GLP-1 and Cancer Risk: What 10 Years of Data Actually Shows

The Rat Study That Started a Decade of Worry

If you have looked into Ozempic, Wegovy, or any other GLP-1 receptor agonist, you have probably encountered the cancer warning. It is right there in the prescribing information, in a black box, the FDA's most serious category of drug label warning. For millions of patients, that box is their first and sometimes only impression of the cancer question. The concern comes down to the thyroid, and it traces back to a specific set of experiments in rodents from the mid-1990s.

To understand why those experiments mattered, and why they may not matter the way most patients think, you need to understand the difference between rat thyroid biology and human thyroid biology. It is a difference that took a decade of human data to fully put in context.

Researchers found that GLP-1 receptor agonists could stimulate the growth of parafollicular C cells in the thyroid glands of rats. These C cells produce calcitonin, and when they proliferate unchecked, the result is medullary thyroid carcinoma. In lab dishes, a rat cell line derived from medullary thyroid carcinoma also responded to GLP-1 by ramping up proliferation. The findings were concerning enough that when GLP-1 drugs moved through the approval process, the FDA took notice.

Think of it like testing a new fertilizer on a patch of soil and finding that it makes one particular weed grow faster. That does not tell you the fertilizer will cause problems in every garden, but it gives you a reason to watch closely.

Then came the adverse event reports. The U.S. Food and Drug Administration maintains a voluntary reporting system called FAERS, and studies mining this database flagged a potential association between GLP-1 receptor agonist use and thyroid cancer. A retrospective FAERS analysis found that GLP-1R agonist use appeared linked to excess thyroid cancer risk, while metformin use did not.

A 2023 nested case-control study led by Bezin and colleagues added fuel. Published in Diabetes Care, it reported greater risk for all thyroid cancers and for medullary thyroid cancers among people currently using GLP-1 receptor agonists. The elevated risk showed up within three years of starting the drugs, raising the possibility that GLP-1s might accelerate existing precancerous changes rather than cause new ones from scratch.

The response from the medical community was swift and critical. Multiple commentary articles followed, each pointing to the same set of problems with the evidence. Patients on GLP-1 drugs see their doctors more often. More doctor visits mean more screening. More screening means more diagnoses, including cancers that might never have been caught otherwise. This phenomenon, known as detection bias, is one of the most common traps in observational research. On top of that, the Bezin study did not exclude patients who already had thyroid nodules or cancer before starting GLP-1 therapy.

The FDA responded by placing a warning on most GLP-1R-targeted therapies for patients with a personal or family history of medullary thyroid carcinoma or multiple endocrine neoplasia syndrome type 2. That warning remains in place today. It was a reasonable precautionary step given the data available at the time. But the warning was issued before the large-scale human studies had been conducted, before hundreds of thousands of patients had been tracked for years, and before multiple independent research groups across multiple continents had weighed in.

The gap between the warning label and the current evidence is substantial. Most patients read the warning, feel alarm, and never see the reassuring data that has accumulated since. That disconnect is why this article exists: to walk through what the research actually shows, study by study, number by number.

One more piece of context before diving into the data. The biology that raised the concern — C-cell proliferation in rodents — involves a specific type of GLP-1 receptor expression in rat thyroid tissue that is far less prevalent in human thyroid tissue. Rats have a much higher density of GLP-1 receptors on their thyroid C cells than humans do. This species difference does not automatically invalidate the concern, but it means extrapolating from rats to humans requires caution in both directions — just as the lack of a perfect animal model for human metabolism limits what any rodent study can prove about human disease.

86,000 People, 94,000 Trial Participants, and One Consistent Pattern

The early warning signals came from rodent biology and voluntary reporting databases. The data that followed came from something far more powerful: large-scale studies tracking real patients over real time periods, plus systematic analysis of randomized controlled trials.

The numbers have been remarkably consistent across different research groups, countries, and study designs. A retrospective cohort study published in JAMA Oncology in 2025 followed 86,632 matched adults from the OneFlorida+ health research network using electronic health records from 2014 to 2024. Half were GLP-1 receptor agonist users, half were matched nonusers. The cancer incidence rates landed at 13.6 per 1,000 person-years for GLP-1RA users versus 16.4 for nonusers, translating to a hazard ratio of 0.83 — a 17% lower overall cancer risk in the GLP-1 group.

That study did not stand alone. A year earlier, Dr. Li Wang and colleagues at Case Western Reserve University published what remains one of the largest analyses on this topic: 1,651,452 patients with type 2 diabetes drawn from a nationwide electronic health records database of 113 million US patients. With a 15-year follow-up window, they compared cancer incidence in people prescribed GLP-1 receptor agonists versus insulin or metformin.

