Is Ozempic Causing Cancer? What the Evidence Shows

Current evidence does not show that Ozempic causes cancer. Across dozens of clinical trials involving tens of thousands of patients, the drug’s active ingredient, semaglutide, has not been linked to a clear increase in any specific malignancy in humans. Some observational data even suggest GLP-1 receptor agonists may lower the risk of several obesity-related cancers. But the story is genuinely complicated: rodent studies raised legitimate thyroid concerns years ago, pharmacovigilance databases keep flagging signals, and a few recent analyses have found small upticks in overall tumor diagnoses that researchers are still working to explain.

Where the Cancer Concern Originated

The worry traces back to animal studies conducted before semaglutide and similar drugs reached the market. In rodents, GLP-1 receptor agonists caused a marked increase in a specific type of thyroid cell called C cells, which produce calcitonin. After 13 weeks of continuous exposure, wild-type mice developed C-cell hyperplasia and elevated calcitonin levels. The effect was clearly mediated through the GLP-1 receptor itself, since knockout mice lacking that receptor did not develop the same changes.¹ Those findings were concerning enough that the FDA required a boxed warning on Ozempic’s label about thyroid C-cell tumors in rodents, and the drug remains contraindicated in people with a personal or family history of medullary thyroid carcinoma.

The critical question has always been whether what happens in mice translates to humans. Rodent thyroid C cells express the GLP-1 receptor at much higher levels than human C cells do, and the signaling pathway the drugs activate in mice does not appear to operate the same way in human thyroid tissue. Still, the animal data created a reasonable basis for caution, and researchers have been tracking thyroid outcomes in every major trial since.

Thyroid Cancer Risk in Humans

A systematic review examining semaglutide specifically across 10 studies found that thyroid cancer incidence was low, with isolated cases of papillary and medullary thyroid cancer each making up less than 1% of study participants, suggesting no significant risk given the large sample sizes involved.² A broader analysis using the U.S. MarketScan insurance database found no increased thyroid cancer risk with GLP-1 receptor agonists compared with another class of diabetes drug, with a hazard ratio of 0.87 that was not statistically significant.³ And a 2025 meta-analysis pooling 48 randomized trials and over 94,000 participants concluded that GLP-1 receptor agonists probably have little or no effect on thyroid cancer risk, with moderate certainty.⁴

That said, pharmacovigilance databases tell a noisier story. An analysis of the FDA’s adverse event reporting system through early 2024 found that thyroid cancer was reported disproportionately often among users of semaglutide, liraglutide, dulaglutide, and tirzepatide compared with other medications.⁵ European pharmacovigilance data showed similar reporting imbalances, with the strongest signals for liraglutide and exenatide.⁶ These databases capture voluntary reports from doctors and patients, and they cannot prove causation. A drug in the news is simply reported more often when something goes wrong. But they do help regulators decide where to look more closely, and they are part of why the thyroid question has not been fully closed.

Cancers Where Risk Appears Lower

If the thyroid data are ambiguous, the colorectal cancer data lean in a more encouraging direction. A study following over 1.2 million patients with type 2 diabetes over 15 years found that GLP-1 receptor agonist users had a substantially lower colorectal cancer risk compared with people on insulin, metformin, and several other diabetes medications. Among patients who were also overweight or obese, the reduction was even larger.⁷ A network meta-analysis of 17 randomized trials reinforced this, finding a roughly two-thirds reduction in colon cancer specifically among GLP-1 receptor agonist users.⁸

Breast cancer incidence also appears lower. A large matched analysis found that GLP-1 exposure was associated with about a 30% reduction in breast cancer odds among women.⁹ A study of over 900,000 overweight or obese adults with type 2 diabetes reported reduced risk for a composite of 13 obesity-related cancer types, with the strongest reductions for breast, colorectal, uterine, and ovarian cancer.¹⁰ Separately, a JAMA Oncology study in adults with obesity found statistically significant reductions in endometrial cancer, ovarian cancer, and meningioma among GLP-1 receptor agonist users.¹¹

