GLP-1 Agonists and Thyroid Cancer Risk
Semaglutide Long-Term Safety
No increased thyroid cancer risk in largest retrospective cohort study to date
GLP-1 receptor agonists carry an FDA boxed warning for thyroid C-cell tumors based on rodent data. The largest human studies published in 2025-2026 found no increased risk of thyroid malignancy.
Morales et al., Diabetes Care, 2025
Morales et al., Diabetes Care, 2025
View as imageEvery GLP-1 receptor agonist sold in the United States carries an FDA boxed warning about thyroid C-cell tumors. The warning exists because liraglutide, semaglutide, and other GLP-1RAs caused dose-dependent C-cell hyperplasia, adenomas, and medullary thyroid carcinomas (MTC) in rats and mice exposed to high doses for two years. The drugs are contraindicated in patients with a personal or family history of MTC or multiple endocrine neoplasia syndrome type 2 (MEN2).[1] For the broader safety picture of these drugs, see the pillar article on semaglutide long-term safety.
The question that the boxed warning cannot answer is whether this rodent signal translates to humans. After more than 15 years of clinical use and millions of treated patients, the human evidence is now substantial enough to evaluate. The answer is more nuanced than either "GLP-1 drugs cause thyroid cancer" or "they're completely safe."
Key Takeaways
- The largest retrospective cohort study (Morales et al., Diabetes Care, 2025) found no increased risk of thyroid tumors with GLP-1RA use compared to SGLT2 inhibitors, DPP-4 inhibitors, or sulfonylureas across international databases
- A meta-analysis of 50 randomized controlled trials found no significant difference in overall cancer risk with GLP-1RA treatment, including no thyroid cancer signal (Silverii et al., Diabetes Obesity and Metabolism, 2025)
- GLP-1 receptors are expressed on nearly all rodent thyroid C-cells but detected in only about 27% of human C-cell neoplasms, a species difference that likely explains why the rodent signal has not translated to humans (Kelly et al., JCEM, 2025)
- In 18 patients with papillary thyroid carcinoma on active surveillance, GLP-1RA exposure did not accelerate tumor growth compared to matched controls (Patrizio et al., Journal of the Endocrine Society, 2026)
- Semaglutide actually suppressed papillary thyroid carcinoma growth in mouse xenograft models by reprogramming tumor-associated macrophages toward anti-tumor phenotypes (Wang et al., JCEM, 2025)
- Among US Medicare beneficiaries with diabetes, incretin-based therapies showed no increased thyroid cancer incidence compared to SGLT-2 inhibitors (Acheampong et al., BMJ Open Diabetes Research and Care, 2025)
The Rodent Signal: What Actually Happened
The concern originated from mandatory two-year rodent carcinogenicity studies required by regulatory agencies before drug approval. In these studies, rats and mice were exposed to liraglutide at doses up to 8-fold higher than the maximum recommended human dose (adjusted for body surface area) for their entire lifespan equivalent of two years.
The results were consistent and dose-dependent: rodents developed C-cell hyperplasia (enlargement of calcitonin-producing cells in the thyroid), C-cell adenomas, and in some cases medullary thyroid carcinomas. The effect was seen across multiple GLP-1RAs, confirming it as a class effect rather than a drug-specific toxicity. The mechanism was straightforward: rodent thyroid C-cells express GLP-1 receptors at high density, and sustained activation of these receptors drives C-cell proliferation through cAMP-dependent pathways.
Kelly et al. (2025) explained why this finding prompted serious regulatory concern: medullary thyroid carcinoma, though rare in the general population (about 3-5% of all thyroid cancers), is aggressive and difficult to treat once advanced. A drug class that might increase MTC incidence even modestly would be clinically significant given the tens of millions of people now taking GLP-1RAs worldwide.[1]
The Species Difference: Why Rodents Are Not Humans
The critical biological question is whether human thyroid C-cells respond to GLP-1R activation the same way rodent C-cells do. The evidence strongly suggests they do not.
