Safety review · YMYL
GLP-1 and thyroid cancer (MTC): the honest evidence read
Every GLP-1 receptor agonist carries an FDA black-box warning for medullary thyroid carcinoma (MTC). Search “Ozempic thyroid cancer” and you will find a mix of fear-driven copy and dismissive reassurance. This is the honest read of what the rodent studies actually showed, what human data through mid-2026 says, and who should still avoid these drugs.
The evidence stack, one source at a time
| Source | Type | Finding |
|---|---|---|
| Rat 2-year carcinogenicity studyWegovy FDA label | Preclinical 2005-2020 | Dose- and duration-dependent thyroid C-cell hyperplasia, adenomas, and carcinomas in rats at exposures relevant to the human therapeutic dose. Basis for the FDA boxed warning. |
| LEADER 5-year follow-up (liraglutide)Marso NEJM 2016 | Randomized trial 2016 + extension | N=9,340 T2D patients on liraglutide vs placebo, 3.8 years median. No increase in thyroid cancer (any type). MTC not increased. |
| SELECT trial (semaglutide 2.4 mg)Lincoff NEJM 2023 | Randomized trial 2023 | N=17,604 non-diabetic adults with CV disease, mean 3.3 years. Total cancer incidence did not differ between arms. No MTC signal. |
| France ANSM cohort (Bezin et al.)Bezin Diabetes Care 2022 | Retrospective cohort 2022 | French SNDS national database. Initial analysis suggested elevated thyroid cancer risk (HR ~1.6) with GLP-1 use. Widely re-analyzed; detection bias was likely explanation. |
| Skov Jensen et al. — Danish nationwide cohortSkov Jensen BMJ 2024 | Retrospective cohort 2024 | N ~50,000 GLP-1-exposed vs matched controls. After adjusting for detection frequency, no increased incidence of thyroid cancer of any type. Contradicted the initial French signal. |
| FDA post-marketing surveillance (2005-2026)FAERS dashboard | Passive surveillance Cumulative | 15+ years of exenatide, liraglutide, dulaglutide, semaglutide, and tirzepatide post-marketing has not led to an FDA labeling change beyond the original rodent-based warning. FAERS shows reports but no established causal signal. |
Why the rat signal did not translate
The rat carcinogenicity finding is real. The question that mattered from 2005 onward (when exenatide launched) was whether it would translate to humans. Fifteen-plus years of the drug class in millions of patients has answered: not yet, and probably not.
Two mechanistic reasons commonly proposed:
- C-cell density. Rat thyroid C-cell density is roughly 40-100× higher than human thyroid C-cell density. The tumor substrate that exists in a rat thyroid is not present in a human thyroid at comparable density.
- Receptor expression. Rat C-cells express GLP-1 receptors at higher density and with greater proliferative response than human C-cells. When GLP-1 receptors are stimulated in a rat C-cell, the pro-proliferative signaling is robust; in human C-cells, it is muted.
Combined, these differences reduce the biological plausibility of the rat-to-human extrapolation. This is not proof that the rodent finding is irrelevant — it explains why 15 years of human exposure has not reproduced the signal.
The 2022 French signal and its re-analysis
In 2022, Bezin et al. published a cohort analysis in Diabetes Care using the French SNDS national health database. Initial finding: a hazard ratio around 1.6 for thyroid cancer in GLP-1-exposed patients. This launched a wave of media coverage.
Subsequent re-analyses identified detection bias as the dominant explanation. GLP-1 patients get more routine bloodwork, more imaging, and more clinical encounters — which surfaces incidental thyroid nodules that were already there. When the Danish nationwide cohort (Skov Jensen et al., BMJ 2024) adjusted for detection frequency, the signal disappeared. The Danish study also showed no increase in medullary thyroid carcinoma specifically.
Regulatory response: the FDA did not update the boxed warning in response to Bezin 2022. The ANSM (French regulator) issued a caution letter in 2023 but did not modify drug labeling.
Who should still avoid GLP-1s
The FDA black-box warning identifies two subgroups where GLP-1 use is contraindicated. These remain valid regardless of the reassuring human data — the risk-benefit calculation is different when your baseline MTC risk is already elevated.
- Personal history of MTC. If you have previously been diagnosed with medullary thyroid carcinoma, GLP-1 receptor agonists are absolutely contraindicated. This applies even after successful treatment.
