The thyroid warning attached to GLP-1 medication has three separate parts, and they are routinely collapsed into one. The warning itself comes from tumours seen in rodents. The contraindication written around it is not the same document to document, and Singapore's differs from the United States'. And the human research, which is the part that would actually settle it, disagrees with itself. This article separates the three.
Medullary thyroid carcinoma (MTC): an uncommon thyroid cancer arising from C cells, the cells that produce calcitonin. It is distinct from the far more common papillary and follicular thyroid cancers, and it is the specific cancer the animal finding concerns.
Where the warning comes from
The warning rests on animal studies, and the label says so directly.
In rodents — mice and rats for semaglutide, rats for tirzepatide — these medicines caused thyroid C-cell tumours (semaglutide, tirzepatide) in a dose- and duration-dependent way. The US prescribing information for both semaglutide and tirzepatide states that the human relevance of that rodent finding has not been determined. It appears there as a boxed warning, which is the most prominent category a US label has. The European and Singapore product information carry no boxed-warning equivalent, and record the rodent finding in the non-clinical section instead.
That difference in presentation is a difference in document structure, not in the underlying science. The same rodent data sits behind both. What it means for a reader is that the visual weight the warning carries depends on which country's document you happen to be reading — and most of what circulates online is the American one.
The contraindication is not the same everywhere
Singapore's registered product information lists one contraindication for both medicines: hypersensitivity to the active substance or to any of the excipients. Neither product lists medullary thyroid carcinoma or multiple endocrine neoplasia type 2 (NDF). The US prescribing information lists a personal or family history of those conditions as a contraindication (semaglutide, tirzepatide).
Two things follow, and the second matters more than the first.
The first is that a Singapore patient reading a US health site is reading a document that does not govern their prescription. The contraindication list they are looking at is not the one their doctor is working from.
The second is that the absence of medullary thyroid carcinoma from Singapore's section 4.3 is not clinical reassurance. Singapore's warnings-and-precautions section is not published on the public formulary portal (NDF), so what it says about the thyroid signal cannot be read from outside. An absence in the one section that is public tells you about that section, not about the medicine.
In practice this changes very little about what happens at a consultation. A personal or family history of medullary thyroid carcinoma or MEN 2 is something a doctor asks about before prescribing either way, and it is the kind of history that leads to a careful conversation rather than a quick answer. The wider set of things assessed is covered in who should not take GLP-1 medication.
What the human studies show
Five substantial analyses have looked at this in people. Two found a signal. Three did not. They are listed here in order rather than ranked, because ranking them is what produces the misleading versions.
A signal, in French national data. A nested case-control study in the French national health insurance database compared 2,562 thyroid cancer cases with 45,184 controls, all adults with type 2 diabetes. One to three years of GLP-1 receptor agonist use was associated with an adjusted hazard ratio of 1.58 (95% CI 1.27–1.95) for thyroid cancer overall, and 1.78 (95% CI 1.04–3.05) for medullary thyroid cancer specifically (Bezin 2023). It drew published disagreement in the same journal. Its design also cannot rule out detection bias — people on a drug with a thyroid warning attached tend to get their thyroid looked at more often, and looking harder finds more.
A borderline signal, in randomised trials. A meta-analysis pooled trials of at least 52 weeks, of which 26 reported any thyroid cancer event, and found an odds ratio of 1.52 (95% CI 1.01–2.29). The authors put the five-year number needed to harm at 1,349 (Silverii 2024). The lower bound of that interval is 1.01, which is as marginal as a significant result gets, and thyroid cancer is rare enough that most trials contributed no events at all.
No signal, in Scandinavian registers. An active-comparator cohort across Denmark, Norway and Sweden followed 145,410 people starting a GLP-1 receptor agonist against 291,667 starting a DPP-4 inhibitor. Thyroid cancer, hazard ratio 0.93 (95% CI 0.66–1.31). Medullary thyroid cancer, 1.19 (95% CI 0.37–3.86) (Pasternak 2024). The active-comparator design addresses the detection-bias problem directly, by comparing against people who also have diabetes and also see doctors. Mean follow-up was 3.9 years, so it says nothing about a long-latency effect.
No signal, in a very large claims cohort. A multi-database study compared 460,032 GLP-1 receptor agonist users against SGLT2 inhibitor, DPP-4 inhibitor and sulfonylurea users. None of the meta-analysed hazard ratios against any of the three comparators was significantly elevated (Diabetes Care 2025).
No signal, in the most comprehensive synthesis. An integrated analysis combined 93 trials and 101,732 participants, more than 18 million patient-years of post-marketing exposure, and a US real-world database study. Its conclusion was that the totality of the data does not suggest an association between semaglutide or liraglutide use and thyroid cancer (Vilsbøll 2026). One number inside it should not be dropped when quoting that conclusion: the all-trials estimate against placebo was a hazard ratio of 1.70, with a confidence interval of 0.99 to 3.03 — non-significant by the width of a hair.
Read together: the largest and best-designed observational studies do not find an association, the trial-level evidence is equivocal, and the two positive findings both come with structural caveats that their own authors name. Anyone presenting this as settled in either direction is going beyond what the studies support.
What to report
Whatever the epidemiology eventually shows, the practical instruction is stable and short.
Patients are counselled to report a mass in the neck, difficulty swallowing, shortness of breath, or hoarseness that persists (US Medication Guides). Those are report-it symptoms, not emergency symptoms — the appropriate response is a conversation with your doctor, not alarm. They appear alongside the more urgent items in red flags on GLP-1.
The other practical item is history. Medullary thyroid carcinoma and MEN 2 are rare, and someone who has either in their family usually knows. Say so before you start rather than after.
Individual circumstances vary, and this article is not a substitute for the product information supplied with your medicine or for your doctor's instructions.
Common questions
Does GLP-1 medication cause thyroid cancer?
The human research disagrees. A French case-control study found a raised relative risk (Bezin 2023); a three-country Scandinavian cohort found none (Pasternak 2024), as did a claims cohort of 460,032 users (Diabetes Care 2025). The question is open rather than answered.
What is the thyroid warning actually based on?
Animal studies. GLP-1 receptor agonists caused thyroid C-cell tumours in rats and mice, and the US prescribing information states the human relevance of that rodent finding has not been determined. It appears there as a boxed warning; the European and Singapore documents have no boxed-warning equivalent.
Is medullary thyroid carcinoma a contraindication in Singapore?
Not in the published section 4.3. Singapore's registered product information lists hypersensitivity to the medicine or its excipients as its contraindication, and neither product lists medullary thyroid carcinoma or MEN 2 (NDF). The warnings section is not published, so that absence is not clinical reassurance.
What thyroid symptoms should you report?
A lump or mass in the neck, difficulty swallowing, shortness of breath, or hoarseness that persists. These are the symptoms patients are counselled to report in the US Medication Guides (US Medication Guides).
Do the human studies agree with each other?
No. Two analyses found signals — a French case-control study (Bezin 2023) and a borderline randomised-trial meta-analysis, odds ratio 1.52 (95% CI 1.01–2.29) (Silverii 2024). Three larger or later analyses did not. Reporting the disagreement is more accurate than picking a side.