The endocrine-muscle facet of the weight-management cluster (orbits the
Low-Carbohydrate vs Balanced-Carbohydrate Diets nucleus). The frame this page installs, and the reason
it is not titled “testosterone”: the serum testosterone level is a surrogate. The
decisions men actually face are about muscle mass and function, fractures, cardiovascular events and
survival — and on the evidence the wiki holds, the intervention that raises the T number most
reliably (replacement therapy) is not the one that best moves those outcomes. Organised around the T
number, this cluster reads as “low T -> replace it.” Organised around outcomes, it reads very
differently, and the reversal is the point.
Three limbs, four sources (a meta-analysis, an umbrella review, and two RCTs), one through-line: judge the axis by patient-important outcomes, not by the hormone.
The surrogate at the centre
Testosterone declines with age and is suppressed by adiposity, and it is easy to measure — so it is the quantity everyone acts on. But T is a marker on the causal chain, not the endpoint. This page holds it to the same discipline as LDL or blood pressure -> Surrogate Outcomes: a marker earns a target only if its transmission to a named patient-important outcome is itself evidenced. For the muscle/bone endpoints, the T->outcome chain fails that test in the hardest possible way (limb 3: bone density rose while fractures rose). For the CV endpoint the RCT shows safety, not benefit. So raising T is licensed narrowly, and the real levers on the outcomes are adiposity reduction and resistance training.
Limb 1 — adiposity suppresses testosterone, and weight loss reverses it
(Corona et al., 2013) Corona’s meta-analysis (24 studies, 13 low-calorie-diet + 11 bariatric-surgery arms, 479 patients, mean follow-up 38 weeks) establishes that obesity-associated low T is reversible by weight loss, and does so on a dose-response:
«Overall, both a low- calorie diet and bariatric surgery are associated with a significant (P!0.0001) increase in plasma sex hormone-binding globulin-bound and -unbound testosterone levels (total testosterone (TT)), with bariatric surgery being more effective in comparison with the low-calorie diet (TT increase: 8.73 (6.51–10.95) vs 2.87 (1.68–4.07) for bariatric surgery and the low-calorie diet, respectively; both P!0.0001 vs baseline).» (Corona et al., 2013)
The determinant is the weight lost, not the method: «Multiple regression analysis shows that the degree of body weight loss is the best determinant of TT rise (BZ2.50G0.98, PZ0.029)» (Corona et al., 2013) — bariatric surgery wins only because it removes more weight (32% vs ~9.8% of body weight). The androgen rise is larger in younger, non-diabetic, more-obese men and those with higher baseline T.
The mechanism names why this is central and not incidental. Obesity-related male hypogonadism is secondary (central), not testicular: «obesity-related male hypogonadism is now considered as a form of hypogonadotropic hypogonad- ism» (Corona et al., 2013) — this central classification is the source’s firm finding. The aromatase/estrogen-feedback pathway on top of it is the source’s hypothesis, not a demonstrated mediator (corrected 2026-08-08). Corona proposes that fat tissue over-expresses aromatase, so estrogen rises and feeds back to suppress the pituitary: «The increased amount of estrogen levels might, in turn, play a negative feedback on both the hypothalamus and the pituitary, leading to decreased LH secretion» (Corona et al., 2013). But Corona’s own meta-regression declines to support circulating E2 as the mediator: it «did not show any association between the D-E2 and TT modifications after weight loss» (Corona et al., 2013), concludes that «other fat-associated factors, besides estrogens, should mediate a weight reduction-induced improvement in testosterone levels» (Corona et al., 2013), and attributes any estrogen role to local rather than circulating aromatization. So weight loss reverses the observed pattern — estradiol falls, LH and FSH rise (+1.31 and +1.79 U/l), T rises — but the E2->LH->T causal chain through it is the source’s hypothesis, left open by its own data, not an established cascade. So in an obese man, a low T reading is often a readout of the adiposity, and the outcome-relevant move is to treat the adiposity (which carries its own large benefits -> Weight-Loss Maintenance and Metabolic Adaptation, Does Weight Loss Reduce Cardiovascular Events), which restores T as a by-product — rather than to treat the number directly.