Rodent thyroid C-cells express GLP-1 receptors abundantly. Human C-cells express GLP-1R at much lower levels, with only about 27% of human C-cell neoplasms showing detectable GLP-1R expression.[1] This receptor density difference is not subtle: it changes the fundamental capacity of the tissue to respond to GLP-1R agonism.
Calcitonin levels provide a functional readout of C-cell activity. In rodent studies, GLP-1RA treatment elevated calcitonin dramatically. In human clinical trials across thousands of patients treated for years, GLP-1RAs have not produced clinically meaningful increases in serum calcitonin levels. This absence of calcitonin elevation in humans is the single most reassuring piece of evidence, because it indicates that human C-cells are simply not being stimulated by GLP-1RAs at therapeutic doses.
The difference likely reflects evolutionary divergence in thyroid C-cell biology. Rodent C-cells constitute a larger proportion of thyroid tissue and have higher baseline proliferative activity compared to human C-cells. Rodent C-cells also replicate more readily in response to hormonal stimulation, a property shared with many rodent cell types that makes two-year rodent carcinogenicity studies produce false-positive cancer signals for multiple drug classes.
The Human Epidemiological Evidence
Large Retrospective Cohorts
Morales et al. (2025) published the largest and most rigorous epidemiological study to date, using international administrative claims and electronic health record databases. The retrospective, active-comparator new-user cohort study compared thyroid tumor incidence in patients with type 2 diabetes initiating GLP-1RAs versus those initiating SGLT2 inhibitors, DPP-4 inhibitors, or sulfonylureas. Using propensity score matching to control for confounders, they found no increased risk of thyroid tumors or thyroid malignancy with GLP-1RA use across any comparator group.[2]
Acheampong et al. (2025) investigated the association between incretin-based therapies and thyroid cancer incidence among US Medicare beneficiaries with diabetes. Comparing GLP-1RA and DPP-4 inhibitor users to SGLT-2 inhibitor users, they found no increased thyroid cancer incidence with incretin-based therapy use. The study used thyroidectomy followed by cancer diagnosis codes as the outcome measure, reducing the risk of detection bias.[3]
Meta-Analysis of Clinical Trials
Silverii et al. (2025) conducted a meta-analysis of 50 randomized controlled trials comparing GLP-1RAs to any comparators for diabetes and obesity. GLP-1RA treatment was not associated with a significant difference in overall cancer risk (MH-OR 1.05, 95% CI 0.98-1.13). No thyroid-specific cancer signal emerged from the pooled trial data. The strength of this evidence is that it comes from randomized trials rather than observational data, reducing confounding.[4]
Active Surveillance Data
Patrizio et al. (2026) addressed a question that no previous study had: what happens when patients who already have thyroid cancer take GLP-1RAs? In a cohort of 18 patients with 19 low-risk papillary thyroid carcinomas (all 1.5 cm or smaller) undergoing active surveillance, GLP-1RA exposure was compared to matched unexposed controls. GLP-1RA-exposed tumors showed no difference in growth rate, size change, or need for intervention compared to unexposed tumors. This finding is clinically important because it suggests that even in patients with existing thyroid malignancies, GLP-1RAs do not accelerate disease progression.[5]
A Surprising Finding: Semaglutide May Suppress Thyroid Cancer
Wang et al. (2025) reported an unexpected result that reframes the thyroid cancer discussion. In human papillary thyroid carcinoma (PTC) xenograft mouse models, semaglutide actually suppressed tumor growth. The mechanism involved reprogramming of tumor-associated macrophages (TAMs) via the GLP-1R/PPARG/ACSL1 pathway, shifting TAMs from a tumor-promoting M2 phenotype toward a tumor-suppressing phenotype. In coculture systems of human THP-1 macrophages and PTC cells, semaglutide inhibited cancer cell proliferation through this immune-mediated mechanism.[6]
This study has limitations (animal model, single drug, one cancer type), but it introduces the possibility that GLP-1RAs could have anti-tumor effects on the very tissue they were suspected of harming. If confirmed in further studies, this would represent a complete reversal of the initial concern.
