Does Semaglutide or Tirzepatide Cause Cancer? The Thyroid Warning, Explained

A boxed warning is the FDA's strongest label warning — reserved for risks the agency judges serious enough to print at the very top of the prescribing information, above everything else. Both semaglutide (Ozempic, Wegovy) and tirzepatide (Mounjaro, Zepbound) carry one, for thyroid C-cell tumor risk, and it's the reason "does GLP-1 cause cancer" and "does tirzepatide cause cancer" are among the most-searched questions about this drug class. Our tirzepatide and semaglutide side-effect pages each mention this warning in one sentence. Here's the full picture: what the rodent studies actually found, why the warning exists despite no confirmed human cases, and what real-world human data has found so far.

What the boxed warning actually says

Wegovy's label states it directly: "In rodents, semaglutide causes thyroid C-cell tumors at clinically relevant exposures. It is unknown whether WEGOVY causes thyroid C-cell tumors, including medullary thyroid carcinoma (MTC), in humans as the human relevance of semaglutide-induced rodent thyroid C-cell tumors has not been determined."¹ Zepbound's label makes the identical statement about tirzepatide, in rats: "In rats, tirzepatide causes thyroid C-cell tumors... It is unknown whether ZEPBOUND causes thyroid C-cell tumors, including medullary thyroid carcinoma (MTC), in humans."² Read that closely — the warning itself states it does not know whether this happens in humans. That's unusual; most boxed warnings describe a confirmed human risk. This one describes an animal finding the FDA judged serious enough to require its strongest label warning anyway, precisely because the uncertainty is the point: the signal couldn't be ruled out, so it gets flagged at maximum visibility instead of buried in a subsection.

The rodent studies behind the warning

The underlying data is specific enough to look at directly rather than take on faith. In a 2-year carcinogenicity study in CD-1 mice, semaglutide produced a statistically significant increase in thyroid C-cell adenomas — plus a numerical increase in carcinomas — in both males and females at every dose level tested, including doses at clinically relevant human exposure¹. A parallel 2-year study in Sprague Dawley rats found the same pattern: statistically significant increases in C-cell adenomas at all doses, and C-cell carcinomas at higher doses, again at exposures relevant to human dosing¹. Tirzepatide's own 2-year rat carcinogenicity study found essentially the same signal — statistically significant increases in thyroid C-cell adenomas, and adenomas-plus-carcinomas combined, in both male and female rats at every dose tested². One detail worth noting: a separate 6-month study in rasH2 transgenic mice — a different model commonly used to screen for tumor-forming potential — found tirzepatide was not tumorigenic in that strain². That doesn't erase the rat finding, but it's a reminder the signal isn't uniform across every animal model tested. Both labels are explicit that whether any of this translates to human risk "has not been determined."

Why this drives the MEN 2 and personal/family MTC screening question

This animal data is exactly why the intake question about personal or family MTC history, or Multiple Endocrine Neoplasia type 2 (MEN 2), is a hard exclusion rather than a soft caution. Both drugs are flatly contraindicated — not "use with caution," but disqualifying — in patients with a personal or family history of MTC or MEN 2¹². That line is drawn from the rodent signal, not from confirmed human cases: someone with a genetic predisposition toward MTC (MEN 2 arises from inherited RET gene mutations) is exactly the population where an unconfirmed but biologically plausible risk stops being worth taking, even absent direct proof it happens in people. It's also why neither label recommends routine calcitonin blood tests or thyroid ultrasounds as ongoing monitoring for patients without that history — both state such screening is "of uncertain value" for early MTC detection in the general treated population, given calcitonin's low test specificity and how common incidental thyroid nodules already are, and warn it could drive unnecessary procedures rather than useful information¹². The screening happens once, at intake, as an eligibility gate — which is exactly why that personal/family thyroid-cancer question on an intake form isn't a formality. A provider skipping it is skipping the one point where this specific risk is actually supposed to be managed.

What the human evidence actually shows — and it's genuinely mixed

Here the honest answer is more complicated than either "no risk" or "confirmed risk." Two large observational studies looked directly at this question in humans and reached different conclusions. A Scandinavian cohort study spanning Denmark, Norway and Sweden followed more than 145,000 GLP-1 users against nearly 292,000 DPP-4 inhibitor users over a mean 3.9 years and found no significantly increased risk of thyroid cancer overall (hazard ratio 0.93, 95% CI 0.66–1.31) or medullary thyroid cancer specifically (hazard ratio 1.19, 95% CI 0.37–3.86)⁴. A separate French nested case-control study using national health insurance data reached the opposite conclusion: GLP-1 use for 1–3 years was associated with an increased risk of thyroid cancer overall (adjusted HR 1.58, 95% CI 1.27–1.95) and specifically medullary thyroid cancer (adjusted HR 1.78, 95% CI 1.04–3.05)³. Those two well-conducted studies genuinely disagree. A 2024 narrative review synthesizing the full evidence base — animal data, randomized trial data, observational cohorts, and postmarketing pharmacovigilance reports — reached a measured verdict: randomized trial data shows thyroid cancer is a rare event with imprecise effect estimates, observational studies are inconsistent with one another, and pharmacovigilance databases do show an elevated reporting signal, but overall there is "no conclusive evidence of elevated thyroid cancer risk" in humans⁵. Separately, both drug labels note that cases of MTC have been reported in patients taking liraglutide, an older GLP-1, during postmarketing surveillance — but state plainly that those reports don't contain enough information to establish or rule out a causal relationship¹.

What this means before you buy

The rodent-tumor finding is real, specific, and the documented basis for the boxed warning — that part isn't in dispute. What isn't established is whether it happens in humans at all: both labels say so explicitly, and the best human evidence available is genuinely split between one large cohort finding no meaningful increase and another finding one, with a synthesizing review calling the overall picture inconclusive rather than confirmed. If you have a personal or family history of MTC or MEN 2, this isn't a medication to work around that screening question for — it's a hard exclusion grounded in real, if unconfirmed-in-humans, biology, and no legitimate provider should let you skip past it. If you don't have that history, the evidence as it stands doesn't show a settled causal link, though it hasn't ruled one out with certainty either, which is why postmarketing monitoring continues. This is a genuine part of clinical screening, not a box to check past on the way to a lower price — see our methodology for how real intake review factors into how we score providers. For the other label warnings covered in similar depth, see GLP-1s, pancreatitis and gallbladder risk; for the broader adverse-event picture on each molecule, see tirzepatide side effects and semaglutide side effects. Not medical advice — see our medical disclaimer and talk to your prescriber about your own personal and family history before starting either medication.