The findings were striking. Compared with insulin, GLP-1RA use was associated with a significantly lower risk of 10 out of 13 obesity-associated cancers. The reductions were not subtle:

Cancer Type	Hazard Ratio vs Insulin	Risk Reduction
Gallbladder cancer	0.35	65%
Meningioma	0.37	63%
Pancreatic cancer	0.41	59%
Hepatocellular carcinoma	0.47	53%
Ovarian cancer	0.52	48%
Colorectal cancer	0.54	46%
Multiple myeloma	0.59	41%
Esophageal cancer	0.60	40%
Endometrial cancer	0.74	26%
Kidney cancer	0.76	24%

An important caveat: when Wang's team compared GLP-1 receptor agonists against metformin instead of insulin, the picture changed. Metformin has its own well-documented cancer-protective properties, and against this tougher comparator, GLP-1 drugs showed no decreased risk for any cancer type. This matters because it tells us that GLP-1 drugs may not be uniquely cancer-fighting — they may be comparable to other metabolic therapies that improve insulin sensitivity and reduce inflammation.

At the 2025 American Society of Clinical Oncology Annual Meeting, another large dataset entered the picture. Lucas A. Mavromatis of NYU Grossman School of Medicine presented findings from 170,030 adults across 43 US health systems, all with a BMI of 30 or higher and a diabetes diagnosis. After propensity score matching, the GLP-1RA group showed a 7% lower risk of obesity-related cancer and an 8% lower risk of death from any cause compared to those taking DPP-4 inhibitors.

Source: Dai H, et al. JAMA Oncology 2025. HR <1.0 indicates lower cancer risk in GLP-1RA users. Only ovarian, endometrial, and meningioma reached statistical significance. Kidney cancer was the only type trending higher.

The sex difference was notable. Women treated with GLP-1 receptor agonists had an 8% lower risk of obesity-related cancer and a 20% lower risk of death from all causes. For men, the difference was not statistically significant. Colorectal cancers showed the most pronounced effect: 16% fewer colon cancers and 28% fewer rectal cancers in the GLP-1RA group. ASCO President Robin Zon, MD, noted that "though this trial does not establish causation, it hints that these drugs might have a preventative effect."

Perhaps most surprising was the data on blood cancers. A 2025 study published in JAMA Network Open by Ashruf and colleagues at the Cleveland Clinic examined hematologic cancers specifically. GLP-1 receptor agonist use for type 2 diabetes was associated with a 54% reduction in all hematologic malignancies compared with insulin. That included lower risk across multiple subtypes: leukemias, lymphomas, myelodysplastic syndromes, myeloproliferative neoplasms, and multiple myeloma.

Key takeaway: Across multiple large studies involving hundreds of thousands of patients, GLP-1 receptor agonist use is consistently associated with equal or lower cancer risk compared to nonusers and insulin users. No study in this group found a significant overall increase.

Then in late 2025, the largest meta-analysis of randomized controlled trials arrived. Dr. Albert Ko and colleagues at Harvard's T.H. Chan School of Public Health pooled data from 48 randomized placebo-controlled trials with 94,245 participants, published in the Annals of Internal Medicine. Their conclusion: GLP-1 receptor agonists "probably have little or no effect" on risk for thyroid, pancreatic, breast, or kidney cancer, with moderate certainty evidence. The results held regardless of which specific GLP-1 drug was used, the dose, or the duration of follow-up.

Dr. Sai Yendamuri of Roswell Park Comprehensive Cancer Center summarized the state of the evidence bluntly: "If you look at the larger body of literature on this topic, it's very consistent."

The gastrointestinal cancer picture

The data on GLP-1 drugs and cancers of the digestive system has been especially encouraging. A recent meta-analysis covering 90 studies found no excess risk for hepatic, biliary tract, pancreatic, colorectal, or gallbladder cancers in people using GLP-1R therapies. That finding held whether the drugs were prescribed for type 2 diabetes or for weight loss, and regardless of the GLP-1 dose.

Liver cancer showed one of the strongest protective signals. A meta-analysis by Pasta and colleagues found that GLP-1R therapies reduced hepatocellular carcinoma risk by 58% compared with insulin or other antihyperglycemic drugs. Given that non-alcoholic fatty liver disease — a direct consequence of obesity and insulin resistance — is one of the fastest-growing causes of liver cancer worldwide, a drug that addresses the underlying metabolic dysfunction might be expected to reduce liver cancer risk as a downstream effect.

Colorectal cancer data tells a similar story. A meta-analysis of five studies including over 2 million people showed significantly lower colorectal cancer risk in GLP-1RA users compared with those using thiazolidinediones, SGLT2 inhibitors, and insulin. The use of exenatide, liraglutide, or semaglutide was associated with an 18% lower risk for colorectal cancer compared with TZDs and 43% lower risk compared with insulin. The insulin comparison is noteworthy because insulin therapy itself appears to be a risk factor: it was associated with a 69% greater risk for colorectal cancer, likely related to the mitogenic and pro-survival effects of insulin on colorectal cancer cells.

