Refining cardiovascular liability - Bayesian analysis of safety pharmacology in vivo studies

Abstract

The ICH E14/S7B Q&As includes Best Practice Considerations for the In Vivo QT Studies. In particular, the nonclinical studies for QTc prolongation should demonstrate adequate sensitivity and power to detect a QTc prolongation effect with a similar magnitude observable by dedicated clinical QT studies i.e., ΔQTc +10 msec. This is in accordance with the 3R (reduce/refine/replace) principle which aims to minimize the number of animals used on studies. In order to use the nonclinical and clinical data for an integrated assessment as a substitute for a thorough QT/QTc (TQT) study, unequivocal evidence of low QTc risk needs to be established using nonclinical in vitro and in vivo studies as defined for a nonclinical double-negative. To this end, we have developed a species-specific Bayesian paradigm that makes use of data collected from historical safety pharmacology studies in dog or monkey to construct the prior of all CV/telemetry endpoints, including QTc, with respect to both the circadian rhythm and dose response. This model was inspired by a publication co-authored by FDA on the subject, and yet considers features of our internal cardiovascular telemetry studies. We have built this model into an R package containing a shiny app to facilitate the analysis by Safety Pharmacology study directors and scientists. For most of the case studies examined, the Bayesian analysis improved the precision of the treatment effect assessment (5 % to 20 % reduction in the least significant difference (LSD)). By randomly removing animals from the study, we demonstrated the Bayesian method's ability to restore the accuracy even with incomplete data. Also, the Bayesian method provides a natural probabilistic statement of the treatment effect to facilitate decision making. All procedures performed on animals were in accordance with regulations and established guidelines and were reviewed and approved by an Institutional Animal Care and Use Committee or through an ethical review process.