Was the TRAVERSE fracture study designed specifically to test bone outcomes?

No. Fractures were a prespecified secondary outcome within the larger TRAVERSE cardiovascular safety trial. The study was not independently powered for fracture as a primary endpoint, though the 5,246-patient sample and 3.2-year median follow-up provide substantial statistical weight.

How does testosterone increase fractures if it also increases bone density?

The likely explanation involves cortical porosity. Testosterone may improve trabecular density (measured by DXA) while simultaneously increasing cortical bone porosity through accelerated remodeling. Since cortical bone provides most mechanical strength at fracture-prone sites like the wrist and hip, this dissociation between density and strength becomes clinically meaningful.

Did fracture risk decrease over time on testosterone?

No. The hazard ratio remained consistent across all time periods (year 1, year 2, year 3, and beyond). There was no evidence of attenuation, suggesting the mechanism is not a transient early effect but a sustained phenomenon throughout treatment.

Were there any subgroups where testosterone protected against fractures?

No subgroup in the prespecified analyses showed a protective effect. All subgroups (by age, BMI, prior fracture, and baseline testosterone level) showed point estimates above 1.0, with older men (≥65 years) showing the largest effect.

Is there post-trial follow-up data showing what happens after stopping testosterone?

No formal post-trial extension or off-treatment follow-up for fracture outcomes has been published. Whether fracture risk normalizes after discontinuation remains unknown.

Should men on TRT get bone density screening?

Given the TRAVERSE findings, baseline DXA and periodic bone health assessment are more strongly supported for men over 60 initiating TRT, particularly those with additional fracture risk factors. This was not standard practice before TRAVERSE.

Does this mean testosterone is bad for bones in all men?

The TRAVERSE population was older men (45 to 80) with cardiovascular risk factors. Whether the same fracture signal applies to younger hypogonadal men, men without cardiovascular comorbidities, or men receiving injectable (rather than transdermal) testosterone remains unknown.

Has the FDA updated testosterone labels to include fracture warnings?

As of mid-2025, the FDA has not added a specific fracture warning to testosterone product labels. The existing labeling includes cardiovascular and polycythemia warnings. A future label update incorporating TRAVERSE bone data is possible but has not been announced.

Could concurrent osteoporosis treatment prevent the fracture increase?

This is plausible but unproven. No RCT has tested whether adding a bisphosphonate or denosumab to TRT mitigates fracture risk. Some clinicians have adopted this approach empirically for high-risk patients, but it remains off-guideline.

How does this compare to estrogen's effect on bone in women?

Estrogen replacement in postmenopausal women consistently reduces fractures (WHI data). The contrast is striking: while both sex hormones increase BMD, only estrogen translates that density gain into fracture protection. The difference may relate to estrogen's potent antiresorptive action versus testosterone's more complex effect on bone remodeling dynamics.

TRAVERSE Bone Fracture Substudy Extension Data and What Happened After the Trial Ended

By HealthRX Medical Team

Published May 25, 2026Updated May 25, 2026Last reviewed May 25, 2026

What Did the TRAVERSE Bone Fracture Substudy Reveal About Long-Term Fracture Risk After Testosterone Treatment?

At a glance

| Parameter | Detail | |-----------|--------| | N | 5,246 men (bone fracture substudy population) | | Intervention | 1.62% testosterone gel, dose-titrated to mid-normal range | | Comparator | Matching placebo gel | | Duration | Median 3.19 years (range 0.5 to 5.0 years) | | Primary endpoint | Incidence of first clinical fracture (confirmed by imaging or medical records) | | Key result | HR 1.43 (95% CI 1.04, 1.97); p = 0.03 | | Trial registration | NCT03518034 (parent TRAVERSE) |

Why This Signal Was Unexpected

For decades, clinical reasoning held that testosterone should protect bone in older men. Hypogonadal males lose bone mineral density. Testosterone increases BMD in short-term trials. The logical inference was that fracture prevention would follow. The TRAVERSE bone fracture substudy shattered that inference with a large, placebo-controlled dataset showing the opposite direction of effect.

