My phone wouldn’t stop buzzing in the weeks leading up to the 2026 American Society for Clinical Oncology annual meeting. Every major outlet, from NPR to Substack, was screaming the same thing: Ozempic might prevent cancer. The headlines were loud. The excitement was palpable.
But as a physician and clinical epidemiologist who spends my life designing these studies, I have to hit the brakes.
The idea that GLP-1 agonists are a cancer-prevention panacea is running far ahead of the proof. It isn’t wrong—yet. But it isn’t earned, either.
Cancer isn’t a single target. It’s more than a hundred distinct diseases with wildly different biology. A drug rarely gets a simple verdict on cancer. It might raise risks for some, lower them for others, and leave most untouched. The reality is messy.
The Early Fears: Thyroid and Pancreas
Before the current hype, the fear was that these drugs caused cancer.
Specifically, thyroid C-cell tumors showed up in rodent studies. That scare led U.S. regulators to add a black box warning in June 2 humans are different animals than rats. Our thyroid cells have far fewer GLP-1 receptors, making them far less sensitive. Monkey studies didn’t show the same tumor growth. By 2023, European regulators had already reviewed the data and found no clear link between GLP-1s and thyroid cancer. A 2025 analysis of 93 trials backed that up.
Pancreatic cancer was another early worry. A 2025 review of 62 studies found no consistent rise in risk there either.
The reassurance feels solid, but it’s provisional. These drugs are relatively young. Cancer often takes decades to surface. We might be looking at the data before the full picture has developed.
The Narrative Flips
Now, the story has inverted. We’re moving from fear of causation to hopes for prevention.
A massive 2024 study tracked over 1.6 million people with type 2 diabetes. It found lower rates for 10 out of 13 “obesity-related” cancers among those on GLP-1 therapy compared to insulin users.
Another 2025 study, involving roughly 87,000 adults, showed an overall cancer risk drop of about 17%. The clearest wins were in endometrial, ovarian, and meningioma brain cancer cases. Interestingly, kidney cancer risk spiked by 38% in this group. That finding needs larger studies to confirm if it’s real or noise.
Then there was breast cancer. A 2026 study of 110,000+ women undergoing imaging found a 30% lower risk for those on GLP-1 drugs versus those who weren’t.
Survival data looks promising, too. In a 2024 colon cancer study, 16% of patients on GLP-1s died within five years, compared to 37% not on them. At the 2026 ASCO meeting, a report claimed a 34% lower risk across six cancer types.
Exciting? Yes. Conclusive? No.
Why the Data is Deceptively Simple
Observational studies are tricky beasts. They have three major blind spots that make GLP-1 benefits look bigger than they probably are.
1. The Healthy User Bias
People who get these prescriptions aren’t a random slice of the population. They tend to be healthier, wealthier, and more engaged with healthcare. If you have insurance, a steady income, and regular doctor visits, you’re likely on Ozempic. Someone with the same BMI but without those advantages probably isn’t.
Those advantages—better diet, less stress, earlier detection of issues—lower cancer risk. The drug might just be getting credit for the lifestyle that got the patient the drug in the first place. It is incredibly hard to scrub this bias out of real-world data.
2. The Comparison Problem
How you measure a drug changes the result. The 2024 diabetes study compared GLP-1 users to those on insulin. Insulin is usually reserved for advanced diabetes, a condition itself linked to higher cancer risk. So, of course, the GLP-1 group looked better. They weren’t being compared to healthy people; they were being compared to sicker people.
When researchers compared GLP-1 users against those taking metformin—a diabetes drug for less severe cases—no clear cancer risk reduction showed up. The benefit wasn’t necessarily the GLP-1; it was the lower-risk group it was measured against.
3. The Timing Issue
Cancer prevention doesn’t happen overnight. If a drug truly changed cellular biology to stop tumors, you’d see a gradual decline over years. But in these studies, the risk drop happens almost immediately.
That speed is suspicious. It suggests the patients already had lower risk to begin with. The GLP-1 isn’t creating a miracle; it’s riding a wave of pre-existing health advantages.
The Randomized Trial Reality
The only way to truly isolate the drug’s effect is through randomized controlled trials (RCTs). These are the gold standard because they balance the groups before the drug even touches them.
So, what do the RCTs say?
Quietly.
Two major 2025 meta-anyses—one covering 50 trials, the other 48—found little to no evidence that GLP-1s change overall cancer risk. They don’t seem to raise it, and they don’t clearly lower it.
Why the discrepancy with the observational data? Observational studies often have larger sample sizes but more bias. RCTs are cleaner but often short-term. The trials included in those 2025 analyses usually followed patients for only a year or two. Not enough time for cancer to develop or disappear.
We need tens of thousands of patients followed for years to answer the question properly. Until then, we’re guessing.
The Bottom Line
Let’s be clear about what we do know. GLP-1 drugs like Ozempic do not appear to increase the overall risk of cancer. That is reassuring. It means we don’t need to panic.
But the more ambitious claims—that they actively prevent cancer or significantly boost survival—remain unproven hypotheses.
Even if they do help, we don’t know why. Is it the weight loss? Improved metabolic function? A direct hit on inflammation? We don’t know.
Most of this research also comes from high-income countries. It’s risky to assume these results apply globally, especially where cancer burdens are highest and healthcare access differs.
The surest fact is also the least satisfying: It is way too soon. These drugs are young. The cancers they are being credited with stopping are old, slow-moving, and complex.
Patience isn’t just a virtue in science. It’s a requirement.
