The GLP-1 Testicular Paradox: When the Drug That Restores Your Testosterone May Also Suppress It

Across five previous articles on this site, I have described GLP-1 receptor agonists as protective of the HPG axis — breaking the metabolic trap, restoring testosterone in obese men, offering a fertility-sparing alternative to TRT. All of that evidence is real. But it is incomplete.

New data from human testicular organoids, pharmacovigilance databases, and serotonergic neuroscience reveals a parallel story: the same class of drugs may simultaneously suppress testosterone production at the cellular level, downregulate the receptor Leydig cells need to respond to LH, and blunt sexual function through a central serotonergic mechanism unrelated to hormones. These effects are masked — overwhelmed by metabolic benefits in the obese men who dominate every clinical trial. But in lean men prescribed these drugs for weight maintenance, the mask comes off.

This is the most densely connected paradox in the series. It touches the metabolic trap, the muscle paradox, the SSRI paradox, the drug map, and the treatment framework. It also forces a correction to what I've written before.

The Three-Layer Paradox

The GLP-1 testicular paradox operates at three distinct biological levels, each with independent evidence, each pulling testosterone in a different direction.

In metabolically unwell men, Layer 2 overwhelms Layers 1 and 3. The net effect is unambiguously positive. In lean or eugonadal men, Layers 1 and 3 have no metabolic benefit to mask them. The net effect inverts.

The Organoid Evidence

Huang et al. (Theranostics, January 2025) constructed human iPSC-derived testicular organoids — three-dimensional tissue models with functional Leydig cells — and exposed them to semaglutide at 10, 20, and 40 ng/mL for 24 hours. Therapeutic semaglutide plasma concentrations peak at 30-40 ng/mL, so these doses are clinically relevant.

The KISS1/KISS1R upregulation is the most intriguing finding. In the intact HPG axis, kisspeptin from Leydig cells signals to spermatids in a paracrine loop. The organoid, isolated from the hypothalamic-pituitary axis, appears to mount a compensatory kisspeptin response to LHCGR loss — but with no pituitary to receive that signal, compensation cannot complete. This is consistent with the GLP-1→PPARα→Kiss1→steroidogenesis pathway demonstrated in diabetic rat models, where liraglutide upregulates testicular KISS1 and KISS1R and restores steroidogenesis. The difference: in the diabetic rat, the whole axis is present. In the organoid — and potentially in a healthy, non-metabolically-compromised testis — it isn't.

Huang's study also included a Mendelian randomization component using cis-eQTL variants of the GLP-1R gene as instrumental variables, providing genetic evidence supporting a causal link between GLP-1R activation and reduced testosterone. This is not observational noise. It is genetically instrumented.

The Clinical Reality: Who Wins, Who Loses

The clinical data is overwhelmingly positive — but the populations studied are overwhelmingly metabolically compromised.

The pattern is unmistakable. Every positive study is dominated by men who have something metabolic to fix. The TriNetX signal — the only large database study to isolate non-diabetic men — shows the opposite: new erectile dysfunction, new testosterone deficiency. The MR studies appear contradictory until you recognize they capture different pathways: the MR instruments select for the metabolic cascade (obesity → insulin resistance → aromatization → low T), while the pharmacovigilance data captures the direct testicular and serotonergic effects that only become visible when the metabolic benefit is small.

The Serotonergic Brake

The third layer connects to territory I covered in the SSRI paradox article: serotonergic suppression of sexual function independent of testosterone.

Tveit, Gelfand, and Simon (J Sex Med, 2025) proposed a biopsychosocial model for GLP-1RA-associated sexual dysfunction. The core pharmacology: GLP-1 agonism activates 5-HT2C receptors in brainstem NTS neurons. 50-80% of PPG (preproglucagon) neurons in the NTS receive serotonergic innervation. This is how GLP-1 RAs suppress appetite — but the 5-HT2C receptor does not discriminate between food reward and sexual reward. The appetite-suppressing effect of GLP-1 is abolished in 5-HT2C knockout mice.

"Reward blunting" affects food, alcohol, AND sex. The mechanism that makes these drugs suppress hunger also suppresses desire.

— Tveit, Gelfand & Simon, J Sex Med 2025

SSRIs activate a different serotonergic brake — the RFRP/GnIH pathway that directly suppresses GnRH pulsatility. GLP-1 RAs activate the 5-HT2C reward blunting pathway. Different molecular targets, convergent clinical result: decreased libido that hormone panels cannot explain. A man on semaglutide with normal testosterone and normal gonadotropins can still experience diminished sexual desire. The hormone system isn't broken; the brain has turned down the volume on wanting.

