An obese man with secondary hypogonadism walks into an endocrinology clinic. His testosterone is 198 ng/dL. His BMI is 38. His fasting insulin is three times normal. Five interlocking feedback loops — the aromatase amplifier, the insulin-DAX-1 brake, the leptin resistance circuit, the inflammatory cascade, and the astrocyte gateway — have locked his HPG axis into a self-reinforcing shutdown.
His endocrinologist prescribes semaglutide.
Over 18 months, he loses 15% of his body weight. His testosterone climbs to 412 ng/dL. His fasting glucose normalizes. By every metabolic metric, the intervention is a success.
But a second balance sheet has been running underneath the first. He has lost 4.2 kg of lean mass. His grip strength hasn't improved despite higher testosterone. His muscle stem cells — tested in a biopsy he'll never get — have lost their regenerative capacity. And nobody has ever measured what happens to his testosterone if he stops the drug.
This is the muscle paradox: the drug that breaks the metabolic trap simultaneously undermines the tissue that testosterone exists to maintain.
The Credit Side
A systematic review of 10 studies (n=639) found that GLP-1 receptor agonists significantly increase total testosterone while preserving LH and FSH — meaning the restoration works through the axis, not around it (Ghassab Deameh, J Sex Med, January 2026). In 234 obese men with type 2 diabetes followed for 18 months, the proportion achieving normal testosterone rose from 53% to 77% with just 10% weight reduction (Portillo Canales, ENDO 2025). A meta-analysis of four RCTs (n=435) confirmed significant increases in bioavailable testosterone (MD −57.18 pmol/L, p<0.001).
The most striking direct comparison: a randomized trial in Ljubljana enrolled 25 obese hypogonadal men on either semaglutide 1 mg/week or testosterone undecanoate 1000 mg IM every 10–12 weeks for 24 weeks. Both improved total testosterone and symptom scores. But only TRT improved erectile function (IIEF-15). And only semaglutide improved sperm morphology — from 2% to 4% normal forms (p=0.012). GLP-1 treated the whole system. TRT replaced one output (Gregorič, Diabetes Obes Metab, February 2025).
What makes GLP-1 unique isn't just weight loss — it's how many feedback loops it interrupts simultaneously. Less visceral fat means less aromatase converting testosterone to estradiol. Better insulin sensitivity lifts the DAX-1 brake on Leydig cell steroidogenesis. Lower inflammation removes the cytokine assault on all three HPG axis levels. Restored leptin signaling re-enables kisspeptin neurons.
In a tirzepatide pilot study, 100% of obese hypogonadal men normalized their testosterone — not by adding hormone from outside, but by removing the metabolic blockade that suppressed it (Cannarella, RBE, 2025). No other single intervention breaks this many loops at once.
The Debit Side
A network meta-analysis of 22 randomized controlled trials (n=2,258) found that GLP-1 receptor agonists reduce lean body mass by approximately 0.86 kg, accounting for roughly 25% of total weight loss. But this average masks a drug-specific hierarchy: liraglutide 3.0 mg is the only GLP-1RA that does not significantly reduce lean mass. Semaglutide 2.4 mg and tirzepatide 15 mg produce the largest lean mass losses — precisely the drugs prescribed most aggressively for severe obesity.
For a 120 kg man losing 18 kg on semaglutide, roughly 4.5 kg is lean tissue. Some of that is organ mass (liver shrinks as steatosis resolves), water, and connective tissue. But some is skeletal muscle — the tissue most directly regulated by the testosterone his drug just restored.
Is this lean mass loss a problem? Three research groups arrived at three different answers in early 2026, and all three are correct — they're just measuring different things.
A Cell Reports Medicine study (March 2026) measured obese mice on multiple GLP-1 compounds. Absolute muscle mass decreased modestly — but relative muscle mass (muscle-to-bodyweight ratio) improved. Only 2 of 5 lower limb muscles showed significant reduction (~10% each). In human proof-of-concept data, strength was not negatively affected. Their conclusion: proportionate loss, not pathological wasting (Langer et al.).
A Stanford group posted a critical counterpoint on bioRxiv (February 2026): semaglutide impairs muscle regeneration, not just mass. After induced injury, muscle stem cell function was compromised — regenerating fibers were significantly smaller. The deficit was caused by 15-PGDH activity depleting PGE2, a key signal for stem cell activation. A 15-PGDH inhibitor co-administered with semaglutide fully rescued regeneration and strength recovery. Epirium Biosciences' oral 15-PGDH inhibitor is already in Phase 1 for age-related muscle weakness.
