Overlap coefficient for assessing the similarity of pharmacokinetic data between ethnically different populations

Abstract

We developed a method to assess the similarity of pharmacokinetic data between ethnically different populations. An evaluation of confidence intervals for the mean difference in pharmacokinetic parameters, such as area under the concentration-versus-time curve (AUC), between populations is often used. We propose the use of the overlap coefficient (OC), which represents the proportion of overlap between two probability distributions, as a measure of the similarity between distributions. We considered five OC estimators--two parametric ones and three nonparametric ones. Simulation studies were conducted to compare the performance of the five OC estimators and their bootstrap confidence intervals. Results showed that nonparametric estimators with fixed-bandwidth kernel density estimation had a smaller mean squared error in almost all situations, and their coverage probabilities were close to the nominal level. The proposed method was applied to pharmacokinetic data from a bridging study of a combination therapy for metastatic colorectal cancer patients in the USA and Japan. From the analyses of this study, it was suggested that the distributions of the logarithmically transformed AUC for leucovorin and 5-fluorouracil were similar between the two populations.