Does a GLP-1 Cause Cancer? The Medullary Thyroid Carcinoma Boxed Warning, the Pancreatic Cancer Signal, and What the Human Evidence Actually Shows

Cancer is the single most common patient-anxiety question about GLP-1s, driven mostly by the FDA black-box warning for medullary thyroid carcinoma (MTC) on the front page of the Wegovy^[1] and Zepbound^[2] labels. That warning is real, comes from rodent studies, and has not been demonstrated in humans. The pancreatic cancer signal has been investigated in multiple large cohorts and found to be null or inconclusive. The 2023 BMJ-published French national cohort generated a fresh thyroid signal that the EMA and FDA both reviewed and did not find sufficient to change labeling. This article walks through the actual evidence — the rodent data, the human cohorts, the regulatory reviews, and the contraindications that genuinely apply.

The boxed warning, in plain language

The Wegovy^[1] and Zepbound^[2] labels carry the FDA's most serious warning category — the “boxed warning” (informally, the “black-box warning”). The wording on both labels is essentially identical:

• Semaglutide and tirzepatide cause thyroid C-cell tumors at clinically relevant exposures in rats. It is unknown whether these drugs cause thyroid C-cell tumors, including medullary thyroid carcinoma (MTC), in humans, as the human relevance of the rodent finding has not been determined.
• The drugs are contraindicated in patients with a personal or family history of MTC and in patients with multiple endocrine neoplasia syndrome type 2 (MEN2).
• Patients should be counseled regarding the potential risk and the symptoms of thyroid tumors (a mass in the neck, dysphagia, dyspnea, persistent hoarseness).

The boxed warning is not the same as the FDA saying “this drug causes cancer in humans.” It is the FDA saying “there's a rodent signal we cannot rule out in humans, and we are taking the precautionary position of contraindicating the drug in patients who are already at high baseline MTC risk.”

The rodent finding

The rodent C-cell tumor finding is real and well-documented. Bjerre Knudsen and colleagues^[3] showed in 2010 that liraglutide (the prototype long-acting GLP-1 receptor agonist) activates GLP-1 receptors on rodent thyroid C-cells and produces calcitonin release, C-cell hyperplasia, and eventually C-cell adenomas and carcinomas at supraphysiologic doses in rats and mice. The same finding has been replicated for semaglutide and tirzepatide in pre-approval rodent toxicology studies, which is why both labels carry the same boxed warning.

Why rodent ≠ human (the scientific argument for skepticism)

Rodent C-cells express GLP-1 receptors at much higher densities than human C-cells, and the receptor signaling pathway that drives the rodent proliferation appears to be species-specific. The functional differences are large:

• C-cell density. Rodent thyroids contain dramatically more C-cells per unit tissue than human thyroids.
• GLP-1 receptor expression. Human C-cells express GLP-1 receptors at low levels or not at all, depending on the assay used. Rodent C-cells express them at high levels.
• Calcitonin response. In humans on long-term GLP-1 therapy, serum calcitonin (the marker that would suggest C-cell activation) does not rise consistently. In rats it rises sharply.

These mechanistic differences are the basis for the argument that the rodent tumor finding is not directly relevant to humans. They are the reason the FDA did not outright contraindicate GLP-1s for the general population — the warning is restricted to patients who already carry a genetic predisposition to MTC.

The human evidence — what the cohorts actually show

LEADER (2016)

The LEADER trial^[6] was a 9,340-patient, 3.8-year cardiovascular outcomes trial of liraglutide vs placebo in type 2 diabetes. As one of the longest and largest GLP-1 trials, it's been examined for cancer signals. There was no statistically significant excess of thyroid cancer or pancreatic cancer in the liraglutide arm compared with placebo. The follow-up was 3.8 years, which is short for solid-tumor latency, but it's the largest randomized dataset we have.

The 2023 BMJ-published French cohort signal^[4]

Bezin and colleagues used the French national health data system (SNDS) to compare thyroid cancer incidence in 2,562 patients on a GLP-1 receptor agonist with matched diabetic controls on other drugs. They found a statistically significant elevated risk of thyroid cancer at 1-3 years of cumulative exposure (adjusted hazard ratio approximately 1.58 for any thyroid cancer, 1.78 for medullary thyroid cancer specifically). This is the strongest human signal published to date.

The signal generated significant attention and triggered formal regulatory reviews at both the EMA and FDA. Limitations the authors and subsequent commentators identified:

• Detection bias. Patients on GLP-1s see their physicians more often and may get more thyroid imaging — “more looking, more finding.”
• Reverse causation / latency. Many of the thyroid cancers diagnosed during follow-up may have been present subclinically at the time of GLP-1 initiation.
• Confounding by indication. Patients eligible for GLP-1s differ in many ways from patients on other diabetes drugs.
• Single national dataset. The signal has not been replicated in subsequent cohorts.

The EMA PRAC review (2023-2024)^[5]

The European Medicines Agency's Pharmacovigilance Risk Assessment Committee (PRAC) reviewed the Bezin signal alongside additional cohort and trial data. PRAC concluded in October 2023 that a causal association between GLP-1 receptor agonists and thyroid cancer has not been established and that the available data do not support a change to the existing product information. The FDA conducted a parallel review and reached the same conclusion.