Breast cancer: neutral risk, more screening

Breast cancer has received particular scrutiny because of its high incidence and the hormonal pathways that link obesity to breast cancer risk. The evidence here is neutral in a specific and informative way.

Large retrospective studies demonstrate that breast cancer risk is not elevated in people using GLP-1 receptor agonists, particularly with follow-up of more than two years. There is also no excess risk for benign or premalignant breast neoplasms. One study found a greater risk of breast cancer diagnosis in patients using GLP-1 therapies who had no prior mammography screening history, but not in those with regular mammograms. The same study showed that mammography screening rates were higher in patients using GLP-1 therapies compared with controls.

A separate study found that excess breast cancer diagnoses increased with weight loss categories. Patients who lost more than 10% of their body weight were most likely to be diagnosed with breast cancer. But referrals for diagnostic breast imaging and mammography screening also increased with weight loss. This fits a well-known pattern: losing a significant amount of weight makes breast tissue less dense and tumors easier to detect on imaging. The cancers being found may not be new — they may have been there all along, hidden behind the signal-dampening effect of dense breast tissue.

The practical upshot: GLP-1 drugs do not appear to cause breast cancer, but they may lead to more breast cancer being found through improved screening and better imaging conditions after weight loss. That is arguably a good outcome for patients — earlier detection typically means better prognosis. The data also challenges a common assumption that finding "more cancer" after a treatment necessarily means the treatment is harmful. Sometimes it means the treatment gave patients access to medical attention they were not getting before, or changed their body in ways that made existing problems visible.

This detection bias pattern runs through the entire GLP-1-cancer literature. GLP-1 drugs bring patients into the medical system more frequently, improve their metabolic health in ways that make diagnostic procedures easier, and motivate both patients and physicians to take a closer look at health that was previously neglected. Accounting for this bias is one of the reasons why randomized controlled trials, which assign treatment regardless of health-seeking behavior, are so important for resolving these questions definitively.

One number that puts it in context

To understand the scale of who this affects: an estimated 137 million individuals in the United States are currently eligible for GLP-1RA therapies, and about 1 in 8 US adults has reported taking a GLP-1 agonist at some point. Roughly 40% of those individuals used the drugs specifically for weight loss. At that scale, even modest changes in cancer risk carry enormous public health implications. The fact that the evidence consistently trends toward benefit rather than harm is why the research community has shifted from asking "do these drugs cause cancer?" to "how much cancer risk do they prevent?"

The Danish Registry Study: When Living Longer Looks Like More Cancer

If the evidence is so reassuring, why did a major Danish study published in The Lancet Regional Health - Europe in 2025 find an increased cancer risk with GLP-1 drugs? The answer involves one of the most counterintuitive phenomena in medical statistics.

The Danish study was a register-based nationwide emulated trial that tracked all Danes who started either a GLP-1 receptor agonist or a DPP-4 inhibitor between 2007 and 2019. They were propensity score matched one-to-one and followed for up to 10 years. After 195,702 person-years, 4,758 participants developed cancer.

Among sustained GLP-1RA users, 4.1 more patients per 100 developed cancer compared to DPP-4i users at the 10-year mark. The hazard ratio for cancer at 6-10 years post-initiation was 1.35 (95% CI 1.05-1.73). Read in isolation, that looks worrying. But the study's own authors pointed to a critical companion finding.

Fewer GLP-1 receptor agonist users died without developing cancer first. The difference was -4.9 per 100 users. And when the researchers looked at the composite outcome of death or cancer combined, there was no difference at all: -1.15 per 100 users (95% CI -4.9 to 2.5). The math tells a clear story. GLP-1RA users were not getting more cancer because the drug caused it. They were getting more cancer diagnoses because they were living long enough to develop cancer that the DPP-4i users never survived to get.

This is sometimes called competing risks bias. If a treatment keeps people alive longer, they have more years of exposure to all the ordinary risks of aging, including cancer. A drug that cured all heart disease tomorrow would, statistically, appear to increase cancer rates. That does not make it dangerous. It makes it effective enough at preventing one cause of death that other causes show up.

The sex difference in the Danish data was also revealing. The excess cancer risk was 6.6 per 100 women but only 2.2 per 100 men (the latter not statistically significant). Given that women tend to live longer than men generally, and that the survival benefit of GLP-1 drugs may amplify that difference, the higher cancer detection in women is consistent with the survival benefit explanation.

