The Thyroid Cancer Warning, Explained

Every product in this class carries a boxed warning — the most serious warning a US label can carry — about thyroid C-cell tumours. People read it, are frightened, and then get told “that was in rats” as though that settled it.

It does not quite settle it, and the honest picture is more interesting than either the alarm or the dismissal.

Where the warning came from

In rodent carcinogenicity studies, GLP-1 receptor agonists caused thyroid C-cell hyperplasia and medullary thyroid tumours in rats and mice, at exposures relevant to human dosing. That finding is consistent and it is why the warning exists.

C-cells are the cells that make calcitonin, and medullary thyroid carcinoma is the cancer that arises from them. It is a rare cancer, making up roughly 1–2% of all thyroid cancers, which are themselves uncommon.

Why rodents may not translate

This is the part that usually gets compressed to four words, and it deserves more.

Rodent and human thyroids differ in ways that are specifically relevant here. Rodents have substantially more C-cells as a proportion of the thyroid — C-cells make up under 1% of human thyroid tissue — and those C-cells express far more GLP-1 receptor than human C-cells do. In human thyroid tissue, GLP-1 receptor expression on C-cells is low and inconsistent, and the calcitonin response seen so clearly in rodents has not been reproduced in humans at therapeutic doses.

So there is a plausible biological reason the rodent finding might not apply. There is also no way to prove a negative, and medullary thyroid carcinoma is rare enough that detecting a small increase in risk requires very large populations and long follow-up.

That is the actual state of knowledge: a real animal signal, a plausible reason it may not translate, and human data that cannot yet fully close the question.

What the human data shows

Two studies frame the current debate.

A French cohort study published in Diabetes Care in 2023 reported an association between GLP-1 receptor agonist use and thyroid cancer, including medullary thyroid cancer, with risk appearing after 1–3 years of use.

A larger Scandinavian cohort study published in the BMJ in 2024, following patients for a mean of 3.9 years across national registries in three countries, found no substantial increased risk of thyroid cancer.

Both are observational. Both are subject to the usual confounders, and one in particular matters a great deal here: detection bias. People who start these drugs see clinicians more often, get more blood tests and more imaging, and thyroid nodules are extremely common incidental findings. Looking harder finds more.

It is also worth noting what the randomised trials found. SUSTAIN-6 followed participants for 104 weeks and LEADER for a median of 3.8 years, and neither identified a thyroid cancer signal — though both were powered for cardiovascular outcomes rather than for a rare cancer, so their reassurance is limited by that.

Where that leaves things: no established causal link in humans, some conflicting signal, and small absolute numbers on any interpretation.

The part that is not uncertain

The contraindication is clear and it is worth separating from the uncertainty above.

Do not take these drugs if you have a personal or family history of medullary thyroid carcinoma, or multiple endocrine neoplasia syndrome type 2 (MEN2).

That is an absolute contraindication on every label in the class, not a caution to weigh. Two practical notes:

• Family history means blood relatives, and it is worth actually asking. Plenty of people know a parent or aunt “had thyroid cancer” without knowing which kind, and medullary is a small minority of thyroid cancers — but it is the one that matters here, and it is frequently hereditary.
• MEN2 is a genetic syndrome that clusters with phaeochromocytoma and parathyroid disease. If that pattern exists in your family, say so.

This sits alongside the other absolute rules in who should not take a GLP-1.

What not to do

Do not get routine calcitonin testing. The FDA has noted its value is uncertain in this setting. Calcitonin is elevated in a range of benign conditions, and a mildly raised result in an asymptomatic person reliably produces a cascade of repeat tests, imaging, biopsies and months of worry — with the overwhelming majority ending in nothing.

Do not get routine thyroid ultrasound if you have no symptoms. Thyroid nodules are found in up to 60% of adults scanned for any reason, almost all are benign, and finding them mostly generates more scans.

Screening tests are only useful when they change outcomes, and neither of these has been shown to here. That is a general principle worth carrying into reading your own labs too — more testing is not the same as better care.

What to do instead

1. Establish your family history before starting — a 5 minute conversation — specifically about medullary thyroid carcinoma and MEN2. Ask what kind, if you do not know.
2. Report new symptoms if they appear: a neck lump, persistent hoarseness, difficulty swallowing, persistent breathlessness. These warrant assessment regardless of what medication you take.
3. Do not let the boxed warning make the decision alone. It is one input among several, and the conditions these drugs treat — with cardiovascular outcome data behind them — carry their own well-quantified risks that are not hypothetical.

How I put it to patients

The warning is real, it comes from a real animal finding, and there is a decent biological argument that it does not translate. Human data has not confirmed a risk and has not entirely excluded a small one. The absolute numbers, if there is any effect, are small.

If you have the contraindication, that is the end of the conversation. If you do not, this is one item on a risk-benefit list rather than a reason on its own to decline a treatment with substantial demonstrated benefits — and it is a reasonable thing to want discussed properly rather than waved away.