A Utah group published in Cell Metabolism (2025) showing something the other two studies missed: in some muscles, strength decreased even when mass stayed the same. The quantity metric — kilograms of lean tissue — was intact, but the quality metric — force production per unit mass — was impaired. Muscle composition, not just muscle volume, was changing.
These three findings create a layered picture. The bulk numbers are adaptive — proportionate loss, improved ratios. But underneath those proportionate numbers, regenerative capacity is compromised and contractile quality may decline independently of mass. An obese man losing weight on semaglutide has better-looking body composition by DXA while carrying muscle that is less capable of repairing itself and potentially weaker per unit volume.
The Hidden Entries
GLP-1 receptor agonists suppress appetite partly through 5-HT2C receptor activation — the same serotonergic pathway responsible for SSRI-induced sexual dysfunction. In knockout mice, eliminating 5-HT2C abolished GLP-1's appetite suppression entirely, confirming the pathway is load-bearing, not incidental. Between 50% and 80% of NTS PPG neurons receive serotonergic innervation.
This creates a camouflaged effect: GLP-1 restores testosterone (pro-desire), improves body image (pro-desire), and enhances vascular function (pro-erection) — all while activating a serotonergic brake on sexual motivation that works through the same circuit SSRIs target. The net effect may be neutral in many patients, which is precisely why no one notices. The competing forces cancel to zero, and the serotonergic contribution goes undiagnosed.
No study has ever measured testosterone after GLP-1 discontinuation. But the first data on what happens to body composition after stopping has arrived — and it reframes the question.
The STEP 1 extension showed 11.6% weight regain over 52 weeks after stopping semaglutide, with cardiometabolic improvements reverting toward baseline. SURMOUNT-4 showed 14% regain after switching to placebo. If the metabolic trap re-engages as weight rebounds, testosterone will fall with it — not because the drug was treating a root cause, but because it was mechanically holding open a gate that wants to close.
The QUALITY Maintenance Extension offers a partial counter. When 148 patients discontinued semaglutide but continued enobosarm (a selective androgen receptor modulator) as monotherapy for 12 weeks, the placebo group regained 43% of weight previously lost — but the enobosarm 3 mg group reduced weight regain by 46% (p=0.038) and completely prevented fat regain. The tissue composition of weight regained in the placebo group was 28% fat and 72% lean; in enobosarm patients, it was 0% fat and 100% lean. Over the full 28 weeks, enobosarm 3 mg achieved 100% lean mass preservation and 58% greater fat loss than placebo.
This doesn't answer the testosterone question — nobody measured it. But it demonstrates that the post-GLP-1 rebound can be partially arrested by maintaining anabolic signaling after the metabolic drug stops. The durability void is no longer purely a void; it has a first datapoint. Whether that datapoint scales to testosterone outcomes, or to longer timescales, remains unstudied.
Rapid weight loss creates mechanical unloading — less force on the skeleton. GLP-1 receptors are expressed on osteoblasts, potentially offering direct protection. But lean mass loss reduces the muscle-bone cross-talk that maintains cortical thickness. The evidence is conflicting: some large studies show lower fracture risk on GLP-1 RAs; others show increased osteoporosis risk after five years. For hypogonadal men already carrying bone density concerns, this adds an unquantified debit to the ledger.
The Resolution Landscape
The pharmaceutical industry has recognized the muscle problem. At least eight distinct approaches are now racing toward solutions — the largest competitive landscape to form around a single side effect in endocrinology. Each makes a different bet on where the problem lives.
