Influence of differential source patterns in the detection of signals of disproportionate reporting for PARP inhibitors

Abstract

IntroductionCurrent individual case safety report (ICSR) databases contain almost 56 million unique spontaneous declarations of drug-event associations by health professionals but also by patients themselves. These databases have become a useful source for detecting signals of disproportionate reporting (SDR). However, since health professionals use a medical jargon that is often distant from the more colloquial terms used by patients, they usually report more frequently certain adverse events than patients and vice versa. The main objective of this work is to illustrate the existence of different reporting patterns among drugs within a class and to analyze their potential impact on SDR detection.MethodsFour ICSR databases were considered, namely, FAERS, VAERS, JADER, and VigiBase, with reports up until March 2024. They were all integrated in a single database following a careful deduplication and COVID-19 correction protocol. Measures of reporting odds ratio, proportional reporting ratio and empirical Bayesian geometric mean were used to evaluate disproportionate reporting.ResultsThe reporting patterns of four marketed oncology drugs, namely, olaparib, rucaparib, niraparib, and talazoparib, and an investigational drug, veliparib, were compared to those of a diverse set of eight clinically observed SDR, namely, fatigue, asthenia, anaemia, thrombocytopenia, neutropenia, insomnia, intestinal obstruction, and pneumonitis. The source pattern analysis revealed that olaparib and talazoparib are most frequently reported by physicians, and physicians are the main reporters of events such as neutropenia and pneumonitis, predisposing these events to be detected as SDR for those PARP inhibitors. In contrast, rucaparib and niraparib are most frequently reported by American consumers, and American consumers are the main reporters of events such as insomnia and intestinal obstruction, facilitating their detection as SDR for those two drugs. SDR detection was found to be robust to ICSR data completeness.DiscussionMatched reporting patterns between drugs and events may predispose certain drugs to be disproportionally associated with adverse events. Therefore, SDR detected from matched drug-event source patterns in ICSR databases should be challenged during signal validation. Class SDR for drugs with differential source patterns (such as fatigue, asthenia, anaemia, thrombocytopenia, and neutropenia for all PARP inhibitors) usually involve correcting opposite drug-event source patterns.