Meta-Analysis Finds Tirzepatide Safe for Pancreatitis Risk but Flags Gallbladder Concerns

This was a systematic review and meta-analysis following PROSPERO registration (CRD42023412400). Researchers searched Embase, PubMed, and the Cochrane Library through March 2023 for randomized controlled trials comparing tirzepatide to placebo or active comparators (basal insulin, GLP-1 agonists) in people with type 2 diabetes or obesity. Heterogeneity was assessed using I² and Cochran's Q test. A fixed effects model was used to estimate safety outcomes including pancreatitis, composite gallbladder/biliary disease, cholecystitis, and cholelithiasis.

As tirzepatide becomes one of the most prescribed medications for diabetes and obesity worldwide, understanding its safety profile is essential. Rapid weight loss from any cause can increase gallbladder disease risk, and GLP-1-class drugs have previously raised concerns about pancreatitis. This meta-analysis provides reassurance on the pancreatitis front but raises a clinically important flag about gallbladder and biliary complications that clinicians and patients should be aware of.

Gallbladder complications have emerged as a consistent safety signal across the GLP-1 and GIP/GLP-1 drug class, likely related to the rapid weight loss these drugs promote. This meta-analysis extends that concern specifically to tirzepatide, the newest entrant in the class. As these drugs move into broader populations — including non-diabetic obesity treatment — monitoring for gallbladder disease becomes even more important since the patient population is expanding well beyond the traditionally studied groups.

The relatively small number of pancreatitis events across trials means the study may be underpowered to detect a true increase in risk (wide confidence interval: 0.59–3.61). The composite gallbladder/biliary outcome was significant but individual components were not, suggesting the signal may be driven by the aggregation. The search only included trials through March 2023, missing more recent large-scale studies. Fixed effects models were used, which may not be appropriate if there is undetected heterogeneity. The control groups varied (placebo, insulin, GLP-1 agonists), introducing potential comparison inconsistencies.