This is the current regulatory position: the rodent signal triggers the boxed warning and the MEN2/MTC contraindication, but the human evidence does not currently support a broader causal claim.

The 2024 real-world data analysis^[8]

Wang and Kim published a 2024 real-world data analysis comparing cancer incidence across multiple GLP-1 cohorts and found no consistent excess thyroid or pancreatic cancer signal across drugs. Risk ratios for thyroid cancer were not statistically elevated, and a small numeric signal for pancreatic cancer was not robust to sensitivity analysis.

Pancreatic cancer — the older signal that didn't pan out

In the early 2010s, a series of FAERS (FDA Adverse Event Reporting System) analyses suggested an association between exenatide (the first GLP-1 receptor agonist) and acute pancreatitis and pancreatic cancer. This generated substantial concern at the time. Subsequent prospective cohort studies and meta-analyses, including Pinto and colleagues 2019^[7], did not confirm the pancreatic cancer signal. The pancreatitis signal is real and is in the labels, but the pancreatic cancer signal is now considered not to have been substantiated.

Other cancers

Breast cancer was investigated in a 2018 cohort study by Funch and colleagues^[9] with no signal found. Other cancers (gastric, liver, kidney, colon) have been examined in subset analyses of the major trials and the large cohort databases without consistent positive findings.

What this means for actual patients

The honest summary of the current evidence:

• If you have a personal or family history of MTC, or if you have MEN2, do NOT take a GLP-1 receptor agonist. The contraindication is real and is on every label.
• If you have a personal or family history of non-medullary thyroid cancer (papillary, follicular), discuss with your prescriber and your endocrinologist. The black-box warning specifically refers to medullary thyroid cancer, not other thyroid cancers, but the 2023 BMJ signal does not distinguish subtypes well, so caution is reasonable.
• If you have a thyroid nodule or persistent neck mass, get it worked up before starting a GLP-1. The label specifically counsels patients about neck masses and difficulty swallowing as warning signs.
• If you have a personal history of pancreatic cancer, the GLP-1 cancer evidence is not the concern (no signal). The pancreatitis warning is the concern and you should discuss with your oncology team.
• If you have no thyroid history, no MEN2, and no first-degree relative with MTC, the current regulatory position is that the residual cancer risk — if any — is small enough that the benefits of weight loss for an eligible patient outweigh it. This is a decision to make with your prescriber.

Symptoms to watch for

The FDA labels specifically counsel patients to report:

• A new lump or mass in the neck (especially in the front, near the thyroid)
• Persistent hoarseness
• Difficulty swallowing
• Difficulty breathing

These symptoms are not specific to thyroid cancer — most will turn out to be benign — but they are the symptoms the labels ask you to flag to your prescriber if they appear while on a GLP-1.

Routine monitoring while on a GLP-1

The FDA labels do not recommend routine monitoring of serum calcitonin or routine thyroid ultrasound in patients on GLP-1 therapy. The rationale is that the positive predictive value of these tests in an asymptomatic population is low, and the rate of false positives would produce more harm (unnecessary biopsies, surgeries) than the baseline cancer rate would justify. If you are at higher baseline risk (family history of thyroid disease, irradiated neck, etc.), individualized monitoring decisions should be made with your prescriber and an endocrinologist.

What this is NOT

This article does not say GLP-1s are risk-free. It does not say the boxed warning should be ignored. It does not say the 2023 BMJ signal is wrong — only that the regulatory agencies that reviewed it did not find it sufficient to change labeling. New data may emerge. Long-term human cohorts are still accumulating, and the longer the follow-up, the better the answer will get.

Bottom line

• The FDA boxed warning for GLP-1s and medullary thyroid carcinoma comes from rodent studies. The human equivalent has not been demonstrated.
• GLP-1s are formally contraindicated in patients with personal or family history of MTC and in MEN2.
• The 2023 French BMJ-published cohort signal generated a fresh thyroid cancer concern, but the EMA and FDA both reviewed it in 2023-2024 and concluded that no causal association has been established.
• The pancreatic cancer signal from the early 2010s has not been confirmed in subsequent cohorts.
• Routine calcitonin or thyroid ultrasound monitoring is not recommended in low-risk patients on GLP-1 therapy.
• New neck mass, persistent hoarseness, dysphagia, or dyspnea should be reported to your prescriber promptly.

Related research and tools

• 17 GLP-1 side effect questions answered — every common patient concern with the trial-data context
• GLP-1 side effects: what the trials actually showed — the full registration trial AE table breakdown
• GLP-1s, pregnancy, PCOS, and women's health — the obstetric and reproductive safety review
• GLP-1 drug interaction checker — search any medication for interaction with your GLP-1
• SELECT trial cardiovascular benefits in non-diabetics — the largest non-diabetic GLP-1 outcome trial to date
• FLOW trial: semaglutide in kidney disease — another major outcomes trial with cancer adjudication

Important disclaimer. This article is educational and does not constitute medical advice. The decision to start, continue, or stop a GLP-1 in any patient with a personal or family history of cancer should be made with the prescribing clinician and, where relevant, the oncology and endocrinology teams. Weight Loss Rankings does not provide medical advice, diagnosis, or treatment recommendations. If you experience a new neck mass, persistent hoarseness, difficulty swallowing, or difficulty breathing while on a GLP-1, contact your prescriber promptly.