- Family history of MTC or known MEN2. First-degree relative with MTC, or a genetic test showing a pathogenic RET mutation, is a contraindication. If you are uncertain about family history, ask about it — MTC is rare and would have been a memorable diagnosis in a family member.
If neither of these applies to you, the general-population evidence supports GLP-1 initiation without routine thyroid screening. AACE 2024 and ADA 2024 both take this position.
What we still need
The current data is reassuring but has real limitations. Follow-up in most human trials is 3-5 years; MTC has a long latency. Longer-term observational data will accumulate through 2030 and beyond as the first large cohorts of chronic semaglutide + tirzepatide users reach 10+ year exposure.
If a genuine human MTC signal exists at very long durations, it will show up in that data. So far, no signal has emerged. This review will be refreshed as new evidence publishes.
Frequently asked
Why does Wegovy have a black-box warning for thyroid cancer?
The Wegovy, Ozempic, Rybelsus, Mounjaro, and Zepbound boxed warnings are all based on the same underlying data: in 2-year rat carcinogenicity studies, semaglutide and liraglutide caused dose-dependent thyroid C-cell tumors (including medullary thyroid carcinoma) at exposures relevant to the human dose. The FDA labels this a precautionary contraindication in patients with personal or family history of MTC or Multiple Endocrine Neoplasia syndrome type 2 (MEN2). It does not mean the drugs cause MTC in humans — human evidence is what section 8 of this review covers.
Do humans actually get thyroid cancer from Ozempic?
As of mid-2026, human data does not show a signal. The SELECT trial (17,604 patients, 3.3 years), LEADER 5-year follow-up, the SUSTAIN/STEP programs, and multiple large EHR cohort studies have all failed to demonstrate an increased incidence of medullary or papillary thyroid cancer above the general-population background. The Ministère de la Santé française (France ANSM) published a widely-discussed analysis in 2022-2023 that initially suggested a signal, but subsequent larger-cohort re-analyses (Skov Jensen et al., Zhou et al.) attributed most of the signal to detection bias — GLP-1 patients get more routine bloodwork and imaging.
Should I get a thyroid screen before starting Ozempic?
Routine thyroid ultrasound or calcitonin screening is NOT recommended for the general population starting a GLP-1, per AACE 2024 guidelines and the ADA 2024 Standards of Care. Screening is warranted only in patients with personal history of MTC, family history of MTC, or known MEN2 — for whom the drugs are contraindicated regardless of screening result.
What is MEN2 and how do I know if I have it?
Multiple Endocrine Neoplasia syndrome type 2 (MEN2) is a rare inherited genetic syndrome (RET proto-oncogene mutation) that predisposes to medullary thyroid carcinoma and other endocrine tumors. Prevalence is ~1 in 30,000. If a first-degree relative has had MTC or a genetic test has identified a RET mutation, you have MEN2. If nobody in your family has had MTC, your risk of MEN2 is effectively zero and no additional screening before GLP-1 initiation is needed.
What about non-medullary thyroid cancer (papillary, follicular)?
The FDA warnings specifically cover medullary thyroid carcinoma (MTC), the C-cell tumor seen in rat studies. Papillary and follicular thyroid cancers (much more common types) are not part of the black-box warning. Human data has not shown an increase in papillary or follicular thyroid cancer with GLP-1 use in any large cohort. Some observational studies have found elevated papillary detection rates, again largely attributable to detection bias.
Why has the rat signal not translated to humans?
Two reasons commonly proposed. First, rat thyroid C-cell density is roughly 40-100x higher than human thyroid C-cell density — the target-tissue substrate for the tumor differs substantially. Second, rat GLP-1 receptor expression on C-cells is much higher and more responsive than in human C-cells. The rodent tumors are real; the extrapolation to humans is what has failed to materialize in 15+ years of post-marketing surveillance across the drug class.
Sources
Editorial notes
- Freshness: reviewed July 17, 2026. Refresh triggered by any new large-cohort MTC study, FDA labeling change, or regulatory position update.
- Not medical advice. Individual decisions about GLP-1 initiation, especially in patients with any thyroid history, are for a prescriber who has assessed the patient.
- Citation: journalists and researchers are welcome to cite — attribution to Glpverdict is the only requirement.