Do GLP-1 drugs cause thyroid cancer? FDA warning vs human data

Every box of Ozempic, Wegovy, Mounjaro, and Zepbound carries the same scary line at the very top of its FDA label: a boxed warning about thyroid tumors. That warning is real, and it is required by law. But it comes from studies in rats and mice, not people, and the human data collected over the past decade tells a calmer story. This guide walks through where the warning came from, what large studies in actual patients have found, and how to read the gap between the two.

Why there is a thyroid warning at all

GLP-1 drugs copy a gut hormone called glucagon-like peptide-1. The hormone tells your pancreas to release insulin when blood sugar rises, slows how fast your stomach empties, and quiets appetite signals in the brain. That is why these drugs lower blood sugar and drive weight loss.

The thyroid worry comes from one specific cell type. Your thyroid gland contains "C-cells," also called parafollicular cells. These cells make a hormone called calcitonin, and they carry receptors that GLP-1 can bind to. When a rare cancer grows out of these C-cells, it is called medullary thyroid carcinoma (MTC). MTC is different from the far more common types of thyroid cancer (papillary and follicular), which start in different cells and make up the large majority of thyroid cancer cases.

So the question on every label boils down to this: if you give animals or people a drug that stimulates C-cell receptors for years, do those C-cells turn cancerous?

It is worth pausing on how rare MTC is to begin with. Medullary thyroid carcinoma makes up only about 1 to 2 percent of all thyroid cancers, and thyroid cancer itself is uncommon. So even if a drug nudged the risk slightly, the absolute number of extra cases would be small. That math matters when you read headlines. A "50 percent increase" sounds huge, but 50 percent more of a very small number is still a very small number. Keeping the baseline rate in mind is the single best defense against being scared by a percentage with no context attached.

What happened in the rodent studies

When liraglutide (the first long-acting GLP-1 drug, sold as Victoza and Saxenda) went through animal testing, rats and mice given the drug for two years developed thyroid C-cell tumors. The effect was dose-dependent and duration-dependent, meaning higher doses and longer exposure produced more tumors. Some of those tumors were the malignant MTC type.

Because of that finding, the FDA placed a boxed warning on liraglutide and then required the same warning on every long-acting GLP-1 drug approved afterward, including semaglutide (Ozempic, Wegovy) and tirzepatide (Mounjaro, Zepbound). The label also makes these drugs contraindicated (meaning do not use) in anyone with a personal or family history of MTC, or with a genetic syndrome called Multiple Endocrine Neoplasia type 2 (MEN 2), which raises baseline MTC risk on its own.

The FDA is careful with its wording. The Wegovy label states plainly that it is unknown whether semaglutide causes thyroid C-cell tumors, including MTC, in humans, because the human relevance of the rodent finding has not been determined (Wegovy FDA label, 2025). That single sentence is the heart of the whole debate. The warning flags a signal from animals; it does not claim the cancer happens in people.

The rat-versus-human problem

There is a biological reason scientists doubt the rodent finding transfers cleanly to humans. Rodent thyroid C-cells are studded with GLP-1 receptors. Human C-cells have far fewer of them, and a large share of human MTC tumors show little or no GLP-1 receptor activity at all. A drug can only directly act on a tissue that carries its receptor. If the human target barely exists, the rodent mechanism may simply not apply.

This is a known pattern in toxicology. Rats are unusually prone to certain hormone-driven tumors that never show up in human users of the same drugs. So the rodent C-cell signal is treated as a flag worth watching, not as proof of human harm.

There is also a difference between how rats and humans clear these tumors. In the rodent studies, the C-cell changes started as benign growths (hyperplasia) and progressed slowly over the animal's whole two-year lifespan with constant high drug levels. Humans are not exposed the same way: doses are lower relative to body size, and most people are not on the drug for their entire life. None of this proves humans are safe. It just explains why scientists treat the leap from "rats got tumors" to "people will get tumors" as a hypothesis to test, not a settled fact.

What the human evidence actually shows

Animal data starts the conversation. Human data finishes it. Over the last few years, researchers have studied very large groups of real patients to see whether GLP-1 users get thyroid cancer more often than people on other diabetes drugs. The short version: the big, careful studies have not found a convincing increase.

