Initiation of glucagon-like peptide-1 receptor agonists was not associated with a higher 1-year risk of acute pancreatitis overall compared with sulfonylureas among patients with type 2 diabetes, according to a target trial emulation published in BMJ Medicine.
“This neutral overall effect is the net result of a small, early risk of drug-induced pancreatitis that is counterbalanced by later reductions in pancreatitis associated with alcohol use and hypertriglyceridaemia,” wrote lead author Yan Xie, MD, with the Clinical Epidemiology Center at the VA Saint Louis Health Care System in Missouri, and colleagues.
Researchers analyzed electronic health records from the US Department of Veterans Affairs health system to compare acute pancreatitis risk among 333,687 patients with type 2 diabetes (T2D) who initiated either a glucagon-like peptide-1 receptor agonist (GLP-1RA; n = 132,551) or a sulfonylurea (n = 201,136) between 2017 and 2023. Patients with prior pancreatitis, pancreatic cancer, or contraindications to either drug class were excluded. Participants were followed for up to one year.
The primary outcome was incident acute pancreatitis requiring emergency department evaluation or hospitalization. Researchers classified events into mutually exclusive categories: suspected drug-induced, alcohol-induced, hypertriglyceridemia-associated, biliary, and idiopathic or other causes.
In the intention-to-treat analyses, GLP-1RA initiation was associated with no clinically meaningful difference in overall acute pancreatitis risk, with a rate difference of -3.64 cases per 100,000 patients at one year compared with sulfonylureas.
Cause-specific analysis showed that, compared with sulfonylureas, GLP-1RA initiation was associated with:
23.45 additional suspected drug-induced pancreatitis cases per 100,000 patients
16.96 fewer hypertriglyceridemia-associated pancreatitis cases per 100,000 patients
10.32 fewer alcohol-induced pancreatitis cases per 100,000 patients
No meaningful differences in biliary or idiopathic pancreatitis.
Per-protocol analyses showed that 42% of the excess one-year risk occurred during the first three months after treatment initiation. By contrast, reductions in alcohol-associated pancreatitis accumulated mainly during months four through six, whereas reductions in hypertriglyceridemia-associated pancreatitis emerged primarily during months 10 through 12. This resulted in overall pancreatitis risk being modestly higher during the first 2 months after GLP-1RA initiation, after which the monthly risk was similar between groups.
A prespecified analysis of semaglutide, which accounted for 66.13% of GLP-1RA prescriptions in the cohort, recapitulated the class-level findings. Compared with sulfonylureas, semaglutide was associated with a higher risk of suspected drug-induced pancreatitis and lower risks of hypertriglyceridemia-associated and alcohol-induced pancreatitis, without increasing overall acute pancreatitis risk.
The study had several limitations. The cohort consisted primarily of older male US veterans, potentially limiting generalizability. Residual confounding is possible despite extensive adjustment for demographic, clinical, laboratory, medication, and health care utilization variables. Acute pancreatitis events that did not result in hospitalization or emergency department care may have been missed. In addition, suspected drug-induced pancreatitis remained a diagnosis of exclusion and could have been misclassified. The study also included only patients with T2D and did not evaluate outcomes by treatment indication.
Both Dr. Xie and Ziyad Al-Aly, MD, reported unpaid consulting relationships with Pfizer. Taeyoung Choi, MD, reported no competing interests.
Expert Insight
GI & Hepatology News invited corresponding author Ziyad Al-Aly, MD, chief of research and development at the VA Saint Louis Health Care System in Missouri, to elaborate on the study results.
Did any of the study findings stand out?
Dr. Al-Aly: We were surprised that GLP-1RAs lowered the risks of pancreatitis due to high triglycerides and alcohol. This is likely related to the effectiveness of GLP-1RAs in reducing triglycerides and curbing alcohol consumption. Even a modest effect on triglycerides still translated into a reduction in the risk of pancreatitis due to hypertriglyceridemia. The other surprise is that the reduced risk of alcohol- and triglyceride-induced pancreatitis was offset by an increased risk of suspected drug-induced pancreatitis.
What are the practical implications of these findings for clinicians?
Dr. Al-Aly: Clinicians should counsel patients on the risk of drug-induced pancreatitis. In patients with high triglycerides or an alcohol use disorder, initiating GLP-1RAs may actually reduce the risk of pancreatitis due to these causes. We also noticed that the increased risk of suspected drug-induced pancreatitis tends to be front-loaded, meaning it manifests shortly in the first few weeks after initiation. The beneficial effects of lowering triglycerides and curbing alcohol consumption tend to occur later in the course of taking GLP-1RAs.
What are the next steps for research in this area?
Dr. Al-Aly: We need to understand who is at most risk of drug-induced pancreatitis. Are there predictors or characteristics that make some patients more or less susceptible to drug-induced pancreatitis? These could be useful in stratifying risk among individuals. Also, the field of GLP-1RAs is growing rapidly, and different GLP-1RAs will likely have different risks.
What else would you like clinicians to take away from this work?
Dr. Al-Aly: We know that GLP-1RAs have GI side effects, including acute pancreatitis. We are still learning about what these drugs do and don't do, their effectiveness, and risks. The main takeaway from our study is that pancreatitis has many causes. GLP-1RAs increase the risk of one cause of pancreatitis and reduce the risk of others. This is a humbling lesson — one drug can act in opposite directions on various causes of pancreatitis.