Do GLP-1 Drugs Cause Cancer? The Boxed Warning and 94,245 Patients

The boxed warning on every one of these drugs is about rodents, and it says so itself: whether the effect applies to humans “has not been determined.”^[2] In 2026 the Annals of Internal Medicine pooled 48 randomized placebo-controlled trials covering 94,245 participants and found no meaningful signal for thyroid, pancreatic, breast or kidney cancer at moderate certainty.^[1] That is the most reassuring evidence available, and the authors are the first to point out why it cannot close the question: none of those trials was built to measure cancer, and cancer takes longer to appear than they ran.

What the boxed warning actually says

It is worth reading rather than summarizing, because the summary people carry around is more alarming than the text.

In rodents, semaglutide causes dose-dependent and treatment-duration-dependent thyroid C-cell tumors at clinically relevant exposures. It is unknown whether WEGOVY causes thyroid C-cell tumors, including medullary thyroid carcinoma (MTC), in humans as human relevance of semaglutide-induced rodent thyroid C-cell tumors has not been determined.

— WEGOVY prescribing information, boxed warning (revised 06/2026)

Two claims, and they differ in kind. The first is an established animal finding, stated confidently. The second concedes that nobody knows whether it carries across to people. Tirzepatide’s label runs nearly word for word, naming rats where this one names rodents.^[3]

Section 5.1 adds the only human evidence there is. After liraglutide reached the market, cases of medullary thyroid carcinoma were reported in people taking it — and the label states plainly that those reports can neither establish a causal link nor rule one out.^[2] Both halves of that sentence matter.

What 48 trials found

The 2026 review searched five databases through August 2025 for randomized placebo-controlled trials reporting any of thirteen cancer outcomes, assessed each for risk of bias, and graded the certainty of every result.^[1] It is the most careful look at this question so far, and it reports absolute risk alongside the ratios, which is the part that makes it usable.

Every one of those intervals crosses 1, which means the data are consistent with no effect in either direction. The thyroid figure is the one people fix on, and 1.37 does sound like something — but the interval runs from 0.82 to 2.31, and in absolute terms the honest statement is somewhere between one case fewer and nine cases more per ten thousand people treated. The review's own wording is that these drugs “probably have little or no effect” on each of the four.^[1]

For colorectal, esophageal, liver, gallbladder, ovarian and endometrial cancer, along with multiple myeloma and meningioma, the finding was the same direction at low certainty. For gastric cancer the evidence was too thin to say anything, and the review says so rather than reporting a number.^[1]

The results held up when the analysis was restricted to trials at low risk of bias, when it was restricted to semaglutide or tirzepatide, and across subgroups split by follow-up length, population, drug class, dose and duration.^[1] Consistency across that many cuts is worth something.

Why this does not settle it

The authors put the limitation in their own conclusion: the trials pooled here were not designed to evaluate cancer, and their follow-up was short.^[1]

That is a serious caveat rather than a formality. Solid tumors generally take years to become detectable. A drug trial running one to two years, counting cancers that happened to be reported as adverse events, is not the instrument that would find a cancer risk emerging over a decade. What this analysis rules out is a large, fast effect. A small or slow one would look exactly like this.

Who should not take these at all

The contraindication is not a caution to weigh, and it is the part of this page to act on. Two histories rule these drugs out altogether: Multiple Endocrine Neoplasia syndrome type 2, and medullary thyroid carcinoma — whether the diagnosis was yours or a blood relative’s.^[2][3] That second half catches people out. A grandparent’s diagnosis counts, and it is worth asking your family before starting rather than assuming you would already have heard.

The label also sets out what should send you to a clinician once you are taking one: a lump in the neck, trouble swallowing, breathlessness, or a hoarse voice that will not clear.^[2]

The screening question

The natural response to a thyroid warning is to want a thyroid scan. The label addresses this directly and the answer surprises most people: routine monitoring of serum calcitonin, or routine thyroid ultrasound, is of uncertain value for detecting medullary thyroid carcinoma early in patients on these drugs.^[2]

That is the label declining to recommend a test, which labels do not do lightly. Screening that cannot be shown to help still produces false positives, biopsies and anxiety, and it is not a substitute for the contraindication doing its job. If you want the scan anyway, that conversation belongs with your prescriber, who can weigh your own history. What we can tell you is that the label does not present it as the answer.

For the risks in this class that are established rather than uncertain, our gallbladder piece covers the one with the clearest signal, and the side-effect timeline sets out the common effects against the rare ones for each drug.