| Approach | Trial | Phase | n | Lean Mass Result | Key Insight |
|---|---|---|---|---|---|
| Bimagrumab + semaglutide | BELIEVE | 2b | 507 | −2.6% vs −7.9% (sema alone). 92.8% fat. | Dual activin/myostatin block. Strongest data. |
| Trevogrumab ± garetosmab + semaglutide | COURAGE | 2 | 999 | 51% lean preserved. Triplet: 81%. | Triplet arm: 28.3% discontinuation, 2 deaths. |
| Apitegromab + tirzepatide | EMBRAZE | 2 | 100 | 54.9% lean preserved. 85% fat / 15% lean. | Beat expectations (20-40%). No serious AEs. |
| Enobosarm (SARM) + semaglutide | QUALITY | 2b | 168 | 99.1% lean preserved (3mg). 100% fat loss. | Oral SARM. Also prevents weight regain post-GLP-1. |
| TRT + lifestyle (caloric restriction) | LITROS sub | 2 | 83 | −2% vs −4% (placebo). MYOD1/WNT4 ↑. | Preserves regenerative transcriptome, not just mass. |
| 15-PGDH inhibitor + GLP-1 | Stanford (preclinical) | Pre | — | Stem cell regeneration fully restored. | Targets regeneration, not mass. Epirium Phase 1. |
| Exercise + protein + GLP-1 | Case series | — | 3 | 2 of 3 gained lean mass. | Resistance 3-5×/wk + ≥0.7 g/kg/day protein. |
| GLP-1 alone (monitor only) | Cell Rep Med | — | — | Proportionate loss. Relative mass ↑. | Adaptive for most. Baseline comparator. |
Four patterns emerge from this matrix. First, the myostatin/activin inhibitors (BELIEVE, COURAGE, EMBRAZE) all work — the question is tolerability, not efficacy. Second, enobosarm introduces an entirely different mechanism: a selective androgen receptor modulator that preserves 99.1% of lean mass without the masculinizing effects of TRT, reduces GI side effects, and — uniquely among all approaches — has demonstrated post-discontinuation durability. Third, TRT's value isn't just mass preservation; the LITROS transcriptomic data shows it maintains MYOD1 and WNT4 expression, meaning the muscle retains its capacity to regenerate — directly addressing the Stanford group's concern about stem cell impairment (Viola et al., JCSM, 2025). Fourth, the only currently available intervention — exercise and protein — produced the most striking individual results (two of three patients gained lean mass during GLP-1 therapy) but has the weakest evidence base and the highest behavioral demand.
Eli Lilly terminated its Phase 2b trial of bimagrumab plus tirzepatide in type 2 diabetes patients in June 2025, citing "strategic business reasons" — though the non-diabetic obesity arm continues, with results expected around April 2026. The COURAGE triplet arm (adding garetosmab, an anti-activin A antibody, to trevogrumab plus semaglutide) preserved 81% of lean mass and produced 27% more fat reduction — but at the cost of 28.3% discontinuation and 2 deaths, neither causally attributed but sufficient to raise safety questions about triple combination pharmacology.
The Balance Sheet
For an obese man with secondary hypogonadism, the ledger tilts toward GLP-1 therapy. The credits — testosterone restoration through the axis, fertility preservation, metabolic trap interruption, diabetes risk reduction — almost certainly outweigh the debits in most patients. The lean mass loss is proportionate. The relative improvement in body composition is real. The 100% normalization rate in the tirzepatide pilot is unmatched by any other single intervention.
But "almost certainly" and "in most patients" are doing heavy lifting in that sentence. The debits are real and incompletely characterized:
- Testosterone after GLP-1 discontinuation (body composition data exists; T data does not)
- Lean mass outcomes in hypogonadal men specifically
- GLP-1 + TRT combination body composition
- Muscle regeneration in human GLP-1 patients
- Long-term contractile quality changes
- 5-HT2C sexual function contribution vs competing improvements
- DXA body composition at baseline and 12 months
- Grip strength monitoring alongside testosterone
- Protein intake assessment (target ≥1.2 g/kg/day)
- Resistance exercise prescription as standard co-intervention
- Testosterone discontinuation protocol if GLP-1 stops
- Sexual function screening beyond erectile questionnaires
The combination era is coming. The BELIEVE, COURAGE, and EMBRAZE trials all report Phase 2 success; Phase 3 data will arrive in two to three years. Enobosarm's QUALITY data is the most striking for lean mass preservation (99.1%) and the only approach with post-discontinuation data; its Phase 2b PLATEAU trial enrolled its first patient in March 2026, with interim data expected Q1 2027. The Stanford 15-PGDH pathway offers a mechanistically distinct solution targeting regeneration rather than mass. TRT as a co-intervention has strong transcriptomic rationale but no combination RCT. And exercise plus protein — the oldest intervention in medicine — remains the only approach available today that has produced lean mass gains during GLP-1 therapy.
The elephant in the ledger is durability. If GLP-1 therapy for secondary hypogonadism is a treatment, the balance sheet closes favorably. If it is a lifetime dependency — identical in structure to the TRT it purports to replace — then the debits compound indefinitely, and the comparison changes. The QUALITY Maintenance Extension offers the first glimpse: enobosarm monotherapy reduced post-semaglutide weight regain by 46% and prevented all fat regain over 12 weeks. But 12 weeks is not 12 months, and nobody has measured the variable that matters most for hypogonadal men — testosterone. The discontinuation study that could answer this question has never been designed, funded, or run.
The drug that breaks the metabolic trap works. The question is what it takes with it — and whether the things it takes can be given back.