Gastric and esophageal cancers show a mixed but generally favorable pattern. A national analysis found that patients with type 2 diabetes on GLP-1 receptor agonists had lower rates of both gastric and esophageal cancer at seven years compared with non-users.¹² A separate case-control study found that GLP-1 receptor agonist use was associated with a reduced risk of esophageal squamous cell carcinoma, though not of esophageal adenocarcinoma.¹³

Findings That Point the Other Direction

Not every study paints a reassuring picture. A meta-analysis focused specifically on semaglutide in patients with type 2 diabetes across 19 randomized trials found a modest but statistically significant increase in overall neoplasms, including benign tumors. For malignant tumors alone, the increase fell just short of statistical significance.¹⁴ That borderline result is hard to interpret cleanly, but it is not nothing.

A presentation at a major diabetes conference compared GLP-1 receptor agonist users head-to-head with users of three other drug classes and found a roughly 12-15% higher rate of new cancer diagnoses among GLP-1 receptor agonist users, with the increase driven mainly by melanoma, male reproductive cancers, and female breast cancers.¹⁵ And the JAMA Oncology study in adults with obesity, while finding reductions in several cancers, noted a borderline trend toward increased kidney cancer risk among GLP-1 receptor agonist users.¹¹

Within the GLP-1 class, individual drugs may not behave identically. A large network meta-analysis of over 270,000 trial participants examined blood cancers and found that dulaglutide was associated with roughly double the risk of overall hematologic malignancy. Tirzepatide, a newer dual agonist, showed the opposite pattern, with a significantly lower risk.¹⁶ Lumping all GLP-1 receptor agonists together may obscure real differences between individual drugs.

Why the Numbers Are So Hard to Read

Two major problems make it genuinely difficult to say whether these drugs are raising or lowering cancer risk: the populations taking them already have elevated cancer risk, and the medical attention that comes with a new prescription can itself generate more diagnoses.

Obesity and type 2 diabetes are both independently associated with higher cancer rates.¹⁷ In men with type 2 diabetes, each additional 5 kg/m² of body mass index raises the risk of obesity-related cancers; in women, the same pattern holds, particularly for sex-specific cancers like uterine and ovarian.¹⁸ That means the comparison group matters enormously. If you compare GLP-1 receptor agonist users against the general population, they will look like they get more cancer, because they were already at higher risk before they ever filled the prescription. Observational studies try to control for this by comparing against users of other diabetes or obesity drugs, but residual confounding is hard to eliminate completely.

Surveillance bias is the other thorn. People who start Ozempic typically enter a period of more frequent doctor visits, bloodwork, and health screenings. Researchers have documented this directly: one study found that GLP-1 receptor agonist users had higher mammography screening rates than controls, and that the apparent increase in breast cancer diagnoses disappeared among women who already had a history of regular mammograms.¹⁹ In other words, the drug may not have caused more cancer; it may just have led to more cancer being found. A recent review summarized the problem plainly: large observational studies generally associate GLP-1 receptor agonists with lower obesity-related cancer incidence, but these findings remain vulnerable to confounding, comparator selection, and surveillance bias.²⁰

Pancreatic Cancer, Specifically

Pancreatic cancer deserves its own mention because it was an early and persistent worry. Rodent studies raised concerns about chronic pancreatitis and pancreatic ductal adenocarcinoma in animals given GLP-1-based therapies, but those findings have not been consistently replicated in human observational studies or clinical trials, and the number of confirmed pancreatic cancer events in trials remains small.²¹ The 2025 meta-analysis of 48 trials found no meaningful signal for pancreatic cancer, with moderate certainty.⁴ This is one area where the gap between animal data and human data is wide enough that most researchers consider the initial alarm to have been overstated, though long-term monitoring continues.