The study's authors were transparent: "Long-term sustained users of GLP-1RA had a small increased risk of cancer; potentially explained by a survival benefit. Residual confounding by body mass index cannot be ruled out." That last point matters too. Patients prescribed GLP-1 drugs tend to have higher BMI at baseline than those on DPP-4 inhibitors, and higher BMI is itself a cancer risk factor that propensity score matching may not fully capture.

This study is also worth understanding in the context of how it differs from the other major studies discussed in this article. The Danish study compared GLP-1RA users to DPP-4 inhibitor users — both active treatments for diabetes. The JAMA Oncology study compared GLP-1RA users to nonusers. The Wang study compared against both insulin and metformin. Different comparators produce different results, and understanding which comparison a study makes is essential to interpreting the number it generates.

When a media headline says "Danish study finds GLP-1 drugs increase cancer risk," the truthful but complicated version is: sustained GLP-1 use was associated with slightly more cancer diagnoses than sustained DPP-4 inhibitor use at 10 years, but GLP-1 users also died less often, and when you count both outcomes together, the groups were the same. That sentence does not fit in a headline. The data does not support the alarm the headline creates.

Bottom line: The Danish study's "increased cancer risk" likely reflects GLP-1RA users living longer, not the drugs causing cancer. When death and cancer are combined into a single outcome, the two groups are indistinguishable.

Six Meta-Analyses, Three Countries, and One Consistent Answer on Thyroid Cancer

Because the original thyroid concern drove the FDA warning and continues to generate patient anxiety, it deserves a section of its own. The amount of research specifically addressing this question since the initial rat studies is substantial, and it points in the same direction.

The largest and most methodologically rigorous study on GLP-1 drugs and thyroid cancer was published in the BMJ in April 2024. Led by researchers at the Karolinska Institutet in Stockholm, with collaborators across Denmark, Norway, and Sweden, it tracked 145,410 patients who started GLP-1 receptor agonists against 291,667 who started DPP-4 inhibitors, all drawn from nationwide cancer registries between 2007 and 2021.

Over a mean follow-up of 3.9 years, 76 thyroid cancers occurred among GLP-1RA users (1.33 per 10,000 person-years) versus 184 among DPP-4i users (1.46 per 10,000 person-years). The thyroid cancer rate in the GLP-1 group was actually numerically lower. The hazard ratio was 0.93 (95% CI 0.66-1.31).

The medullary thyroid cancer subtype, the specific concern from the rodent data, showed a hazard ratio of 1.19 (95% CI 0.37-3.86). That wide confidence interval tells you everything you need to know: there were so few cases of medullary thyroid cancer in either group that no meaningful conclusion could be drawn. This is actually the expected result. Medullary thyroid carcinoma accounts for only about 2-4% of all thyroid cancers in the general population. Even in a study of over 400,000 people, there simply were not enough cases to detect a small effect if one existed.

The Scandinavian study was one of many reaching the same conclusion. Here is the trajectory of evidence on thyroid cancer:

Year	Study	Finding
2012	Alves et al. meta-analysis	No significant increase in thyroid cancer with liraglutide vs placebo/insulin/other drugs
2022	Hu et al. meta-analysis of RCTs	Corroborated 2012 findings; no risk regardless of underlying condition or comparator
2023	Nagendra et al. meta-analysis	Semaglutide specifically did not increase thyroid or pancreatic cancer risk
2024	Scandinavian BMJ cohort	HR 0.93 for thyroid cancer in 145K GLP-1RA users across 3 countries
2024	Silverii et al. meta-analysis	Fixed-effect: borderline significant, but fragility index of 1; random-effect: NOT significant
2025	JAMA Oncology cohort	Thyroid cancer HR 0.77 (0.54-1.11) — trending lower
2026	Ko et al. Ann Intern Med (48 RCTs)	OR 1.37 (0.82-2.31) — NOT significant; moderate certainty evidence

The Silverii finding deserves special attention because it is the one meta-analysis sometimes cited to support the thyroid concern. Their fixed-effect analysis did show a statistically significant increase in thyroid cancer risk. But the fragility index was 1, meaning a single additional thyroid cancer case in the comparator group would have made the association disappear. When they switched to a random-effects model, the association was no longer significant. A finding that fragile does not form a reliable foundation for clinical conclusions.

Meanwhile, preclinical work on newer GLP-1 drugs has actually flipped the script. A 2025 study found that semaglutide had tumor-suppressive effects on papillary thyroid carcinoma in mice, working through immune cell reprogramming rather than direct cytotoxicity. Semaglutide shifted macrophage polarization from a tumor-promoting M2 phenotype to a tumor-fighting M1 phenotype. The researchers showed this was a direct effect on immune cells, occurring through suppression of a protein called PPARy.