Here is the human evidence laid out side by side.

Study (year)	Design	Population	Main finding	Confidence in result
Bezin et al., Diabetes Care (2023)	Nested case-control, French national database	Type 2 diabetes, 2nd-line drugs	Higher thyroid cancer with 1-3 yr use (HR ~1.58); also higher MTC	Weaker — single country, screening/detection bias likely
Pottegård/multisite, Thyroid (2025)	Cohort across 6 countries	~92,500 GLP-1 users vs ~2.5M DPP-4i users	No increased risk (pooled HR 0.81, 95% CI 0.59-1.12)	Stronger — largest, multi-country, dose-response checked
Morales et al., Diabetes Care (2025)	Retrospective cohort	~460,000 GLP-1 users vs SGLT2i/DPP-4i/SU	No increased risk (all HRs near or below 1.0)	Stronger — large, multiple comparison drugs, lag analysis
Silverii et al., Diabetes Obes Metab (2025)	Meta-analysis of randomized trials	Pooled RCT participants	No increase in overall cancer, but a thyroid cancer signal (OR ~1.55) was detected	Moderate — gold-standard design, but rare event, short follow-up

A few things stand out when you read this table honestly.

The one study that found a signal — and why it is shaky

The 2023 French study by Bezin and colleagues is the study that worried people. It reported that 1 to 3 years of GLP-1 use was tied to roughly a 58% higher rate of all thyroid cancer and a higher rate of MTC specifically (Bezin et al., Diabetes Care 2023).

That sounds alarming. But the finding has serious weak spots, and even the authors and commentators flagged them:

It was a single country with one health database. Findings that do not repeat elsewhere are treated cautiously.
Detection bias is the big one. People starting a new diabetes drug often get more medical attention, more lab work, and more imaging. Thyroid nodules are extremely common and usually harmless, but the more you scan, the more cancers you find that would never have caused trouble. So extra screening, not extra cancer, can fake a signal.
The number of MTC cases was tiny, which makes the estimate unstable.

When other large studies tried to reproduce it, the signal did not hold up.

The studies that found nothing — and why they carry more weight

The 2025 international cohort study, led by researchers including Anton Pottegård, pooled six population-based databases from Canada, Denmark, Norway, South Korea, Sweden, and Taiwan. It compared about 92,500 GLP-1 users against roughly 2.5 million users of DPP-4 inhibitors (another diabetes drug class used as a fair comparison). The pooled result showed no increased risk of thyroid cancer (hazard ratio 0.81, 95% CI 0.59 to 1.12), and no rising risk as the cumulative GLP-1 dose went up (multisite cohort, Thyroid 2025). A hazard ratio under 1.0 actually points slightly the other way, though the wide confidence interval means the most honest reading is simply "no clear effect."

A second large 2025 retrospective cohort by Morales and colleagues looked at over 460,000 GLP-1 users and compared them against three different diabetes drug classes. Across all comparisons, GLP-1 exposure was not associated with higher thyroid tumor risk, and the result held even after a one-year lag analysis designed to filter out early detection bias (Morales et al., Diabetes Care 2025).

Pulling the randomized trials together, a 2025 meta-analysis of 50 trials found no rise in overall cancer risk, but it did flag a statistically significant thyroid cancer signal (odds ratio 1.55, 95% CI 1.05 to 2.27), more evident in longer trials (Silverii et al., Diabetes Obes Metab 2025). Randomized trials are the cleanest design because the drug is assigned by chance, which removes the bias problems that haunt database studies. But the thyroid result rests on only a handful of cases — thyroid cancer is so rare that even pooling thousands of trial participants gives very few events, so the estimate has wide error bars and the authors themselves called for further specific studies rather than treating it as settled.

Looking at cancer more broadly, a 2025 analysis in JAMA Oncology studied GLP-1 use in adults with obesity and did not find the drugs raising overall cancer risk; if anything, weight loss itself is tied to lower rates of several obesity-related cancers (JAMA Oncology 2025). That wider lens matters. Obesity is a known driver of multiple cancers, so a drug that treats obesity could plausibly lower some cancer risks even while it carries a thyroid warning. The full cancer picture is not one-directional.