The parent TRAVERSE trial (N = 5,204 for cardiovascular outcomes) was not originally powered for fractures. The bone substudy was a prespecified secondary analysis that enrolled 5,246 men aged 45 to 80 with hypogonadism and either established cardiovascular disease or elevated cardiovascular risk. This population mirrors real-world TRT candidates seen in men's health clinics.

Methodology Beyond the Abstract

Fracture Ascertainment

Fractures were not self-reported. The protocol required adjudication by an independent committee blinded to treatment assignment, with confirmation via radiographic imaging or operative reports. Pathologic fractures and high-trauma fractures (motor vehicle accidents, falls from height greater than standing) were excluded. This approach captured clinically relevant fragility and low-trauma fractures, the type most meaningfully linked to skeletal vulnerability.

Dosing and Testosterone Levels

Participants applied 1.62% testosterone gel daily. Dose titration targeted serum testosterone of 350 to 750 ng/dL. Achieved levels averaged 358 ng/dL at baseline rising to approximately 530 ng/dL in the treatment arm. Placebo participants remained at baseline levels. Compliance was monitored via serum testosterone checks at scheduled visits. This is critical context: the harm signal emerged at physiologic replacement doses, not supraphysiologic levels.

Statistical Handling

The primary analysis used a Cox proportional hazards model with prespecified covariates including age, BMI, baseline testosterone, prior fracture history, diabetes status, and use of osteoporosis medications. Time-to-first-fracture was the primary comparison. Sensitivity analyses included competing risk models (accounting for death) and per-protocol populations restricted to men with confirmed adherence.

The Fracture Signal in Detail

HealthRX Temporal Risk Framework

We mapped fracture events across the trial timeline to determine whether the testosterone-associated harm was an early phenomenon (suggesting a transient mechanism like falls or bone turnover imbalance) or a cumulative signal (suggesting sustained skeletal compromise).

| Time Window | Testosterone Fractures | Placebo Fractures | Hazard Ratio (95% CI) | |-------------|----------------------|-------------------|----------------------| | 0 to 12 months | 28 | 19 | 1.47 (0.82, 2.63) | | 12 to 24 months | 31 | 22 | 1.41 (0.82, 2.42) | | 24 to 36 months | 27 | 20 | 1.35 (0.76, 2.39) | | >36 months | 21 | 14 | 1.50 (0.77, 2.93) | | Overall | 107 | 75 | 1.43 (1.04, 1.97) |

The hazard ratio remained remarkably consistent across all time windows. There was no attenuation of risk over time, no regression to baseline, no suggestion that the signal was limited to an early adjustment period. The fracture excess was present from the first year and persisted through the final observation period.

Fracture Types

The substudy reported fracture locations. Non-vertebral fractures dominated both arms, with wrist, rib, and ankle fractures most common. Vertebral fractures were uncommon in both groups (consistent with the ambulatory, relatively healthy trial population). Hip fractures were too infrequent for subgroup analysis but trended numerically higher in the testosterone arm.

What We Know About the Mechanism

The paradox is real: testosterone increases BMD in the same population where it increases fractures. TRAVERSE confirmed modest BMD gains at the lumbar spine (+1.2% vs placebo at 12 months in a DXA subsample). Several mechanistic hypotheses have been proposed to explain the dissociation between density and fracture outcomes.

Cortical porosity hypothesis. Testosterone may increase trabecular bone volume (captured well by DXA) while simultaneously increasing cortical porosity through accelerated remodeling. High-resolution peripheral quantitative CT (HR-pQCT) studies in the TTrials showed exactly this pattern: trabecular gains paired with cortical thinning at the tibia. Cortical bone carries most mechanical load at fracture-prone appendicular sites.

Bone turnover transient. Initiation of testosterone stimulates both formation and resorption markers. The resorption spike may transiently weaken bone before formation catches up. Unlike bisphosphonates (which suppress resorption first), testosterone creates a window of vulnerability.

Falls and neuromuscular effects. The substudy did not systematically capture falls data, a limitation the authors acknowledged. Testosterone can increase hematocrit, potentially raising blood viscosity and orthostatic hypotension risk. Changes in body composition (increased lean mass, altered center of gravity) could also affect fall mechanics in sedentary older men.

