GLP-1 receptor agonists cause pancreatic cancer. The FAERS database is flooded with pancreatitis and pancreatic cancer reports — these drugs are giving people cancer.

The objection has a real published origin and deserves a steelman before it gets disambiguated. Singh et al., JAMA Intern Med 2013, 173:534–539 ran a population-based matched case-control study on a US administrative-claims database, identifying 1,269 patients hospitalized for acute pancreatitis against 1,269 matched controls, and reported an adjusted odds ratio of 2.24 (95% CI 1.36–3.68) for current exenatide or sitagliptin use within the prior 30 days, with a 2.01 (95% CI 1.37–3.18) ratio for recent use. That signal was published in a top-quartile general-medicine journal, drew from real claims data, and triggered FDA action. In parallel, Butler AE et al., Diabetes 2013, 62:2595–2604 examined eight pancreata from organ donors with type 2 diabetes treated with incretin therapy against twelve from non-incretin-treated diabetics and fourteen non-diabetic controls, reporting an approximately 40% increase in pancreatic mass, increased exocrine cell proliferation, increased pancreatic intraepithelial neoplasia, α-cell hyperplasia, glucagon-producing microadenomas in three of eight incretin-treated specimens, and one neuroendocrine tumor. The mechanistic concern about chronic GLP-1R agonism driving ductal proliferation was not invented out of nothing.

What the controlled-trial body has since shown. Azoulay et al., BMJ 2016, 352:i581 — a Canadian Network for Observational Drug Effect Studies international multicentre cohort of 972,384 patients initiating antidiabetic drugs between 2007 and mid-2013 — reported a pooled adjusted hazard ratio of 1.02 (95% CI 0.84–1.23) for pancreatic cancer with incretin-based drugs compared to sulfonylureas. The point estimate sits at the null. Cao C et al., Endocrine 2020, 68:518–525 pooled seven cardiovascular outcome trials (n=56,004 with type 2 diabetes, median follow-up 1.3–5.4 years) and found Peto odds ratios of 1.05 (95% CI 0.78–1.40) for acute pancreatitis and 1.12 (95% CI 0.77–1.63) for pancreatic cancer — both crossing the null, with no signal robust to sensitivity analysis. Liu Y et al., Diabetes Metab Res Rev 2018, 34:e3061 had earlier pooled four CVOTs (n=33,457) and reported Peto ORs of 0.89 (95% CI 0.63–1.27) for acute pancreatitis and 0.84 (95% CI 0.53–1.35) for pancreatic cancer — directionally below the null. Pinto LC et al., Sci Rep 2019, 9:2375, a meta-analysis with trial-sequential analysis covering 12 trials and 36,397 randomized patients, reported a pooled OR of 1.06 (95% CI 0.67–1.67) for pancreatic cancer. The cardiovascular outcome trials in this site's corpus — Marso et al. 2016 LEADER on liraglutide and Lincoff et al. 2023 SELECT on semaglutide — adjudicated pancreatitis as a prespecified safety endpoint and did not find excess events in the GLP-1 RA arms.

The Butler 2013 histology has not held up to independent re-review. Bonner-Weir et al., Diabetes Obes Metab 2014, 16:661–666 re-examined the same 34 organ-donor cases plus Network for Pancreatic Organ Donation cases that were not included in the original analysis and identified methodological problems sufficient to preclude meaningful conclusions: heterogeneity in subject age, variable incretin drugs and dosages, variable diabetes duration, and the confounding fact that intra-pancreatic fat proportion increases with age — which would artifactually inflate apparent β-cell and α-cell relative areas in the older incretin-treated subjects. The original sample size (n=8 incretin-treated) was small for a histopathology comparison drawing dysplasia conclusions. The autopsy series remains in the literature, but the subsequent independent histology work has not reproduced it.

Pancreatitis is not pancreatic cancer. The two conditions are pathologically and prognostically distinct. Acute pancreatitis is an inflammatory event of the exocrine pancreas; chronic pancreatitis is a fibrotic remodeling syndrome; pancreatic ductal adenocarcinoma is a malignant epithelial neoplasm with one of the worst five-year survival rates in oncology. Even granting a residual pancreatitis signal that the modern controlled-trial evidence does not strongly support, the inferential step from "small pancreatitis association" to "drug causes cancer" is not the same step. The mechanistic question — does chronic GLP-1R signaling drive ductal-cell proliferation at clinically relevant exposures in humans — has been examined across multiple model systems with mixed and largely null results, and the clinical question of whether this happens at observable rates in human exposure has been answered no by the controlled trials referenced above.

