Obesity is linked to at least 13 types of cancer, and some researchers now expect it to overtake tobacco as the leading modifiable cancer risk factor. So when GLP-1 medications started showing up in cancer data, oncologists paid attention.
Last week, Healio pulled together the strongest set of those signals so far: five studies presented at the 2026 ASCO Annual Meeting, plus news that researchers are organizing a prospective study to test the idea properly. The numbers are striking. The researchers who produced them are also unusually direct about what they cannot prove. Both halves matter.
What the studies found
All five are retrospective analyses of health records. Four used the TriNetX research network; one used Penn Medicine’s own records. Each compared people with a GLP-1 prescription against matched people without one.
- Breast cancer (Penn Medicine). Among 111,646 women aged 45 to 80 with a BMI of 25 or above who had breast imaging between 2022 and mid-2025, those with a GLP-1 prescription had about 30% lower odds of a breast cancer diagnosis in the matched analysis: 1.62% versus 2.31%.
- Breast cancer (high-risk women). In a separate cohort of women with a family history, genetic predisposition, or dense breasts, the difference was smaller: 2.95% versus 3.01%, a hazard ratio of 0.84.
- Lung cancer. In 34,187 matched pairs at elevated risk from tobacco, lung disease, or occupational exposure, the GLP-1 group had a 31% lower relative risk. In absolute terms, that was 2.0% versus 2.4%.
- Liver cancer. In about 296,000 matched pairs with hepatitis, cirrhosis, or alcohol use disorder, hepatocellular carcinoma was 73% less common in the GLP-1 group.
- Leukemia. In about 313,000 matched pairs at elevated risk, acute myeloid leukemia was 51% less common among GLP-1 users.
Five cancers, five studies, all pointing the same direction. It’s easy to see why the headline was “time to act.”
What these studies do not establish
They show association, not cause. One of the lead investigators put it plainly in the Healio piece: the signal is as strong as it has ever been, but it is still an association. Records can match people on age, BMI, and diabetes status. They cannot match them on everything. People who get a GLP-1 prescription tend to be more engaged with the healthcare system, better insured, and judged suitable for the medication by a clinician. Any of those could lower measured cancer rates on its own.
Relative and absolute numbers tell different stories. “31% lower lung cancer risk” and “2.0% versus 2.4%” describe the same result. The second is the more useful one for an individual. In the high-risk breast cancer cohort, the absolute difference was six-hundredths of a percentage point.
Exposure was often short. In the leukemia study, average GLP-1 treatment lasted roughly three months. Cancer develops over years. A large effect from a short exposure is a reason for curiosity, and also a reason to suspect that something other than the drug is part of the picture.
Screening can distort cancer data. People who see clinicians more often get screened more often. That can find cancers earlier, or it can shift when a diagnosis lands inside a study window. It doesn’t always push the numbers in the flattering direction, but it adds noise that retrospective data can’t fully remove.
The Penn study didn’t separate drugs or doses. It didn’t look at which GLP-1, how long someone took it, cancer stage, or cancer subtype. “GLP-1 users” is a broad bucket.
The metformin lesson
The researchers themselves raised a cautionary example. About a decade ago, observational studies suggested metformin, a common diabetes drug, might prevent cancer. The signal was consistent and biologically plausible. When prospective trials tested it, the benefit largely failed to appear.
That doesn’t mean GLP-1s will follow the same path. It means consistent observational data has been wrong before, and the only way to find out is a study designed to test the question directly.
What INSPIRE is meant to answer
That’s the proposed INSPIRE study, led by the Penn team behind the breast cancer analysis. It would sit inside ECOG-ACRIN’s TMIST breast screening trial, which has enrolled more than 108,000 women, and prospectively follow breast cancer incidence in higher-risk women, including survivors, along with cardiovascular outcomes and blood and tissue biomarkers. The researchers say it could begin within about six months of being fully funded.
A prospective design sets the question and the measurements before outcomes are known, which removes a lot of the guesswork that record-based studies carry. It will still take years to produce an answer. Until then, “GLP-1s may lower cancer risk” is a hypothesis worth testing, not a reason to start a medication.
The part that applies to compounded medications
Every study above drew on prescription records for branded, manufactured GLP-1 products. None of them studied compounded preparations, and INSPIRE won’t either.
Compounded semaglutide and tirzepatide, the medications NoBsRx provides provider-guided access to, are prepared by state-licensed 503A pharmacies for an individual patient. They are not FDA-approved, and no cancer data of any kind exists for them. Findings about branded products do not transfer to a compounded preparation. If anyone markets a GLP-1 to you on cancer prevention, they’re selling past the evidence for the branded drugs, never mind the compounded ones.
What this means in practice
For now, this research is a reason for continued study, not a factor in whether someone should use a GLP-1. That decision rests on the things that are actually known: your health history, your goals, and a licensed provider’s review of whether treatment is appropriate at all. If you have a personal or family history of cancer, that belongs in the conversation with the clinician managing your care, and the questions we get most often are a reasonable place to start preparing for it.
This post is general information, not medical advice. It does not describe a benefit of any NoBsRx program and it is not a claim that any medication prevents or treats cancer. Compounded medications are not FDA-approved. Treatment is never guaranteed. Eligibility and every treatment decision are made by independent licensed providers based on an individual clinical assessment, and availability varies by state. Do not start, stop, or change any treatment, including cancer screening, based on an article. If you have concerns about your cancer risk, talk to a licensed clinician.