The honest catch: follow-up is still short

Here is where this guide will not oversell the reassurance. Thyroid cancer, and MTC in particular, can take many years to develop. The big reassuring cohort studies have average follow-up of only about 1.8 to 3 years. That is long enough to rule out a fast, dramatic effect. It is not long enough to rule out a small risk that only shows up after a decade of use.

The American Thyroid Association said as much when it covered the multisite study: the data offer "some reassurance" about short-term safety, while noting the follow-up is too short to settle long-term risk (American Thyroid Association). That is the accurate place to land. Not "proven safe forever." Closer to "no signal so far, and we keep watching."

Sorting fact from fear

It helps to separate what is established from what is hype.

Claim	Status	What the evidence says
Rats given GLP-1 drugs got thyroid C-cell tumors	True	Reproduced finding; basis for the FDA boxed warning
The FDA put a boxed warning on these drugs	True	Class-wide; also contraindicated with MTC/MEN 2 history
GLP-1 drugs are proven to cause thyroid cancer in people	Not established	FDA says human relevance "has not been determined"
One human study found a thyroid cancer signal	True but weak	French 2023 study; likely detection bias, not reproduced
Large multi-country studies found no increased risk	True	2025 cohorts; short follow-up is the main limit
Long-term (10+ year) risk is fully ruled out	False	Follow-up so far is only ~2-3 years

The takeaway is not "the warning is fake." The warning is a legitimate, evidence-based caution about an animal finding that no one has yet been able to confirm or fully dismiss in humans. The takeaway is that the loudest version of the fear — that these drugs are giving people thyroid cancer right now — is not supported by the human data we have.

Who should be extra careful

Even with reassuring population data, some people genuinely should avoid GLP-1 drugs or talk through the risk carefully first.

Personal or family history of medullary thyroid carcinoma (MTC). This is a hard stop. The FDA label lists it as a contraindication.
Multiple Endocrine Neoplasia type 2 (MEN 2). This inherited syndrome raises MTC risk on its own and is also a contraindication.
A history of other thyroid cancers or unexplained thyroid nodules. Not an automatic no, but a reason to involve an endocrinologist before starting.
A high blood calcitonin level. Calcitonin is the hormone C-cells make, and a high level can be an early MTC sign worth sorting out first.

If none of those apply to you, your individual risk based on current evidence appears very low. Deciding whether GLP-1 therapy fits your situation is part of a broader eligibility picture covered in our guide to who is a good candidate for GLP-1 medications.

Symptoms that deserve a doctor's look

Routine thyroid cancer screening (like ultrasound or calcitonin checks) is not recommended for everyone on these drugs, partly because over-screening creates the exact detection-bias problem described above. But you should get checked if you notice:

A lump or swelling in the front of your neck
Trouble swallowing
A hoarse voice that does not go away
Shortness of breath that is new and unexplained

These can have many harmless causes, but a thyroid that is changing shape is worth a professional look.

The calcitonin testing debate

Calcitonin is the hormone C-cells make, and a high blood level can be an early sign of MTC. So why not test everyone before and during GLP-1 treatment? Because routine calcitonin screening in the general population causes more problems than it solves. Mild elevations are common and usually have nothing to do with cancer. They come from kidney issues, certain other medications, smoking, and even just lab-to-lab variation. A borderline-high result can launch a chain of biopsies and worry that almost always ends in "nothing was wrong."

For that reason, major guidelines do not recommend blanket calcitonin testing for people starting GLP-1 drugs. Testing makes sense when there is a real reason to look: a suspicious nodule, a family history, or symptoms. This is the same logic that keeps the population-wide thyroid ultrasound off the table. More testing of low-risk people finds more harmless findings, and those findings drive the very detection bias that muddied the French study in the first place.

