The largest clean test of whether GLP-1 drugs raise skin cancer risk just came back, and for melanoma the answer is no. Across roughly 390,000 matched pairs of adults with type 2 diabetes, people who started a GLP-1 receptor agonist were no more likely to develop melanoma over the next few years than people who started a different diabetes drug.
The study was published online July 28 in Diabetes, Obesity and Metabolism ↗. It was led by Huilin Tang and senior author Yong Chen at the University of Pennsylvania's Center for Health AI and Synthesis of Evidence, with skin cancer epidemiologist Jiali Han of Indiana University. It pulled electronic health records from the TriNetX network covering 2014 to 2025 and ran what researchers call a target trial emulation. That means it takes messy real-world records and forces them into the shape of a randomized trial. Define who is eligible, fix the moment treatment starts, and follow everyone forward. The comparison then behaves less like a database fishing trip and more like an experiment.
The comparison is the smart part. Measuring GLP-1 users against people on no diabetes drug at all stacks the deck, because untreated people differ in a hundred ways. So the authors instead matched GLP-1 starters against people starting two other diabetes drug classes. Those were SGLT2 inhibitors (the flozins, small-molecule pills) and DPP-4 inhibitors (the gliptins, also pills). Both comparison groups have diabetes, both are under a doctor's care, and both get their skin looked at at roughly the same rate. That removes most of the sick-user distortion that wrecks weaker observational studies.
What the numbers say
For melanoma, the aggressive skin cancer that spreads, there was nothing. Against SGLT2 inhibitors the hazard ratio was 1.04 (95% confidence interval 0.93-1.17); against DPP-4 inhibitors it was 1.09 (0.95-1.25). A hazard ratio of 1.0 means no difference, and both intervals sit squarely across it. In plain terms, GLP-1 users and their matched controls developed melanoma at statistically indistinguishable rates.
Non-melanoma skin cancer, the far more common basal-cell and squamous-cell type that rarely kills, was the one place a hint showed up. Against SGLT2 inhibitors the risk was 6 percent higher (HR 1.06), and the confidence interval just touched the line of no effect at its low end (1.00-1.11). Against DPP-4 inhibitors even that faint signal was gone (1.03, 0.97-1.09).
Then the authors did the thing that separates a careful study from a scary headline. They ran a negative control outcome calibration. The idea: pick outcomes the drug could not plausibly cause, measure the false association the data produces for those anyway, and use that residual bias to correct the real estimate. When they did, the borderline non-melanoma bump attenuated. The 6 percent was mostly leftover confounding, not the drug.
Why it matters
The worry was not idle. GLP-1 receptors turn up on some melanoma cells in lab work. And any drug now taken by tens of millions of people invites the question of whether it quietly nudges cancer risk. This is the largest and most bias-resistant look at skin cancer specifically, and it lands on reassurance rather than alarm.
It fits a pattern the section has tracked. A 100,000-pair study of US veterans ↗ found no blood cancer signal for the same drug class earlier this month. Different cancer, same design logic, same null. Two independent well-powered looks at whether these drugs cause cancer keep coming back empty.
The honest caveats are about time and detection. Mean follow-up ran only 1.8 to 2.2 years, short for a cancer that can take a decade to surface. So this rules out a fast effect, not a slow one. And non-melanoma skin cancer is so common and so surveillance-driven that a small apparent difference can be an artifact of who gets looked at. That is exactly what the calibration suggested here.
The drugs at the center of the class are semaglutide ↗, sold as Ozempic and Wegovy, and tirzepatide ↗, sold as Mounjaro and Zepbound, both peptides that act on the GLP-1 receptor ↗. The comparators, the flozins and gliptins, are small molecules with no peptide cards. For a reader weighing one of these shots, this study moves skin cancer off the list of things to lose sleep over, with the asterisk that nobody has watched long enough to close the book.