The practical implication: the rat C-cell biology that sparked the original concern does not appear to translate to meaningful thyroid cancer risk in humans. Rat thyroid C cells express far more GLP-1 receptors than human C cells, and the mechanism that causes C-cell proliferation in rats has not been demonstrated in human thyroid tissue at clinically relevant drug concentrations.

The weight of evidence is now sufficient that several expert groups have called for updating clinical guidance. The question is no longer whether GLP-1 drugs cause thyroid cancer in the general population — the data says they do not. The remaining question is whether specific high-risk subpopulations (such as those with pre-existing thyroid nodules or genetic predisposition to medullary thyroid cancer) warrant extra monitoring, which is what the current FDA warning already recommends. For the vast majority of GLP-1 users, thyroid cancer should not factor into their risk-benefit calculus.

Pancreatic Cancer and Kidney Cancer: Where Questions Remain

While the thyroid story has reached something close to resolution, the evidence on pancreatic cancer and kidney cancer is more mixed. These are the areas where intellectual honesty requires acknowledging genuine uncertainty.

The pancreatic cancer story begins in the same place as the thyroid story: adverse event reports. A 2011 analysis of FAERS data reported elevated pancreatic cancer risk with exenatide, one of the earliest GLP-1 receptor agonists. The FDA launched an investigation.

What followed was a decade of conflicting findings. Some cohort studies and meta-analyses found no excess risk. Others reported an elevated risk, but with an important caveat: the risk appeared influenced by the severity of the patient's type 2 diabetes. More advanced diabetes means more years of hyperinsulinemia, higher baseline inflammation, and more cumulative metabolic damage to the pancreas. Disentangling the effect of the drug from the effect of the disease it treats has been the central challenge in this area.

This is a problem called confounding by indication. Patients prescribed GLP-1 drugs tend to have more severe metabolic disease than those on older treatments. More severe disease carries its own elevated cancer risk. If you compare GLP-1 users to a healthier population without properly adjusting for disease severity, the GLP-1 group will look worse — not because the drug is harmful, but because the patients were sicker to begin with. Modern studies use propensity score matching and target trial emulation to reduce this bias, but they cannot eliminate it entirely.

The most reassuring data come from two directions. Wang's 1.65-million-patient study found a 59% reduction in pancreatic cancer risk with GLP-1RA versus insulin (HR 0.41). And the Ko meta-analysis of randomized trials found a pancreatic cancer odds ratio of 0.84 (0.53-1.35) — not significant, but if anything, trending toward protection rather than harm.

A more recent target trial emulation study from Wang and colleagues, published in the Journal of the National Cancer Institute in 2025, specifically examined pancreatic cancer. GLP-1RA use was associated with lower pancreatic cancer risk compared with several other antihyperglycemic drugs, including insulin, metformin, DPP4 inhibitors, SGLT2 inhibitors, sulfonylureas, and thiazolidinediones. The risk reduction was greater in patients with obesity than without.

One finding from the JCI review adds a wrinkle worth noting. When GLP-1 receptor agonists were combined with insulin, the pancreatic cancer risk was higher than with GLP-1RA alone. This may reflect insulin's own tumor-promoting properties in the pancreas (a well-documented phenomenon in preclinical models) or the fact that patients needing both drugs have more advanced metabolic disease.

Kidney cancer is the other persistent outlier. In the JAMA Oncology study, kidney cancer was the only type trending higher among GLP-1RA users, with a hazard ratio of 1.38 (95% CI 0.99-1.93). The Wang study found an even more pronounced signal when GLP-1 drugs were compared with metformin: HR 1.54 (95% CI 1.27-1.87) for kidney cancer.

This pattern is worth paying attention to because it is consistent across studies, even when other cancer types show no increase. But it is also puzzling because GLP-1 receptor agonists have been shown to improve kidney function in other contexts — they reduce proteinuria, slow the decline of kidney filtration rate, and are being actively studied as kidney-protective agents. The JAMA Oncology researchers noted that the kidney cancer risk was higher in patients younger than 65 and those who were overweight but not obese (BMI 27-29.9), which does not fit a straightforward biological explanation. In the Ko meta-analysis of randomized trials, the kidney cancer signal was not significant (OR 1.12, 95% CI 0.78-1.60).

One possibility is that GLP-1 drugs alter kidney physiology in ways that make existing preclinical kidney tumors more detectable, similar to the detection bias mechanism proposed for breast and thyroid cancers. Another is that the signal is confounded by the higher baseline BMI of GLP-1 users, since obesity itself is a significant kidney cancer risk factor. A third possibility is that there is a genuine biological mechanism we have not identified. Until randomized trials or more controlled analyses can tease these explanations apart, kidney cancer remains a category worth monitoring rather than one that should change prescribing decisions.

It is also worth noting what did not appear in the data. Across all these studies, there was no signal for bladder cancer, lung cancer, prostate cancer, breast cancer, or liver cancer that suggested increased risk. The pattern is overwhelmingly one of neutral or reduced risk, with kidney cancer as the lone exception.