This is obscured in clinical practice because the competing effects pull in opposite directions. The man who loses 30 kg on semaglutide feels more energetic, more confident, has better erections from improved vascular function — and these improvements mask the serotonergic blunting. The man who was already lean and started GLP-1 RAs for cosmetic weight loss has no metabolic upside to mask the central effect.

The Tirzepatide Divergence

Not all GLP-1 RAs are equivalent. The Cannarella tirzepatide pilot produced the strongest T restoration of any study in this literature — 100% hypogonadism reversal, significant LH and FSH increases, better outcomes than TRT at 2 months. Tirzepatide is a dual GIP/GLP-1 receptor agonist. The GIP receptor co-agonism may provide additional metabolic benefits with less serotonergic penalty.

Supporting signals are scattered but consistent. In FAERS pharmacovigilance data, tirzepatide accounted for 14.3% of male sexual dysfunction reports versus semaglutide's 21.4%. One case report describes a woman with GLP-1RA-induced anorgasmia that resolved when switching from semaglutide to tirzepatide. The GIP receptor activates partially overlapping but distinct intracellular pathways. Whether this translates to a genuinely different testicular or serotonergic profile requires head-to-head comparison data that does not yet exist.

The Desensitization Question

A critical unknown: what happens to testicular GLP-1R signaling over months and years?

In pancreatic beta cells, chronic GLP-1R stimulation leads to beta-arrestin-mediated receptor desensitization — the cell internalizes the receptor to protect itself from overstimulation (Moo et al., FEBS J 2025). Arrestins are essential for this process: in arrestin-knockout cells, GLP-1R loses the ability to desensitize. GLP-1R is expressed at its highest testicular levels in Leydig cells. But no study has examined beta-arrestin-mediated GLP-1R desensitization in testicular tissue.

The clinical implication: if GLP-1R desensitizes in Leydig cells the way it does in beta cells, the acute stimulatory effects (the cAMP/PKA boost that some rodent studies show is protective) may fade over time, while the LHCGR downregulation documented in organoids may persist or accumulate. This would create a time-dependent trap: early treatment looks protective, late treatment becomes suppressive. No long-term study has measured testicular function biomarkers (INSL3, inhibin B) on GLP-1 RAs beyond 2 years.

The Durability Void

What happens to testosterone when GLP-1 RAs are stopped?

No study has directly measured it. What we know from weight regain data is not encouraging:

If testosterone restoration is driven primarily by metabolic improvement (Layer 2), and metabolic improvement reverses on discontinuation, then testosterone gains likely reverse too. But here's the darker possibility: if LHCGR downregulation (Layer 1) accumulates during long-term use, the Leydig cells may return to a worse baseline than before treatment — they've regained the metabolic burden AND lost some LH receptor sensitivity. This is speculative but mechanistically plausible, and no one is studying it.

The Resolution

The paradox resolves along a single axis: metabolic burden.

The more metabolically compromised a man is, the more Layer 2 (metabolic restoration) dominates. GLP-1 RAs are powerfully beneficial for the population they were designed for. But the prescribing frontier has expanded far beyond that population. The 120% surge in prescriptions for men under 25, the DTC platforms selling semaglutide for cosmetic weight loss, the off-label use in men with BMI 27-30 and no metabolic comorbidities — these populations sit in the uncertain middle of the spectrum, or worse, on the left side where Layers 1 and 3 may dominate.

Six Gaps That Need Filling

What This Changes

This article corrects the narrative I built across five previous posts. GLP-1 RAs remain the most promising non-TRT intervention for obesity-associated hypogonadism. That conclusion stands. What changes is the scope of that conclusion: it applies to metabolically compromised men, not to all men on GLP-1 RAs.

For clinicians: the TriNetX signal in non-diabetic obese men warrants testosterone monitoring in men prescribed GLP-1 RAs without clear metabolic indications. The serotonergic pathway warrants proactive sexual function screening, especially in men without significant obesity. And the lack of post-discontinuation data means that men who stop GLP-1 RAs should have testosterone checked during the weight regain period, not assumed to be fine.

For the field: the Huang organoid data and the emerging lean-population signals are exactly the kind of evidence that should slow the expansion of GLP-1RA prescribing into populations where the risk-benefit calculus may be fundamentally different. Fifteen million prescriptions per year and growing. The metabolic indication is clear. The cosmetic indication is not — and the testicular consequences of widespread use in non-indicated populations remain unmeasured.

This is article 33 of an ongoing investigation into secondary hypogonadism in males. It connects to: The Metabolic Trap, The Muscle Paradox, The SSRI Paradox, The Drug Map, The Treatment Framework, and Obesity-Associated HH. Full source list available on request.