The GLP-1 hair-loss question got a second answer on September 3, and it is not the same answer as the first one. Researchers at NYU Langone Health published a two-sample Mendelian randomization analysis in the Journal of Investigative Dermatology reporting that genetically elevated GLP1R expression is associated with a modest increase in male-pattern hair loss risk — around 7% in men already carrying genetic risk for androgenetic alopecia. The signal held after adjustment for hypertension, insulin resistance and lowered testosterone.
That lands on top of a much larger dataset from July. A BMJ target trial emulation using Penn Medicine electronic health records found alopecia diagnoses running 37% higher on GLP-1 receptor agonists than on SGLT-2 inhibitors and 68% higher than on DPP-4 inhibitors. The two papers agree that something is happening. They disagree, usefully, about what — and the difference between "hair you shed and regrow" and "hair that recedes on a genetic clock" is the whole story here.
Research-context information only. Compounds discussed below are research peptides and supplements; some are investigational drugs not approved by the FDA. Semaglutide and tirzepatide are active ingredients in FDA-approved prescription products; the research-peptide and compounded forms discussed and linked below are not FDA-approved and are sold for research purposes only. Findings reported come from published research and self-reported community sources. This article reports what has been documented, not what should be done. Consult a licensed physician for personal medical decisions.
What the September genetic study actually tested
Mendelian randomization is not a trial and it is not a cohort. It uses inherited genetic variants as a natural randomizer: if variants that raise expression of a gene also track with an outcome, that supports a causal direction, because the variants were assigned at conception and cannot be confounded by later behaviour, diet or weight change.
The NYU group applied that design to GLP1R — the gene encoding the receptor that semaglutide, tirzepatide and the rest of the class act on. Genetic instruments for GLP1R expression came from the eQTLGen database (31,684 participants); the male-pattern hair loss outcome came from a Complex Traits Genetics group dataset of 205,327 men. Both samples were predominantly of European ancestry.
| September study (JID) | July study (BMJ) | |
|---|---|---|
| Design | Two-sample Mendelian randomization | Target trial emulation, EHR cohort |
| Population | 31,684 + 205,327 men, genetic | ~12,000 GLP-1 initiators vs. matched comparators |
| Exposure measured | GLP1R gene expression | Actual drug initiation |
| Outcome | Androgenetic alopecia | Any alopecia diagnosis |
| Effect | ~7% increased risk | 37% vs. SGLT-2; 68% vs. DPP-4 |
| Weight loss involved? | No — genetic, not drug | Yes, unavoidably |
The design choice is what makes it interesting. Because the exposure is a genetic variant rather than a prescription, weight loss cannot be the explanation. Nobody's inherited GLP1R expression made them drop 15% of their body mass. If the association is real, it points at receptor activity in or around the follicle rather than at the caloric deficit.
Senior investigator Lynn Petukhova, PhD, framed the finding as "the first genetic link between GLP-1 use and an increased risk of male-pattern hair loss," and noted the possibility of "combination treatments to prevent hair loss" being prescribed alongside GLP-1 therapy. The analysis was restricted to men, because androgenetic alopecia genetics are far better characterised in men; whether the same holds in women is explicitly unanswered.
The July cohort, and the number the headlines dropped
The BMJ paper (Tang and colleagues, BMJ 2026;394:e100077, published July 22) is the one supplying the percentages in circulation. It drew adults with type 2 diabetes from Penn Medicine records who newly started a GLP-1 receptor agonist, an SGLT-2 inhibitor or a DPP-4 inhibitor between January 2019 and September 2024, then emulated the randomisation a trial would have performed.
Two comparisons, both unfavourable to the GLP-1 arm:
| Comparison | GLP-1 rate | Comparator rate | Relative increase |
|---|---|---|---|
| GLP-1 (12,004) vs. SGLT-2 (15,221) | 6.91 / 1,000 py | 5.04 / 1,000 py | 37% |
| GLP-1 (11,964) vs. DPP-4 (11,233) | 6.53 / 1,000 py | 3.89 / 1,000 py | 68% |
Those absolute rates are the number almost every summary leaves out. Seven cases per thousand people per year versus five is a real difference and a small one. A 68% relative increase on a base rate of 0.39% per year is not the same shape of risk as a 68% increase on something common, and reading the percentage without the denominator overstates it substantially.
The authors' own explanation leaned on the caloric side: "Weight loss and changes in nutritional status, common consequences of GLP-1 receptor agonist therapy, are established risk factors for telogen effluvium." Rapid weight loss and low iron or zinc status are long-documented triggers for diffuse shedding, and the diagnoses captured were non-scarring — the reversible category. The stated limitations are the usual ones for records-based work: no data on severity, extent, duration or reversibility, and no ability to exclude unmeasured confounding.
Two mechanisms, and why it matters which one you have
Put side by side, the papers describe different problems that happen to share a headline.
Telogen effluvium is a stress response. A rapid caloric deficit pushes a large fraction of follicles out of the growth phase at once; the shedding shows up two to four months later, is diffuse rather than patterned, and typically resolves once weight and nutrition stabilise. This is what the BMJ cohort's non-scarring diagnoses and the authors' own mechanistic reasoning describe.
Androgenetic alopecia is patterned and progressive. It follows a temporal-recession-and-crown route, does not resolve on its own when the trigger goes away, and is the condition the September genetic analysis points at. A 7% risk increment on a genetically predisposed background is not a large number, but it is a different kind of number — it describes acceleration of an existing trajectory, not a temporary shed.
The practical consequence is that the two produce opposite expectations about time. One resolves; one does not. Anyone reading their own shedding through the July study's lens will expect regrowth; anyone whose pattern matches the September study's outcome will not get it by waiting. Neither paper offers a way to tell which is which from the outside, and that gap is where most of the internet's confident advice is currently being generated.
What it changes for anyone running a GLP-1 protocol
Not the dose, and not the compound. Nothing in either paper compares one GLP-1 against another for hair outcomes, and nothing supports switching molecules on this basis — the BMJ analysis treated the class as a class, and the genetic analysis is about the receptor common to all of them.
What it does change is what is worth tracking. The nutritional pathway is the one with a documented mechanism and the one that is measurable: iron status, ferritin, zinc and protein intake are all things that move during aggressive weight loss and all things with established links to diffuse shedding. Bloodwork answers questions here that a forum thread cannot. The same question comes up around lean mass on a GLP-1 and the bone-density signal reported earlier this year. In each case authors have raised the rate and magnitude of weight loss as a candidate explanation alongside the drug itself; no published work has tested that as a shared mechanism across all three outcomes.