- Bound the claim: the outcome here is the T level itself — a surrogate. Corona shows weight loss raises T; it does not show that the T rise causes any muscle/function/mortality benefit. Its value is that it relocates the cause (adiposity) and offers a lever that is not TRT.
- The diet-route effect is modest, and the pooled magnitude is likely an over-estimate (corrected 2026-08-08). Corona itself concludes «Body weight loss and lifestyle interventions should be the first approach offered to obese hypogonadal men. However, their effect on testosterone levels is modest» (Corona et al., 2013) — the large TT rise is the 32%-loss bariatric arm; at ~10% diet-driven loss the restoration is small. The pooled figures are paired before/after estimates, and Corona notes «the mean (paired) analysis, which excludes these data, is likely to overestimate the effect of treatments» (Corona et al., 2013). So «restores T» reads as modest restoration by the realistic (non-bariatric) route.
- The one non-health confound worth naming: a co-author is employed by a testosterone manufacturer (Bayer Schering Pharma) — noted, not weighted; the finding is a within-subject before/after effect robust across 24 studies.
Limb 2 — testosterone therapy builds muscle mass, but function lags and it is a surrogate
(López Hilario et al., 2026) The gold-tier umbrella review (11 SRs, 121 unique RCTs, ~20,846 men — none women) is the best synthesis of what giving testosterone does to muscle. The honest headline is mass yes, function maybe:
- Lean body mass rises, moderate-to-high certainty: «Lean body mass changes in response to testosterone reported in 5 SRs of moderate to high quality ranged from 1.6 kg (95% CI 0.6---2.6; I2 = 0.0%) to 3.59 kg (95% CI 2.38---4.81» (López Hilario et al., 2026).
- Lower-limb strength rises modestly: leg press «+91.23» N (95% CI 0.23---182.22) and leg extension +144.10 N (44.21---244.00); handgrip only +1.58 kg and upper-limb measures often non-significant (López Hilario et al., 2026).
- The structural gain does not reliably become a functional one — the key extract:
«improvements in muscle strength and functional performance were modest and heterogeneous, suggesting that increases in muscle mass do not neces- sarily translate into clinically meaningful improvements in physical function.» (López Hilario et al., 2026)
This is the Surrogate Outcomes problem inside a single therapy: lean mass (a surrogate) moves well; physical performance (closer to what a person values — independence, not falling) moves little and uncertainly. Intramuscular routes out-perform transdermal and oral for both mass and strength. The umbrella’s own bottom line is narrow: «TRT may represent a therapeutic option for carefully selected older men with confirmed hypogonadism, particularly to improve body composition» (López Hilario et al., 2026) — body composition, i.e. the surrogate, for confirmed hypogonadism, not a general anabolic. The abstract reports 20,846 male participants; the body text reports 28,046 (both give 11 SRs / 121 unique RCTs). An internal source discrepancy; the ~20,846 headline is used and the conflict flagged, not resolved.
Limb 3 — the hard outcomes: cardiovascular SAFETY (not benefit), and fractures UP
This is where the surrogate story is refuted by the outcomes, and both come from the same trial family (TRAVERSE), so they are one evidence base read on two endpoints — not independent confirmations.
Cardiovascular: non-inferior — a safety clearance, not a benefit
(Lincoff et al., 2023) TRAVERSE (5198 hypogonadal men 45-80 with preexisting/high CV risk, T <300 ng/dl, transdermal T gel, mean 22 months) was a noninferiority trial:
«testosterone-replacement therapy was noninferior to placebo with respect to the inci- dence of major adverse cardiac events. The haz- ard ratio for a primary end-point event was 0.96 (95% CI, 0.78 to 1.17)» (Lincoff et al., 2023)
Read “noninferior” precisely — it clears a fear, it does not add a benefit. MACE was 7.0% vs 7.3%; death from any cause HR 0.98. But TRT carried new harms: «A higher incidence of atrial fibrillation, of acute kidney injury, and of pulmonary embolism was observed in the testosterone group» (Lincoff et al., 2023) (AFib 3.5% vs 2.4%; AKI 2.3% vs 1.5%; PE 0.9% vs 0.5%). The trial’s own framing is the decision anchor: «because testosterone deficiency is not a life-threatening condition, uncertainty about cardiovascular outcomes has weighed on treatment decisions» (Lincoff et al., 2023). A modified-ITT design (analysing men who took >=1 dose) biases toward noninferiority — acknowledged in source.