Honest assessment: Pancreatic cancer risk with GLP-1 drugs appears neutral to reduced based on the weight of current evidence, though confounding by disease severity remains a challenge to fully resolve. Kidney cancer shows a small, unexplained signal in observational studies that does not appear in randomized trial data. Both warrant continued monitoring but do not support changes to clinical practice.

The Weight Loss Explanation Is Only Part of the Story

The most intuitive explanation for why GLP-1 drugs might lower cancer risk is straightforward: obesity increases cancer risk, GLP-1 drugs reduce obesity, therefore cancer risk drops. And there is plenty of evidence supporting that chain of logic. More than 40% of cancer deaths among US adults are linked to modifiable lifestyle factors including excess body weight, according to the American Cancer Society. At least 13 cancer types have elevated risk in people with obesity.

Obesity drives cancer through at least three parallel biological pathways, and GLP-1 drugs touch all three. Dr. Dawn Hershman, deputy director of the Herbert Irving Comprehensive Cancer Center at Columbia University, laid out the biological logic: "Obesity affects cancer risk in several biological ways. It can raise estrogen levels, which fuel certain hormone-related cancers, increase insulin and insulin-like growth factors that help tumors grow, and drive chronic inflammation that creates an environment where cancer is more likely to develop and spread."

Each of these mechanisms is a separate cancer accelerator, and GLP-1 drugs address all of them simultaneously through metabolic improvement. But weight loss alone does not account for everything the data shows. If it did, the drugs that produce the most weight loss would show the biggest cancer risk reductions. That is not what we see.

The most striking evidence comes from comparing GLP-1 drugs against bariatric surgery, which produces far more weight loss.

Wolff Sagy and colleagues in Israel tracked 3,178 matched pairs of patients (mean BMI 41.5) for a median of 7.5 years, comparing first-generation GLP-1RA users against those who had bariatric metabolic surgery. Despite bariatric surgery producing superior weight loss, the cancer rates were nearly identical: 5.62 per 1,000 person-years for surgery versus 5.89 for GLP-1RA (HR 1.11, not significant).

Here is where it gets interesting. The researchers ran a mediation analysis to estimate how much of the GLP-1 drugs' cancer risk reduction was attributable to weight loss versus other mechanisms. The result: 41% of the cancer risk reduction was estimated to be independent of weight loss (95% CI 15-59%). The drug appears to be doing something beyond just shrinking waistlines.

The preclinical research reviewed in the JCI 2025 review points to at least three mechanisms operating beyond weight loss:

Immune system reprogramming. Across multiple tumor models — thyroid, pancreatic, colorectal, breast — GLP-1 receptor agonists shifted the tumor immune environment. Macrophages were pushed from an M2 (tumor-promoting) to M1 (tumor-fighting) phenotype. T-cell infiltration into tumors increased. In one breast cancer study, semaglutide's antitumor effect was eliminated when CD8+ T cells were depleted but persisted when NK cells were removed, pinpointing adaptive immunity as the critical mediator.

Hyperinsulinemia reduction. Chronically elevated insulin is more than a marker of metabolic disease. Insulin is a mitogenic growth factor that directly promotes cancer cell proliferation and survival. Fasting insulin levels can be elevated for years before blood sugar rises, and GLP-1 receptor agonists reduce fasting insulin as part of their metabolic effects. This may partly explain why GLP-1 drugs showed the strongest cancer risk reductions against insulin as a comparator — insulin therapy itself may promote cancer through hyperinsulinemic effects.

Tumor microenvironment remodeling. In a pancreatic cancer model, semaglutide reduced collagen deposition and the mesenchymal stromal cell signature in the tumor microenvironment, making it easier for T cells to infiltrate. Think of it as the drug breaking down the walls of a fortress that tumors build around themselves.

These mechanisms were observed even in lean mice that were not fed high-fat diets, reinforcing that the effects are not simply downstream consequences of weight loss.

The bariatric surgery comparison tells the story

The comparison with bariatric surgery is worth dwelling on because it offers a natural experiment. Bariatric surgery causes dramatic, rapid weight loss — often far more than GLP-1 drugs achieve. A systematic review and meta-analysis of 32 bariatric surgery studies showed a 38% reduction in risk of any type of cancer and 41% for obesity-related cancers. If weight loss were the only mechanism through which cancer risk drops, bariatric surgery should produce far better cancer outcomes than GLP-1 drugs.

It does not. The Wolff Sagy data shows nearly identical cancer rates between the two groups, despite bariatric surgery producing more weight loss. This mismatch is the strongest indirect evidence that GLP-1 receptor agonists have pharmacological effects on cancer risk that operate through pathways other than fat reduction.