The FDA and EMA reached a joint conclusion. Egan AG, Blind E, Dunder K et al., NEJM 2014, 370:794–797 — the joint FDA/EMA Perspective titled Pancreatic Safety of Incretin-Based Drugs — reported that the two agencies had together reviewed more than 250 toxicology studies in approximately 18,000 animals, the clinical-trial and observational evidence then available, and the Butler 2013 histology, and stated that the assertion of a causal association between incretin-based drugs and pancreatitis or pancreatic cancer was inconsistent with the data. The 2013 FDA Drug Safety Communication on possible pancreatitis and pre-cancerous findings (FDA: FDA investigating reports of possible increased risk of pancreatitis and pre-cancerous findings of the pancreas from incretin mimetic drugs for type 2 diabetes) opened the review without imposing new restrictions; the 2014 joint statement closed it without adding pancreatic cancer warnings to package inserts. The pancreatitis warning, which had been present from the original approvals, remained. The labeling distinction was deliberate, not accidental.

What current package inserts actually say. Wegovy (semaglutide) prescribing information, 2025 revision carries a boxed warning for thyroid C-cell tumors including medullary thyroid carcinoma, derived from rodent carcinogenicity studies showing dose- and duration-dependent C-cell tumors in mice and rats at lifetime exposure. The boxed warning is for thyroid neoplasm — a different organ, a different cell type, a different mechanism. The pancreatitis warning is present in Warnings and Precautions rather than as a boxed warning, with the standard instruction to discontinue promptly if pancreatitis is suspected and not to restart if confirmed. There is no pancreatic-cancer warning. The Ozempic, Mounjaro, Zepbound, and Saxenda labels follow the same pattern. The "GLP-1s cause pancreatic cancer" framing collapses a thyroid C-cell rodent signal, a clinical pancreatitis-discontinuation warning, and a pancreatic-cancer claim that does not appear in any of these labels into a single sentence that none of the regulatory documents actually support.

The FAERS framing is methodologically broken. Disproportionality analyses on the FDA Adverse Event Reporting System have detected signals for pancreatic neoplasms in GLP-1 RA reports — a 2022 FAERS pharmacovigilance study calculated a proportional reporting ratio of approximately 9.86 for malignant pancreatic neoplasms while finding no overall increase in tumor reporting at the system-organ-class level (PRR 0.83). The interpretive problem is that FAERS is a passive spontaneous-reporting system with no denominator, no comparator, and no protection against reporter bias, channeling bias, or confounding by indication. Patients with type 2 diabetes carry a roughly 1.5- to 2.0-fold elevated baseline pancreatic-cancer risk versus non-diabetic populations, with the highest relative risk concentrated in the first two years after diabetes diagnosis — a pattern driven substantially by reverse causation, because new-onset hyperglycemia is itself an early symptom of subclinical pancreatic adenocarcinoma, with 74–88% of pancreatic-cancer patients diagnosed with diabetes within 24 months of cancer diagnosis (Magruder et al., Diabetes Metab J 2011 reviews the bidirectional epidemiology). A drug class prescribed selectively to patients in exactly the demographic window where occult pancreatic cancer is most likely to be discovered will accumulate FAERS reports linking the drug to the cancer regardless of any causal relationship between the two. The Boniol cohort study captures this directly — Boniol M et al., Diabetes Care 2018, 41:286–292 examined two retrospective cohorts (n=525,733 non-insulin antidiabetic users plus 33,292 incretin users in Belgium and Lombardy) and found an adjusted HR of 2.14 (95% CI 1.71–2.67) for short-term pancreatic-cancer risk with incretin therapy, but the authors interpreted this as reverse causation reflecting occult tumor presence at the time of prescribing rather than drug causation, with the elevated risk decreasing as follow-up lengthened. The same pattern — apparent association front-loaded in the first months of exposure, attenuating over time — is the signature of reverse causation rather than carcinogenesis.

What this page does and does not assert. It does not assert that GLP-1 receptor agonists are without risk. The label-listed pancreatitis precaution is real, the post-marketing reports of acute pancreatitis in GLP-1 RA users are real, and the discontinuation guidance in the prescribing information should be followed when clinical pancreatitis develops. The microdosing GLP-1 critic, the GLP-1 short-cycle critic, and the compounded-GLP-1 equivalence critic walk other parts of the risk-benefit landscape this site treats in detail; the semaglutide, tirzepatide, liraglutide, and exenatide peptide pages cover the molecule-specific evidence. The claim this page rejects is narrower: that GLP-1 RAs cause pancreatic cancer, that FAERS counts establish this, and that the early Singh and Butler reports have been confirmed rather than substantially walked back. The decade of controlled-trial and large-cohort evidence accumulated after 2013 — Azoulay 2016, Boniol 2018 (interpreted by its own authors as reverse causation), Liu 2018, Pinto 2019, Cao 2020, plus the prespecified pancreatitis adjudication in the CVOTs — converges on the null for pancreatic cancer. That is a different sentence from "GLP-1s are giving people cancer," and it is the sentence the published evidence currently supports.