(048) Exploring early signals for gliptins and gliflozins: A pharmacovigilance study on recent oral antidiabetic groups in Spain.

Background: It is well known that pharmacological actions of new oral antidiabetic agents commercialized and spreadly used in the last decades have led to new adverse drug reactions or contraindications, such as empagliflozin and kidney failure or urinary infections; and alogliptin and heart or hepatic failure.

Objectives: To analyze suspected adverse drug reactions of Dipeptidyl peptidase 4 inhibitors (iDPP-4, gliptins) and Sodium-glucose co-transporter 2 inhibitors (iSGLT2, gliflozins) reported through the national database of the Spanish Agency of Medicines and Medical Devices to detect new early signals.

Methods:

Design: Data mining study following the Bayesian methodology applied by the Uppsala Monitoring Center belonging to WHO for searching early signals in databases of suspected adverse drug reaction reports.
Setting: Data of all reported adverse events for Dipeptidyl peptidase 4 inhibitors (iDPP4; ATC: A10BH) and Sodium-glucose co-transporter 2 inhibitors (iSGLT2; ATC: A10BK) from the national Spanish Agency of Medicines and Medical Devices repository were analyzed. In this database, information is reported following MedDRA terminology, and we extracted data aggregated by Preferred Term (Pt). We analyzed as control suspected adverse events reported for metformin (ATC: A10BA02).
Main outcome measures: early positive signals of adverse drug reaction candidates to be analyzed in specific clinical trials or post-marketing studies.
Statistical analysis: a validated adaptation of the data mining Bayesian Confidence Propagation Neural Network (BCPNN) methodology extended to the multiple comparison setting was applied. The threshold of the estimator false discovery rate (FDR) equal to or lower than 0.050 implies an early positive signal for the active ingredient, and among 0.050-0.100 a signal to follow up. The adaptation performed contrasts all the adverse drug reactions of specific ATC subgroups; and separately from the integral database.

Results: The pair active ingredient-suspected adverse drug reaction with a clear signal was metformin-diarrhea (FDR=0.052; control), followed by sitagliptin-xerostomia (FDR=0.059); both reported in fact sheets and exercising as controls for the method. We observed highlighted signals not reported in the Summaries of Product Characteristics for the pairs empagliflozin-asthenia (FDR=0.067), linagliptin-bone fracture (FDR=0.076), and dapagliflozin-hyperglycemia (FDR=0.087).

Conclusions: These signals detected closely to the threshold and not reported in fact sheets are candidates to be followed up to observe the evolution and preventively considered for health professional notifiers.