What the preclinical models show

The animal data fills in mechanistic detail that observational human studies cannot provide. In mice with pancreatic cancer, semaglutide slowed tumor growth and development of advanced lesions, reduced collagen deposition in the tumor environment, and increased T cell infiltration. Tumors build a protective barrier of fibrotic tissue and immunosuppressive cells — semaglutide appeared to break that barrier down.

The triple agonist retatrutide (which targets GLP-1, GIP, and glucagon receptors simultaneously) showed even stronger effects. In obese mice with pancreatic cancer, retatrutide reduced tumor engraftment rate, delayed tumor onset, and suppressed growth. When the drug was stopped, weight returned but established tumors continued to be suppressed — suggesting the drug had already changed the tumor's biological trajectory.

Breast cancer models added another layer. In lean mice (no obesity, no metabolic disease), semaglutide significantly delayed tumor emergence, reduced tumor size, and cut metastatic spread to the liver without affecting tumor blood vessel formation. Critically, semaglutide was not directly toxic to the cancer cells in lab dishes, confirming that the antitumor effect works through the immune system rather than through killing cancer cells directly. The drug increased dendritic cells (which present threats to the immune system), reduced regulatory T cells (which suppress immune responses), and boosted CD8+ killer T cells that directly attack cancer.

These are not abstract findings. They map onto what the human studies show: broad-spectrum cancer risk reduction that does not depend on weight loss, strongest in cancers where the immune system plays a known surveillance role. The question remaining is whether these preclinical mechanisms operate at the same scale in humans receiving standard doses for obesity or diabetes.

The Practical Calculus for People on GLP-1 Drugs

If you are currently taking semaglutide, tirzepatide, liraglutide, or another GLP-1 receptor agonist, the accumulated evidence offers more reassurance than the prescription label might suggest. But it also comes with honest caveats worth understanding.

The FDA black box warning about medullary thyroid carcinoma applies specifically to people with a personal or family history of medullary thyroid cancer or multiple endocrine neoplasia syndrome type 2. If you have neither of these rare conditions, the warning is not addressed to you. For the general population, multiple studies involving hundreds of thousands of patients have found no increased thyroid cancer risk.

GLP-1 drugs are not cancer prevention medications. As Dr. Hershman at Columbia emphasized, they are not designed to be taken forever, and many people regain weight after stopping. The cancer risk reduction seen in studies may depend partly on sustained metabolic improvements that could reverse with discontinuation. If you are taking a GLP-1 drug for weight management, the cancer data is a reassuring secondary signal, not a primary reason to take or continue the medication.

Several clinical trials are now specifically testing GLP-1 drugs in cancer patients. Tirzepatide will be evaluated in women receiving breast cancer treatment (NCT06517212), with both weight loss and cancer recurrence as endpoints over three years. A separate trial will examine tirzepatide's effect on disease-free survival in high-risk breast cancer (NCT06517212). Semaglutide is being studied alongside androgen deprivation therapy in prostate cancer patients (NCT06908694). These trials will provide the first prospective evidence on whether GLP-1 drugs affect cancer outcomes in people already diagnosed.

The kidney cancer signal is small and inconsistent between observational studies and randomized trials, but it is worth awareness. If you are taking a GLP-1 drug and have other kidney cancer risk factors, this is something to discuss with your physician. The biological explanation remains unclear and no clinical action is recommended based on current evidence.

There are things we still do not know. Most of the data comes from people with type 2 diabetes or obesity. Whether the same risk profile applies to people using GLP-1 drugs off-label, or who are on the lower end of the BMI scale, has not been rigorously studied. The newest drugs — particularly higher-dose semaglutide and tirzepatide — have less long-term follow-up than older agents like liraglutide and exenatide. And no randomized controlled trial has been specifically designed with cancer as a primary endpoint.

What the evidence does support: if you are taking a GLP-1 receptor agonist for diabetes or obesity management, you can do so without evidence-based concern that the drug itself increases your cancer risk. The largest meta-analysis of 48 RCTs concluded these drugs have "little or no effect" on cancer risk. The observational data, if anything, trends toward benefit. As Dr. Yendamuri at Roswell Park put it: "If you say obesity is increasing the risk of cancer, then if you decrease obesity or fix it, that should decrease the risk, right? That should make sense."

What we still genuinely do not know

Intellectual honesty requires listing the gaps that remain in the evidence. These are not rhetorical hedges — they are real questions that current data cannot answer.

First, nearly all the evidence comes from people with type 2 diabetes, obesity, or both. The cancer risk profile in people using GLP-1 drugs without these conditions — for example, off-label use for modest weight management — has not been studied at scale. The biology may differ because these patients start from a different metabolic baseline.

