What serious adverse event signals (pancreatitis, thyroid, eye) have been reported for tirzepatide in FAERS and EudraVigilance, and how should clinicians interpret disproportionality findings?
Executive summary
Post-marketing pharmacovigilance analyses of tirzepatide identify disproportionate reporting for pancreato-biliary events (including pancreatitis), thyroid-related events, and eye-related events across FAERS and EudraVigilance, but these are signal-detection findings—not proofs of causation—and must be interpreted in the context of reporting bias, confounding and limited denominator data [1] [2] [3].
1. What the databases actually show about pancreatitis
EudraVigilance lists pancreatitis among the most commonly reported adverse events for tirzepatide—one extract cites 363 reports of pancreatitis and notes a higher reporting frequency of gastrointestinal and pancreato-biliary adverse events versus several comparator GLP-1 receptor agonists (ROR/PRR analyses provided) [3]; FAERS-based studies likewise flag pancreato‑biliary disorders as disproportionately reported for tirzepatide versus all other drugs in the database (forest plots and RORs presented in a FAERS analysis) [1]. Multiple FAERS mining studies covering different windows since approval (Q2 2022 onward) have detected pancreatitis or pancreato-biliary PTs among “positive signals” identified by ROR or Bayesian methods [2] [4].
2. Thyroid signals: signal presence, not incidence or causality
Thyroid‑related PTs (including those of regulatory concern such as medullary thyroid carcinoma in hypothesis-driven reviews) appear among disproportionally reported categories in FAERS analyses: published FAERS pharmacovigilance studies explicitly investigated thyroid‑related adverse events and included them in forest-plot comparisons of RORs for tirzepatide versus other drugs [1]. These analyses raise regulatory and clinical interest but do not provide incidence rates or demonstrate causal linkage; the FAERS and EudraVigilance outputs are aggregate spontaneous reports without verification of diagnosis or denominators [1] [4].
3. Eye events—diabetic retinopathy and visual harms—what’s in the signal files
Several analyses have called out eye-related events, including diabetic retinopathy, among signals in FAERS and EudraVigilance; one FAERS study explicitly investigated diabetic retinopathy and other eye PTs and shows disproportionate reporting for tirzepatide in subgroup plots [1], and EudraVigilance analyses list eye‑related AE clusters in comparative disproportionality work across incretin agents [3]. Importantly, the raw signal literature notes that some eye events may relate to rapid glycemic change or underlying diabetes progression—an important confounder when interpreting any association [1] [3].
4. How clinicians should interpret disproportionality analyses
Disproportionality (ROR, PRR, BCPNN) identifies statistical excesses in spontaneous reporting, useful for hypothesis generation but not proof of causality; FAERS/EudraVigilance lack exposure denominators, are subject to underreporting, stimulated reporting, and confounding by indication (patients with diabetes already at elevated baseline risk for pancreatitis, thyroid disease and retinopathy), and many papers explicitly caution about these limits while reporting their signal findings [5] [6] [2]. Clinicians must therefore treat signals as prompts for vigilance: assess baseline risk (history of pancreatitis, medullary thyroid carcinoma risk factors, active proliferative retinopathy), counsel patients about warning symptoms (severe abdominal pain, new neck masses or persistent dysphagia, visual changes), and report suspected events to regulators to improve signal fidelity [5] [4].
5. Practical monitoring and reporting steps grounded in the evidence
Given the signal patterns in FAERS and EudraVigilance, pragmatic actions are warranted: consider baseline pancreatic history and avoid in active pancreatitis, maintain guideline‑concordant retinal screening for diabetic patients prior to and during therapy, remain alert for thyroid‑related symptoms and evaluate suspicious findings rather than assuming causality, and promptly submit adverse event reports to pharmacovigilance systems to help refine risk estimates—recommendations that mirror the caution offered by the primary pharmacovigilance studies themselves [1] [3] [2].
6. What the signal literature still cannot tell clinicians
None of the cited disproportionality studies provides a true incidence, controlled risk ratio, confirmed adjudicated case series, or mechanistic proof linking tirzepatide to these serious events; the papers emphasize the need for corroboration by pharmacoepidemiology with denominators, prospective registries, or adjudicated case–control studies to establish magnitude and causality [5] [6] [2].