Inside the TRAVERSE Bone Fracture Substudy Methodology: What Most Summaries Skip

At a glance

| Parameter | Detail | |-----------|--------| | N | 5,204 randomized (of 5 to 246 in bone substudy population) | | Intervention | 1.62% testosterone gel, dose-adjusted to target 350-750 ng/dL | | Comparator | Matching placebo gel | | Duration | Median 3.19 years follow-up | | Primary endpoint | First clinical fracture (confirmed by central adjudication) | | Key result | HR 1.43 (95% CI 1.04-1.97); p = 0.03 |

The Parent Trial Architecture

TRAVERSE (Testosterone Replacement Therapy for Assessment of Long-term Vascular Events and Efficacy Response in Hypogonadal Men) was an FDA-mandated cardiovascular safety trial enrolling 5,246 men aged 45-80 with hypogonadism and established or high-risk cardiovascular disease. The bone fracture analysis was pre-specified as a secondary endpoint in the original protocol, not a post-hoc add-on. This distinction matters because pre-specification protects against the multiplicity problems that plague exploratory subgroup analyses.

The trial was conducted across 316 sites in the United States, required by FDA's 2015 labeling update mandating post-market cardiovascular outcome data for testosterone products. Participants had two fasting morning testosterone levels <300 ng/dL and at least one symptom of hypogonadism.

Randomization and Blinding Design

Randomization was 1:1, stratified by site, using a central interactive web-response system. The testosterone gel (AndroGel 1.62%) and placebo were packaged identically in pump dispensers. Dose titration occurred at months 1, 2, 3, 6, and then every 6 months, targeting serum testosterone between 350 and 750 ng/dL.

A critical methodological note: the dose-adjustment algorithm was managed by an unblinded pharmacist at each site, while all investigators, participants, and outcome assessors remained blinded. This firewall preserved integrity while allowing the clinically necessary dose titration that a fixed-dose design would not support. The primary TRAVERSE publication confirms that mean achieved testosterone was 358 ng/dL in the treatment arm versus 210 ng/dL in placebo, indicating modest physiologic replacement rather than supraphysiologic dosing.

Endpoint Definition and Adjudication

The primary bone endpoint was time to first clinical fracture, defined as any fracture confirmed by radiographic imaging or surgical report, excluding pathologic fractures from malignancy. Vertebral fractures required clinical presentation (pain or height loss prompting imaging) rather than morphometric screening, meaning asymptomatic compression fractures were not captured.

An independent, blinded Clinical Events Committee adjudicated every fracture event using source radiology reports. This central adjudication process reduces site-level reporting bias that might otherwise inflate or suppress fracture counts differentially between arms. The published substudy results report that 91 fractures occurred in the testosterone group versus 64 in placebo, yielding the hazard ratio of 1.43.

Inclusion and Exclusion Criteria: Who Was Actually Studied

The men enrolled were not representative of all TRT candidates. Key inclusion criteria:

• Age 45-80
• Two morning testosterone levels <300 ng/dL
• Pre-existing cardiovascular disease OR high cardiovascular risk (≥3 risk factors)
• BMI ≥ 19

Key exclusions relevant to bone interpretation:

• Current bisphosphonate or denosumab use
• Known osteoporosis with T-score < -3.5
• Recent (within 6 months) fragility fracture
• Use of systemic glucocorticoids exceeding prednisone 5 mg/day

These exclusion criteria mean the trial population was neither the healthiest nor the most osteoporotic. Men with severe bone disease were excluded, and men already on bone-protective therapy were removed. The result applies to a specific window: older men with moderate hypogonadism, cardiovascular risk, and unprotected bones. The Endocrine Society's 2018 guideline on testosterone therapy in men with hypogonadism had previously cited bone density improvement as a potential benefit of TRT, making this fracture signal particularly unexpected.

The Statistical Approach

The substudy used a Cox proportional hazards model with treatment assignment as the sole covariate, following intention-to-treat principles. The proportional hazards assumption was tested and met. Time zero was randomization; censoring occurred at death, withdrawal, or administrative end of study.

No adjustment for multiple comparisons was applied to the bone fracture endpoint because it was pre-specified as a key secondary endpoint with its own hypothesis. The study protocol registered on ClinicalTrials.gov confirms this pre-specification.

