Feasibility of Using Population Pharmacokinetics‐Based Virtual Control Groups in Organ Impairment Studies

Abstract

This work aimed to assess the feasibility of using population pharmacokinetics (popPK) to generate virtual healthy control groups in organ impairment studies. Data from 11 organ impairment studies containing 18 organ impairment arms and 13 healthy control groups across 7 drugs were analyzed. Area under the concentration‐time curve (AUC) and maximum concentration (Cmax) were calculated from popPK‐simulated individual concentration‐time profiles for participants in the healthy control groups, accounting for the participant's specific covariate(s) (N = 1000 replicates). The AUC and Cmax geometric mean ratios (GMRs; simulated healthy control/observed healthy control and observed organ impairment/simulated healthy control) were calculated. The simulated healthy control group geometric mean exposures were within 30% of the observed geometric mean exposures in 8 of the 11 studies (73%). The number of organ impairment arms for which the observed GMR (observed organ impairment/observed healthy control) and median of simulation‐based GMRs (observed organ impairment/simulated healthy control) for AUC and Cmax being within the same fold change were 12 (67%) and 13 (72%) arms, respectively. The number of organ impairment arms for which the median of simulation‐based AUC and Cmax GMRs were within the 90% confidence interval of the observed GMRs were 14 (72%) and 15 (83%), respectively. Poor concordance was observed for 1 drug (3 arms), where healthy participants' data were not incorporated in the popPK model. This work supports using popPK‐based virtual control groups in organ impairment studies. Subsequent work should aim to establish best practices for constructing popPK‐based virtual control groups.