What the Biology Suggests

If GLP-1 receptor agonists turn out to genuinely reduce some cancer risks, researchers think several mechanisms could be involved beyond simple weight loss. In people with obesity, GLP-1 therapy has been shown to restore the function of natural killer cells, a type of immune cell that patrols the body for abnormal cells. Treated patients showed improved natural killer cell cytotoxicity and increases in a metabolic pathway critical for the cells’ cancer-fighting ability.²² Broader reviews have described how GLP-1 receptor agonists influence several immune cell types and aspects of the tumor microenvironment, including tumor metabolism, cell cycle progression, and how tumor tissue remodels itself.²³

An intriguing mouse study tested what happens when a GLP-1-based therapy is stopped partway through treatment. After withdrawal, the metabolic benefits reversed quickly: body weight rebounded, fat mass returned, and food intake increased. But the tumor suppression was more durable. Mice that had been treated and then taken off the drug still showed delayed tumor onset and significantly smaller tumors compared with untreated controls.²⁴ If that finding holds up in further research, it would suggest the anti-cancer effects involve something beyond keeping weight off, perhaps lasting immune changes or shifts in the tumor environment that persist even after the drug is gone.

The Gallbladder and Bile Duct Question

GLP-1 receptor agonists are known to slow gallbladder emptying, which raises theoretical concerns about gallbladder disease and bile duct cancers. A population-based cohort study examined the risk of cholangiocarcinoma among patients using GLP-1 receptor agonists and found an elevated point estimate compared with other second- or third-line diabetes drugs, but the confidence interval was wide and crossed the line of no effect.²⁵ This is too imprecise to draw conclusions from, but it is an area researchers have flagged for continued attention, especially as more patients use these drugs for longer periods.

Compounded Semaglutide and Safety Unknowns

A separate but related concern involves compounded versions of semaglutide sold outside the brand-name supply chain. An analysis of semaglutide products purchased from online sellers without a prescription found major quality problems. Measured semaglutide purity ranged from about 8% to 14%, a dramatic departure from the 99% purity claimed on product labels. The actual semaglutide content exceeded labeled amounts by roughly 29-39%, and endotoxin, a bacterial contaminant that can trigger inflammation, was detected in all samples tested.²⁶ A separate laboratory comparison of follow-on injectable semaglutide products found new impurities and impurity patterns not present in the originator product, including high molecular weight proteins, trace metals, and residual solvents.²⁷

None of this research specifically links compounded semaglutide to cancer. But when patients take a product with unknown impurities, inconsistent dosing, and bacterial contaminants, the long-term safety profile becomes impossible to evaluate. The cancer safety data collected in clinical trials applies to pharmaceutical-grade semaglutide made under strict manufacturing controls. People using gray-market or compounded products are essentially in uncharted territory, and whatever reassurance the clinical trial data provides does not extend to those formulations.

How Long Is Long Enough to Know

Many cancers take a decade or more to develop from the earliest cellular changes to a detectable tumor. Semaglutide received FDA approval for type 2 diabetes in 2017 and for weight management in 2021. The longest randomized trials included in current meta-analyses ran about three years. That is a meaningful amount of follow-up for catching fast-growing cancers or early signals, but it is genuinely too short to rule out a slow-developing risk that might only become visible after 10 or 15 years of widespread use.

This is not a unique problem. The same limitation applied to statins, hormone replacement therapy, and many other widely prescribed drugs in their early years. Post-marketing surveillance, long-term registry studies, and insurance claims analyses will keep accumulating data. The 2025 meta-analysis of 48 trials represents the best snapshot available right now, and its conclusions are cautiously reassuring for most cancer types, but researchers are explicit that longer follow-up is needed.⁴ The honest answer is that the data are good enough to say Ozempic is unlikely to be a potent carcinogen, but not mature enough to close the book entirely.

Differences Between Individual Drugs in the Class

Public conversation tends to treat “Ozempic” and “GLP-1 drugs” as interchangeable, but the cancer data suggest the individual agents within this class may not carry identical risk profiles. Dulaglutide’s elevated hematologic malignancy signal stands out against tirzepatide’s protective signal, and pharmacovigilance reporting ratios differ substantially between semaglutide, liraglutide, and exenatide for thyroid cancer.⁵ Some of these differences probably reflect how long a drug has been on the market and how many people take it, both of which inflate reporting ratios. But some may reflect real pharmacological differences in receptor binding, duration of action, or tissue distribution. As the field matures, expect cancer safety data to be reported agent by agent rather than as a class average, which should clarify whether any individual drug stands out as genuinely riskier than the rest.