The confidence interval (1.04-1.97) barely excludes 1.0. The p-value of 0.03 crosses the conventional threshold but would not survive a Bonferroni correction if applied across all TRAVERSE secondary endpoints. The authors appropriately present this as a signal requiring confirmation rather than definitive proof.

The Estimand Framework

TRAVERSE used a treatment-policy estimand: the effect of being assigned to testosterone regardless of adherence, dose changes, or discontinuation. Approximately 15% of participants discontinued study drug before trial completion. This conservative approach dilutes the treatment effect estimate, meaning the true on-treatment fracture risk could be higher than the observed 43%.

An alternative per-protocol analysis restricted to adherent participants was not presented in the primary publication, which limits understanding of dose-response. Whether men who achieved higher testosterone levels experienced more fractures remains unanswered in the published data.

Comparator Choice and Its Implications

Placebo gel was the correct comparator for a safety trial. Some clinicians have questioned whether the comparator should have included calcium/vitamin D supplementation in both arms. The protocol did not mandate bone-protective co-interventions, and baseline 25-hydroxyvitamin D levels were not systematically reported.

This design choice means the fracture signal exists in a context where neither arm received optimized bone care. Whether testosterone still increases fractures when vitamin D is replete and calcium intake is adequate remains unknown. The American Urological Association guideline recommends monitoring bone density in men on TRT but does not mandate co-administration of calcium or vitamin D.

Mechanistic Plausibility

The fracture increase seems paradoxical given that testosterone stimulates osteoblast activity and the Testosterone Trials (TTrials) bone substudy showed improved volumetric BMD with 12 months of testosterone. Several mechanisms could explain why density gains do not translate to fracture reduction:

| Hypothesis | Evidence Level | |---|---| | Increased physical activity and fall-prone behavior | Plausible; TRAVERSE reported higher activity levels in testosterone arm | | Bone turnover acceleration with transient fragility | Supported by analogy to teriparatide early-phase remodeling | | Cortical porosity from high-turnover state | Shown in preclinical models | | Confounding by increased muscle mass shifting fall dynamics | Speculative |

The TTrials bone data demonstrated that testosterone increased trabecular volumetric BMD by 6.8% at the spine over 12 months but had variable cortical effects. TRAVERSE's 3.2-year follow-up may capture longer-term remodeling consequences that short-duration density studies miss.

Limitations the Authors Acknowledged

The investigators explicitly noted:

1. Clinical fractures only (no systematic vertebral imaging), potentially missing asymptomatic events equally in both arms
2. The cardiovascular risk enrichment limits generalizability to younger or healthier hypogonadal men
3. No pre-randomization DXA was performed, so baseline BMD distribution between arms is unknown
4. The achieved testosterone levels (mean 358 ng/dL) represent low-normal replacement, and results may differ at higher targets
5. Falls were not systematically tracked, preventing separation of skeletal fragility from fall frequency

What This Means for Clinical Practice

The FDA-approved testosterone labeling does not yet include a fracture warning based on TRAVERSE, as of the most recent label revision. Whether this signal will prompt a labeling change depends on FDA's assessment of the totality of evidence and whether confirmatory data emerge.

For prescribers, the TRAVERSE bone findings do not invalidate TRT for symptomatic hypogonadism. They do suggest that the assumption of bone protection from testosterone replacement requires reassessment, particularly in older men with cardiovascular comorbidity. Baseline DXA, fall-risk assessment, and consideration of bone-protective co-therapy are reasonable clinical responses while confirmatory data accumulate.

Frequently asked questions

›

›

›

›

›

›

›

›

›

›

References

1. Lincoff AM, Bhasin S, Flevaris P, et al. Testosterone replacement therapy and bone fractures in men with hypogonadism. N Engl J Med. 2024. PubMed
2. Lincoff AM, Bhasin S, Flevaris P, et al. Cardiovascular safety of testosterone-replacement therapy. N Engl J Med. 2023;389(2):107-117. PubMed
3. Snyder PJ, Kopperdahl DL, Stephens-Shields AJ, et al. Effect of testosterone treatment on volumetric bone density and strength in older men with low testosterone: a controlled clinical trial. JAMA Intern Med. 2017;177(4):471-479. PubMed
4. 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
5. Mulhall JP, Trost LW, Brannigan RE, et al. Evaluation and management of testosterone deficiency: AUA guideline. J Urol. 2018;200(2):423-432. PubMed
6. FDA. AndroGel (testosterone gel) 1.62% prescribing information. FDA Label