Fractures: the surrogate inverted — BMD up, fractures UP
(Snyder et al., 2024) The TRAVERSE bone substudy is a striking refutation of a surrogate by its outcome. Prior trials had shown T raises bone density — «Testosterone treatment in men with hypogonadism improves bone density and quality, but trials with a sufficiently large sample and a sufficiently long duration to determine the effect of testosterone on the incidence of fractures are needed.» (Snyder et al., 2024) — and the trial was powered (80%) to detect a 30% reduction in fractures. It found the opposite direction:
«A total of 91 of 2601 participants (3.50%) in the testosterone group and 64 of 2603 participants (2.46%) in the placebo group had one or more clinical fractures, excluding fractures of the ster- num, fingers, toes, facial bones, and skull (haz- ard ratio, 1.43; 95% confidence interval [CI], 1.04 to 1.97)» (Snyder et al., 2024)
The numerical increase was consistent across fracture endpoints, and the authors state the surprise plainly: «We did not expect these results, because most previous studies showed that testosterone im- proved many measures of bone structure and quality» (Snyder et al., 2024). Read the signal at the hedge the authors themselves set (corrected 2026-08-08): the primary endpoint CI does exclude 1 (HR 1.43; 1.04-1.97), but the trial’s own conclusion is deliberately weaker — «The fracture incidence was numerically higher among men who received testosterone than among those who received placebo» (Snyder et al., 2024) — and the statistics are unadjusted: «no adjustment was made for multiple com- parisons. All confidence intervals are unadjust- ed and are not a substitute for hypothesis tests» (Snyder et al., 2024) (most secondary-endpoint CIs cross 1). So this is a directional, unadjusted increase in the primary endpoint — a concerning signal that refutes the expected benefit, not a settled fracture harm. Bone density is exactly the surrogate GRADE names as proximate to fractures (rate down only one level) — the wiki’s methodology point, not from Snyder (corrected 2026-08-08); here the proximate, credentialed surrogate moved the “right” way while the patient-important outcome moved the wrong way -> Surrogate Outcomes. Mechanism is unknown (bone density/structure were not measured in this trial; falls and risk-taking were not assessed).
Independence note (strict): TRAVERSE-CV (Lincoff) and the fracture substudy (Snyder) are the same
trial, and the umbrella’s CV-safety conclusion pools meta-analyses (e.g. Jaiswal 2024) that plausibly
include TRAVERSE. So the CV-safety signal appears in two of the four sources by shared evidence, not
independent replication — no [E-independent] is claimed. The genuinely separate evidence base is
Corona (a different intervention on a different limb) and the fracture endpoint (measured nowhere else
here).
Synthesis — the axis, read by outcomes
(inferred from Corona et al., 2013; Lincoff et al., 2023; López Hilario et al., 2026; Snyder et al., 2024) Line up the three limbs against the outcome menu and the T number stops being the hero:
| Patient-important outcome | What raising T (TRT) does | What treating adiposity / training does |
|---|---|---|
| Muscle mass | rises +1.6 to 3.6 kg (surrogate) | RT + adequate protein builds it (surrogate) -> Protein and Resistance Training for Muscle and Strength |
| Strength / function | modest, does not reliably translate | RT is the proven driver of strength gains |
| Fractures | numerically higher — primary HR 1.43 (1.04-1.97), CIs unadjusted; a signal, not a settled harm | resistance/balance training reduces falls & probably fractures -> Big Rocks (Elderly) |
| CV events | non-inferior (safe), + AFib/AKI/PE | weight loss: own benefits; hard-CV-event benefit unproven via lifestyle route |
| All-cause mortality | no benefit shown (trial not powered for it) | weight-loss MA: modest reduction (held elsewhere) |
| The T level itself | rises (the point of TRT) | rises as a by-product of fat loss (Corona) |
The Layer-1 reading -> Layer 1 - Ranking Interventions for a Stratum. For the muscle/function goal in an older or obese man, the lifestyle levers lead the hormone on the outcomes that matter (function, fractures, survival — muscle mass is itself a surrogate): resistance training is the established driver of strength and carries no fracture/AFib signal, and losing visceral fat restores T naturally and moves liver, glycaemia and blood pressure. TRT buys a surrogate (mass) plus a fracture harm, with no mortality benefit shown and function gains that do not reliably translate. This is a synthesis across the umbrella and the held resistance-training pages, NOT a head-to-head trial — see the guard below.