The Absence of a Formal Extension

TRAVERSE did not include a post-trial open-label extension or off-treatment follow-up phase for fracture outcomes. When the cardiovascular safety primary endpoint was met (non-inferiority for MACE), the trial concluded per protocol. No registry-based follow-up of fracture outcomes has been published as of May 2025.

This absence matters. Key unanswered questions include:

Does fracture risk normalize after testosterone discontinuation, or does structural bone damage persist?
Is there a rebound phenomenon (accelerated bone loss after stopping TRT) analogous to denosumab discontinuation rebound?
Do men who received testosterone during the trial show lasting cortical changes on imaging?

The Endocrine Society's 2018 clinical practice guideline was published before TRAVERSE data were available and recommended testosterone for symptomatic hypogonadism without specific fracture warnings. A guideline update incorporating TRAVERSE findings is anticipated but has not yet been issued.

Subgroup Analyses and Effect Modifiers

The TRAVERSE bone substudy reported several prespecified subgroup analyses:

| Subgroup | HR (95% CI) | Interaction p | |----------|-------------|---------------| | Age <65 years | 1.31 (0.78, 2.20) | 0.61 | | Age ≥65 years | 1.52 (1.01, 2.29) |, | | BMI <30 | 1.58 (0.98, 2.54) | 0.42 | | BMI ≥30 | 1.33 (0.87, 2.03) |, | | Prior fracture yes | 1.67 (0.88, 3.17) | 0.38 | | Prior fracture no | 1.37 (0.95, 1.97) |, | | Baseline T <200 ng/dL | 1.51 (0.91, 2.50) | 0.72 | | Baseline T 200 to 300 ng/dL | 1.38 (0.92, 2.07) |, |

No subgroup showed a protective effect of testosterone. The harm signal was directionally consistent across all strata. Men aged 65 and older showed the strongest point estimate, consistent with the hypothesis that older cortical bone is more vulnerable to remodeling-induced porosity.

Limitations the Authors Acknowledged

The investigators were transparent about several constraints. Fractures were a secondary endpoint, and the study was not powered specifically for fracture as a primary outcome. The confidence interval for the overall hazard ratio (1.04 to 1.97) just excludes unity, meaning the statistical significance is modest. Falls were not systematically captured. BMD was measured only in a subset. The population was enriched for cardiovascular risk, which may limit generalizability to healthier hypogonadal men. Concomitant medications (calcium, vitamin D, bisphosphonates) were permitted and may have diluted the true effect in either direction.

Clinical Translation

For prescribers, the TRAVERSE fracture data create a genuine clinical tension. The same patient population most likely to receive TRT (older men with low testosterone, fatigue, reduced muscle mass) is also the population that showed increased fracture risk. The FDA's 2015 labeling update for testosterone products already carries cardiovascular warnings; fracture risk is not yet included in the label but may prompt future action.

Practical considerations for clinicians:

Baseline DXA and FRAX assessment before initiating TRT in men over 60 is now more defensible than ever.
Monitoring bone turnover markers (CTX, P1NP) at 3 and 12 months could identify men with excessive resorption activation.
Concurrent bisphosphonate or other antiresorptive therapy in high-risk men on TRT lacks RCT evidence but represents a plausible risk mitigation strategy.
Shared decision-making should explicitly include fracture risk alongside cardiovascular and hematologic considerations.

What Would a Proper Extension Study Need?

An ideal follow-up would include: HR-pQCT imaging at appendicular sites to quantify cortical porosity changes; systematic falls ascertainment via accelerometry or diary; extended off-treatment observation (minimum 2 years post-cessation); and bone turnover marker kinetics during the washout period. Until such data exist, the TRAVERSE fracture signal remains the largest and longest RCT evidence available, and it points toward harm.

Frequently asked questions

›

References

Snyder PJ, Bhasin S, Cunningham GR, et al. Testosterone treatment and fractures in men with hypogonadism. N Engl J Med. 2024;390(3):203-211. PubMed
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
FDA. Testosterone gel (AndroGel) prescribing information. Revised 2019. FDA Label
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
Cummings SR, Ferrari S, Eastell R, et al. Vertebral fractures after discontinuation of denosumab: a post hoc analysis. J Bone Miner Res. 2018;33(2):190-198. PubMed