Second, the newest GLP-1 drugs have the least long-term data. Semaglutide was approved for diabetes in 2017 and for weight management in 2021. Tirzepatide's weight management approval came in 2023. The long-term studies drawing on 10-15 years of follow-up are mostly based on older drugs like liraglutide and exenatide. Whether the newer, more potent agents carry the same risk profile over decades of use is an assumption, not a demonstrated fact.

Third, most studies measure cancer incidence (whether cancer develops), not cancer outcomes (whether people survive cancer). A few early signals suggest GLP-1 drugs might improve cancer survival — the 6,800-patient UC system analysis showing GLP-1 users with colorectal cancer were less than half as likely to die within five years, for instance — but this evidence is preliminary.

Fourth, the weight regain problem. Most patients who stop GLP-1 drugs regain a significant portion of lost weight. If the cancer risk reduction depends partly on sustained metabolic improvement, discontinuation could erase that benefit. No study has specifically tracked cancer outcomes in people who used GLP-1 drugs, stopped, and regained weight.

Fifth, the interaction with cancer treatment. The JCI reviewers flagged an important clinical scenario: once a patient is diagnosed with cancer and begins treatment, physicians might choose to stop GLP-1 therapy. If that leads to rapid weight regain, it could theoretically promote cancer progression during the very period when the body is most vulnerable. The ongoing clinical trials (breast cancer NCT06517212, endometrial cancer NCT06751589) will help answer this, but results are years away.

A measured conclusion

The arc of evidence over the past decade bends toward reassurance. Rat C-cell biology did not translate into human thyroid cancer risk. Adverse event databases generated signals that controlled studies did not confirm. Large populations tracked over years showed equal or lower cancer rates. Multiple independent research groups in multiple countries arrived at compatible conclusions. And the mechanisms researchers are uncovering — immune reprogramming, insulin reduction, microenvironment remodeling — provide plausible biological explanations for why these drugs might genuinely protect against cancer.

None of this makes GLP-1 drugs cancer treatments. The effect sizes are modest, the evidence is observational, and no randomized trial has been designed with cancer prevention as a primary goal. But for the tens of millions of people taking these drugs for diabetes and obesity, the cancer safety question has been answered more thoroughly than most pharmaceutical questions ever are: these drugs do not appear to cause cancer, and they may reduce it.

For more on GLP-1 medication safety beyond cancer concerns, see our guides to Ozempic safety and semaglutide side effects and the broader GLP-1 weight loss drugs safety guide.

Frequently Asked Questions

Do GLP-1 drugs like Ozempic cause thyroid cancer?

Based on the current evidence, no. The thyroid cancer concern originated from rodent studies in the 1990s where rat thyroid C cells responded to GLP-1 by proliferating. However, multiple large human studies, including a Scandinavian study of 145,410 GLP-1RA users and a meta-analysis of 48 randomized trials with 94,245 participants, have found no significant increase in thyroid cancer risk. The FDA warning remains in place for people with a personal or family history of medullary thyroid carcinoma or MEN2, but population-level data does not support a general thyroid cancer risk.

Can GLP-1 receptor agonists actually reduce cancer risk?

Observational studies consistently show lower cancer incidence among GLP-1RA users, with overall cancer rates of 13.6 versus 16.4 per 1,000 person-years in one large study. However, these are not randomized cancer prevention trials, so causation is not proven. The reduction likely reflects a combination of weight loss, reduced hyperinsulinemia, and immune system changes. Ongoing clinical trials are testing whether GLP-1 drugs improve cancer outcomes in patients already diagnosed with breast and prostate cancers.

What about pancreatic cancer risk with GLP-1 drugs?

Early concerns from adverse event databases have not been confirmed by rigorous studies. The largest observational study found a 59% lower pancreatic cancer risk with GLP-1RA versus insulin in 1.65 million patients, and a meta-analysis of randomized trials found a non-significant odds ratio of 0.84, trending toward protection. The one caveat is that patients on insulin often have more advanced diabetes, making clean comparisons difficult.

Why does the Danish study show increased cancer with GLP-1 drugs?

The Danish 10-year registry study found a modestly higher cancer rate in sustained GLP-1RA users, but it also found fewer deaths without cancer in that group. When death and cancer were combined, there was no difference. The most likely explanation is that GLP-1RA users lived longer and therefore had more time to develop cancer — a survival benefit that looks like a cancer increase in isolated analysis.

Is kidney cancer a concern with GLP-1 receptor agonists?

Kidney cancer is the only type that has shown a consistently elevated (though mostly non-significant) signal in observational studies, with hazard ratios of 1.38 in one large cohort and 1.54 versus metformin in another. However, the largest meta-analysis of randomized trials found no significant increase (OR 1.12). The biological mechanism is unknown, and this remains an area for monitoring rather than clinical alarm.