How the thyroid worry compares to other GLP-1 risks

Thyroid cancer gets the boxed warning and the headlines, but in day-to-day use it is one of the least likely problems a patient will actually face. The far more common issues are nausea, vomiting, and constipation, which most people feel in the first weeks. Rarer but more serious concerns include pancreatitis and gallbladder problems.

Putting the thyroid question in context next to the side effects people actually experience helps with perspective. Our complete guide to GLP-1 side effects walks through the full list and how often each one shows up. For two of the risks that, like thyroid cancer, get more attention than their frequency warrants, see our reviews of the real evidence on pancreatitis risk and the study on rare eye/vision risk. The pattern repeats: a real but rare signal, often amplified beyond what the numbers support. For the broader rundown, see our overview of GLP-1 side effects and risks.

If the thyroid warning rules you out: alternatives

If you have an MTC or MEN 2 history, GLP-1 drugs are off the table, but you still have options for weight and blood sugar. None of these carry the C-cell tumor warning, because none of them stimulate the GLP-1 pathway.

Option	Type	Best for	Notes
SGLT2 inhibitors (e.g. empagliflozin)	Diabetes pill	Type 2 diabetes, heart/kidney protection	Modest weight loss; no thyroid warning
Metformin	Diabetes pill	First-line diabetes, mild weight benefit	Decades of safety data; very cheap
Bariatric surgery	Procedure	Higher BMI, durable weight loss	Most effective long-term; surgical risks
Structured lifestyle program	Diet/exercise	Anyone	Foundation for all other treatments
Phentermine-based options	Appetite suppressant	Short-term weight loss	Different risk profile; not for everyone

These are not perfect swaps. GLP-1 drugs are unusually effective, and nothing on this list matches their weight-loss numbers head to head except surgery. But for the small group the thyroid warning genuinely excludes, an honest plan with a doctor beats forcing a contraindicated drug. The right choice depends on your blood sugar, your weight goals, your heart and kidney health, and your tolerance for side effects.

What this means for you

If you do not have an MTC or MEN 2 history, the current human evidence does not show that GLP-1 drugs raise your thyroid cancer risk. The boxed warning reflects a real rodent finding whose human relevance is still unproven, and the largest patient studies to date have come up empty for a thyroid signal. The one human study that found a signal is weak and has not been reproduced.

The honest caveat is time. We have only a few years of careful follow-up, and thyroid cancer can be slow. That is a reason for ongoing monitoring of the data, not a reason for panic today. If you have a thyroid history or a family history of MTC, that is a real reason to choose a different treatment, and your doctor can help you weigh it.

Frequently Asked Questions

Do GLP-1 drugs like Ozempic and Wegovy cause thyroid cancer?

No human study has proven they do. The FDA boxed warning is based on thyroid C-cell tumors in rats and mice, and the FDA itself states it is unknown whether semaglutide causes these tumors in humans. The largest multi-country patient studies in 2025 found no increased thyroid cancer risk, though follow-up is still only a few years.

Why is there an FDA boxed warning if humans probably are not at risk?

The warning is required because rodents reliably developed C-cell tumors at clinically relevant doses, and that is a serious enough animal signal to disclose. The FDA flags the signal and notes the human relevance has not been determined. It is a precaution, not a finding of proven human harm.

Who absolutely should not take a GLP-1 drug because of thyroid risk?

Anyone with a personal or family history of medullary thyroid carcinoma (MTC), or with Multiple Endocrine Neoplasia type 2 (MEN 2). The FDA lists both as contraindications. People with unexplained thyroid nodules or high calcitonin levels should also talk to an endocrinologist first.

Did any human study actually find a thyroid cancer link?

Yes, a 2023 French database study reported a higher rate of thyroid cancer with 1 to 3 years of use. But it was a single country, likely affected by detection bias from extra screening, and larger international studies have not reproduced it, so its result is considered weak.

Should I get my thyroid screened while on a GLP-1 drug?

Routine screening is not recommended for everyone, because over-screening finds harmless nodules and can create false alarms. But see a doctor promptly if you notice a neck lump, trouble swallowing, a lasting hoarse voice, or new shortness of breath.

This article is for general information only and is not medical advice. Talk to a licensed healthcare provider before starting, stopping, or changing any medication.