A large database study found that obese adults without diabetes who took GLP-1 drugs developed fewer obesity-linked cancers than similar adults who were told to diet and exercise. The protection showed up almost everywhere the authors looked. There was one exception they flagged plainly: it did not appear in black patients.
The study, published online June 7 in Annals of Oncology ↗, drew on TriNetX, a network of health records covering 113 million US patients. The authors ran what is called a target trial emulation, a way of using real-world records to imitate a randomized trial. They took obese, nondiabetic adults with no prior cancer, found the ones prescribed a GLP-1 receptor agonist, and matched each to a similar patient who got diet or exercise counseling instead. After matching, 161,798 patients remained, split evenly between the two groups, with a median of two years of follow-up.
Across that cohort, GLP-1 users had a hazard ratio of 0.59 for developing any of 13 obesity-associated cancers. In plain terms, they were about 41 percent less likely to be diagnosed with one of those cancers during the study window. The 13 are the cancers epidemiologists tie to excess body fat: colorectal, postmenopausal breast, endometrial, kidney, liver, pancreatic, and others on the same list.
Where it held and where it did not
The subgroup breakdown is where the study earns attention. The authors split the cohort by sex, by body mass index (under 40 versus 40 and above), by race (white versus black), and by drug, comparing semaglutide ↗, the molecule in Ozempic and Wegovy, against tirzepatide ↗, the one in Mounjaro and Zepbound. Every subgroup showed the same lower cancer incidence, except black patients, where the association did not hold. A second statistical method, inverse probability weighting, returned the same overall result.
The authors do not claim to know why the signal skipped that group. A null result in one subgroup can come from a real biological difference, from differences in screening and access, or simply from too few cancer cases in that slice of the cohort to measure anything. The honest reading is that the headline protection is an average that did not distribute evenly, and the group where it failed to show is one already underserved by cancer screening and priced out of these drugs more often. That is worth naming rather than burying in a supplement.
The platform thread
Both drugs in the subgroup analysis sit on peptidemodel against the GLP-1 receptor ↗: semaglutide and tirzepatide are the two best-studied agonists at that target. The platform covered a narrower version of this question last week, when a separate cohort found about 30 percent fewer breast cancers ↗ among women on these drugs. This study widens the lens to 13 cancer types and to a population the diabetes and heart trials never enrolled: people who are obese but whose blood sugar is still normal.
The limits are the ones that come with every records study. Two years is a short window for cancer, which can take a decade to surface, so this captures early diagnoses, not lifetime risk. Matching on records cannot rule out that people who get prescribed these drugs differ from people who get told to exercise in ways the data never recorded. The authors say as much, calling for prospective trials to test whether the drugs cause the drop or merely travel with it. For now the signal is large, consistent, and pointed at exactly the cancers obesity drives. The exception is the part to remember.