The association of pancreatitis with antidiabetic drug use: gaining insight through the FDA pharmacovigilance database

In patients with diabetes, disease per se, co-morbidities and drugs, including novel agents acting on the incretin system, have all been associated with pancreatitis with controversial data. We investigated the publicly available FDA Adverse Event Reporting System (FDA_AERS) database to gain insight into the possible association between antidiabetic agents and pancreatitis. To this aim, a case/non-case method was retrospectively performed on the FDA_AERS database (2004-2009 period). Cases were defined as reports of pancreatitis according to the Medical Dictionary for Regulatory Activities (MedDRA) terminology. All other reports associated with antidiabetics were considered non-cases. The Reporting Odds Ratio (RORs), with corresponding 95% confidential interval (CI) and Mantel-Haenszel corrected P value, was calculated as a measure of disproportionality, with subsequent time-trend analysis. We retrieved 86,938 reports related to antidiabetics, corresponding to 159,226 drug-report combinations: 2,625 cases and 156,601 non-cases. Disproportionality was found only for exenatide (number of cases, 709; ROR, 1.76; 95% CI, 1.61-1.92; P (MH) < 0.001) and sitagliptin (128; 1.86; 1.54-2.24; < 0.001). For exenatide, significant disproportionality appeared in the first quarter of 2008 (ROR, 1.24; 95% CI, 1.10-1.40; P (MH) < 0.001), soon after the FDA alert; for sitagliptin in the second quarter of 2008 (1.41; 1.05-1.90; 0.021). This temporal analysis found a striking influence of relevant FDA warnings on reporting of pancreatitis (the so-called notoriety bias) and is, therefore, recommended to avoid transforming a pharmacovigilance signal of alert automatically into an alarm. The precise quantification of the risk of pancreatitis associated with antidiabetics deserves assessment through specific disease-based registries.