Guard — “exercise beats T” is not a head-to-head result (parameter check)
| Parameter | TRT (Lopez umbrella) | Resistance training (Morton, held) | Same quantity? |
|---|---|---|---|
| Intervention | testosterone by any route | progressive resistance training + protein | No |
| Lean-mass effect | +1.6 to 3.6 kg | RT is the driver; protein a small adjunct | different comparators |
| Strength effect | modest, translation uncertain | RT builds 1RM; protein adds ~2.5 kg | not measured against each other |
| Direct comparison held? | — | — | No head-to-head trial in the corpus |
So the claim is not “a trial showed exercise beat testosterone.” It is that the two levers were measured separately, and the lifestyle one carries the proven-driver status without T’s fracture and arrhythmia signals — an evidence-weighting judgment, appropriately, that a head-to-head trial could still overturn.
Decision relevance
- A low T reading in an obese man is often a readout of the adiposity (secondary hypogonadism). Treat the fat first: weight loss raises T dose-dependently and delivers benefits TRT does not. This is structural leverage over point-optimization (Layer 1).
- TRT is a narrow option, not a general anabolic or anti-ageing move. On the best evidence it improves body composition in carefully selected men with confirmed hypogonadism; it does not reliably improve physical function, shows no mortality benefit (the safety trial was not powered for it), and in TRAVERSE fractures were numerically higher (primary-endpoint HR 1.43, unadjusted CIs — a signal, not a settled harm; corrected 2026-08-08) and it added AFib/AKI/PE while being CV-non-inferior. The heavy marketing of testosterone optimization runs inversely to its measured outcome benefit (attention is an anti-signal).
- For muscle and function, the lever is resistance training + adequate protein -> Protein and Resistance Training for Muscle and Strength, plus weight-bearing/balance work for falls and fractures.
- Out of scope (appraise, do not prescribe): who is a candidate for TRT, dosing, monitoring hematocrit/PSA, managing the AFib/PE/AKI risks — prescriber acts needing this person’s labs and history.
Limits and gaps
- Men only. The umbrella included no women; nothing here transports to female populations
(
type-Ggap the source names explicitly). The female mirror of this axis is now held -> Menopause and the Shifting Levers: the same adiposity <-> sex-hormone link, running the other direction (loss of estrogen raises the testosterone/estradiol ratio -> central fat), with the same outcome-relevant levers (visceral-fat reduction, resistance training) and the same surrogate discipline applied to estrogen’s BMD->fracture chain -> Hormone Therapy After Menopause. - Function and trajectory under-measured. Every source grades muscle mass well and physical function poorly (the Surrogate Outcomes streetlight effect on the outcome side); the outcome a person most wants — sustained independence, compressed decline — is the least measured.
- No muscle-preserving-weight-loss trial in the elderly — the trade-off between fat loss and lean/bone
loss in older adults is reasoned from mechanism, not measured (
type-G, also flagged in Big Rocks (Elderly)). - Sarcopenia is now a held page -> Sarcopenia Definition and Diagnosis. EWGSOP2 (the consensus the umbrella cites) resolves the mass-vs-function split limb 2 surfaces in the opposite order to the older definition: low muscle strength is now the primary criterion, low muscle mass only confirms — «strength is better than mass in predicting adverse outcomes» (Cruz-Jentoft et al., 2018). This sharpens limb 2’s finding: TRT builds mass reliably but function unreliably — i.e. it moves the demoted, confirmatory parameter, not the primary one. Mass is the surrogate; strength is closer to the outcome -> Surrogate Outcomes. (That demoted mass parameter still independently predicts mortality — Low Muscle Mass and Mortality — so TRT-driven lean-mass gain moves a marker that tracks a hard outcome; but predictor is not target, and TRT’s own outcome record is the fracture increase above, not a mortality benefit.)
- The loop is open. This grades coherence and source-fidelity, never validity; no operation here checks a recommendation against a realized outcome.