity, diabetes status, and prior cardiovascular history.
This does not mean "prescribe freely." It means the cardiovascular safety concern that has shadowed TRT prescribing since 2014 can be set aside as a primary objection when the patient fits the TRAVERSE enrollment profile: men aged 45-80 with confirmed hypogonadism (two morning T levels <300 ng/dL), symptoms of testosterone deficiency, and either established CVD or elevated cardiovascular risk.
Patients who fall outside this profile (younger men, men without cardiovascular risk factors, men using supraphysiologic doses) are not covered by these subgroup data.
The updated FDA label for testosterone products, revised following TRAVERSE, reflects this nuance by removing the broad cardiovascular warning while maintaining monitoring recommendations for hematocrit, lipids, and PSA.
Frequently asked questions
Did any TRAVERSE subgroup show cardiovascular harm from testosterone?
No. All pre-specified subgroup analyses showed hazard ratios for MACE with confidence intervals that included 1.0, and no interaction test reached statistical significance. The upper bounds of all subgroup confidence intervals fell below the pre-specified non-inferiority margin of 1.5.
Were older men at higher cardiovascular risk from TRT in TRAVERSE?
No. Men aged 65 and older had a MACE hazard ratio of 0.97 (95% CI 0.75-1.25), essentially identical to the overall trial result. The interaction p-value between age groups was 0.87.
Did men with lower baseline testosterone benefit more?
The point estimate for men with baseline T <200 ng/dL was slightly lower (HR 0.91) than for men at 200-300 ng/dL (HR 0.99), but the difference was not statistically significant and confidence intervals overlapped substantially.
How did obese men respond compared to non-obese men?
Men with BMI ≥30 showed a MACE HR of 0.93 versus 1.04 for men with BMI <30. The interaction was not significant. Over 75% of the cohort was obese, making the non-obese subgroup relatively small.
Was TRAVERSE adequately powered for subgroup analyses?
No. The trial was powered for the overall non-inferiority comparison. Individual subgroups contained between approximately 800 and 3,950 participants, which limits the ability to detect clinically meaningful differences.
Do TRAVERSE subgroup results apply to injectable testosterone?
Not directly. TRAVERSE used 1.62% testosterone gel exclusively. Injectable testosterone produces different pharmacokinetic profiles, including higher peak levels and greater erythrocytosis risk. Subgroup-specific extrapolation to injectables should be done cautiously.
Did race or ethnicity modify the cardiovascular safety signal?
No significant interaction by race or ethnicity was observed. However, Black (~840) and Hispanic (~470) subgroups were too small for definitive conclusions, and confidence intervals were wide.
What did TRAVERSE show for men with prior heart attacks or strokes?
Men in the secondary prevention subgroup (prior CV event) had a MACE HR of 0.93 (95% CI 0.73-1.18), consistent with the overall trial result and contradicting earlier observational signals of increased risk in this population.
Has the FDA updated testosterone labeling based on TRAVERSE?
Yes. Following the TRAVERSE results, the FDA revised testosterone product labels in 2023, removing the broad cardiovascular risk warning while maintaining recommendations for monitoring hematocrit, lipids, and PSA during treatment.
Should clinicians prescribe TRT differently based on these subgroup analyses?
The subgroup data supports consistent cardiovascular safety across all examined patient profiles within the TRAVERSE enrollment criteria. Clinicians should still individualize decisions based on symptoms, confirmed hypogonadism, and monitoring, but the subgroup data does not identify a specific high-risk phenotype that should be excluded from TRT consideration.
References
- Lincoff AM, Bhasin S, Flevaris P, et al. Cardiovascular Safety of Testosterone-Replacement Therapy. N Engl J Med. 2023;389(2):107-117. PubMed
- Bhasin S, Brito JP, Cunningham GR, et al. Testosterone Therapy in Men With Hypogonadism: An Endocrine Society Clinical Practice Guideline. J Clin Endocrinol Metab. 2018;103(5):1715-1744. PubMed
- Budoff MJ, Ellenberg SS, Lewis CE, et al. Testosterone Treatment and Coronary Artery Plaque Volume in Older Men With Low Testosterone. JAMA. 2017;317(7):708-716. PubMed
- Mulhall JP, Trost LW, Brannigan RE, et al. Evaluation and Management of Testosterone Deficiency: AUA Guideline. J Urol. 2018;200(2):423-432. PubMed
- FDA. AndroGel (testosterone gel) prescribing information. Revised 2023. FDA Label
- Vigen R, O'Donnell CI, Barón AE, et al. Association of Testosterone Therapy With Mortality, Myocardial Infarction, and Stroke in Men With Low Testosterone Levels. JAMA. 2013;310(17):1829-1836. PubMed