Early postmarketing reporting patterns of resmetirom in FAERS: disproportionality analyses with indication-restricted and active comparators

Background Resmetirom is the first approved liver-directed thyroid hormone receptor-beta agonist for noncirrhotic metabolic dysfunction-associated steatohepatitis (MASH) with fibrosis. Early liver-event reporting may be distorted by indication leakage (the treated disease or its manifestations recorded as adverse events), disease progression, and surveillance bias. Methods We analyzed FDA Adverse Event Reporting System (FAERS) data from 2024Q1 through 2026Q1. Deduplicated primary- or secondary-suspect reports were assessed at preferred-term, system-organ-class, and nine predefined clinical-cluster levels using frequentist and Bayesian disproportionality metrics. References included all FAERS, US MASH-indication reports, and an exploratory 144-report active comparator. Sensitivity analyses examined cirrhosis indications, serious outcomes, onset-data selection, and co-medications. Results Among 3,155,054 deduplicated cases, 1,429 resmetirom reports were analyzed. Versus all non-resmetirom reports, RORs (95% CIs) were 12.13 (10.61–13.88) for hepatocellular injury and 11.46 (9.24–14.21) for cholestasis/biliary events. Gastrointestinal, thyroid/metabolic, and rash/hypersensitivity reporting was also enriched. Hepatic estimates remained elevated after excluding 35 cirrhosis-indication reports and in restricted comparisons, although sparse cells produced wide confidence intervals. The active comparator comprised 142 GLP-1/GIP and two pioglitazone reports. Against this reference, RORs were 1.36 (0.93–1.99) for gastrointestinal intolerance, 0.93 (0.59–1.47) for nausea, 1.42 (0.18–11.18) for cholelithiasis, and 1.95 (1.12–3.41) for diarrhoea. The first three attenuated, whereas diarrhoea remained elevated. Only 206 reports (14.4%) had valid exact onset dates, with systematic differences from reports lacking dates. Sex was missing in 94.6%. Gemfibrozil and clopidogrel were co-reported in three and 25 cases, respectively. Conclusion Early FAERS reporting patterns recapitulated recognized gastrointestinal, hepatic, biliary, cutaneous, and thyroid/metabolic domains. Comparator-dependent attenuation suggests that crude all-FAERS estimates partly reflect treatment-context reporting. However, the small GLP-1/GIP-dominant comparator, missing data, target-organ confounding, and spontaneous-report limitations preclude incidence, comparative-risk, or causal inference. These hypothesis-generating patterns support clinical monitoring and denominator-based confirmation.