The Detection Bias Problem
One complication in interpreting thyroid cancer epidemiology with GLP-1RAs is detection bias. Patients starting GLP-1RAs may receive more medical attention, more blood tests, and potentially more thyroid imaging than patients on other diabetes drugs. This increased surveillance can lead to incidental discovery of thyroid nodules and cancers that would otherwise have gone undetected for years or decades. Most thyroid cancers are indolent and never cause clinical symptoms; the prevalence of subclinical papillary thyroid cancer at autopsy is 10-36%, far exceeding the clinical incidence of thyroid cancer.
Kelly et al. (2025) specifically discussed how to approach thyroid nodule management in GLP-1RA users. They recommended against routine thyroid cancer screening for GLP-1RA users, noting that the evidence does not support a causal link between GLP-1RAs and thyroid cancer in humans. Incidentally discovered thyroid nodules should be evaluated according to standard guidelines (ACR TI-RADS or ATA criteria) regardless of GLP-1RA use.[1]
The Boxed Warning: Proportionate or Excessive?
The FDA boxed warning for GLP-1RAs regarding thyroid C-cell tumors is based entirely on the rodent carcinogenicity data. No human case of medullary thyroid carcinoma has been causally linked to GLP-1RA use in any clinical trial or post-marketing surveillance system. The warning serves as a precautionary measure, reflecting the seriousness of MTC as a diagnosis and the regulatory principle that identified animal carcinogenicity signals warrant disclosure even when human relevance is uncertain.
The clinical consequence of the boxed warning is that GLP-1RAs are contraindicated in patients with personal or family history of MTC and in patients with MEN2. For the general population, the warning creates awareness but does not restrict prescribing. There is an active debate in the endocrinology community about whether the warning should be revised given the accumulating human evidence, but regulatory inertia and the precautionary principle make removal unlikely in the near term.
Other Thyroid Effects of GLP-1RAs
Beyond the cancer question, GLP-1RAs have other thyroid-related effects that are clinically relevant.
Humaida et al. (2025) reported a case of tirzepatide-associated biphasic thyroiditis in a 32-year-old woman with no prior thyroid disease. The patient developed initial thyrotoxicosis followed by hypothyroidism, consistent with drug-induced thyroiditis. While this is a single case report, it raises the possibility that dual GLP-1/GIP agonists may have immune-modulating effects on thyroid tissue that differ from selective GLP-1RAs.[7]
Mazza et al. (2025) reviewed how GLP-1 agonists influence autoimmune thyroid care, noting interactions between GLP-1RA treatment and thyroid autoimmunity that clinicians should monitor.[8]
For the parallel safety question about pancreatic effects, see do GLP-1 agonists cause pancreatitis. For the complete adverse event profile, see GLP-1 agonist adverse events. For how these drugs interact with other medications, see GLP-1 drug interactions.
The Bottom Line
The GLP-1 agonist thyroid cancer concern originated from rodent studies showing dose-dependent C-cell tumors at high doses over two years. The biological basis for this effect is high GLP-1 receptor expression on rodent thyroid C-cells, a characteristic not shared by human C-cells, which express GLP-1R at much lower levels. After 15+ years of clinical use, the human epidemiological evidence is broadly reassuring: the largest retrospective cohort studies and a meta-analysis of 50 RCTs show no increased thyroid cancer risk. A 2026 active surveillance study found no tumor acceleration in patients with existing papillary thyroid carcinoma, and preclinical data suggest semaglutide may actually suppress papillary thyroid cancer growth through immune reprogramming. The FDA boxed warning remains in effect as a precautionary measure, and GLP-1RAs remain contraindicated in patients with MTC or MEN2 family history.