Development and Verification of Virtual Population Models for Predicting Drug Pharmacokinetics in Ethnic North American Populations.

Ethnic variabilities can affect the outcome of drug pharmacokinetics (PK) and drug-drug interactions (DDI). This work aimed to develop four North American (NA) sub-populations: White, African American, Asian American, and Hispanic_Latino suitable for physiologically based pharmacokinetic (PBPK) modeling and simulations. Demographic data and tissue weight/volume, blood flows, cardiac output, plasma protein levels, hematocrit, enzyme and transporter abundances/frequencies, serum creatinine, glomerular filtration rate, and gastrointestinal transit times for the different populations were collated. Equations describing various covariate relationships for these physiological parameters were developed for each ethnicity. Some key population differences that can affect drug PK were higher CYP3A5 and OATP1B1 population mean abundances for African Americans and lower CYP3A4 and OATP1B1 population mean abundances for Asian Americans. The most common CYP2C9 alleles in the White as well as the Asian populations are the *1, *2, and *3 alleles. However, additional lower-activity alleles (*5, *6, *8, and *11) were also shown to occur among the African Americans and Hispanic_Latino subjects. Clinical studies reporting population specific PK profiles (6 studies) or results from a mixed NA population (14 studies including 6 DDI studies) were simulated, and the simulated data were within two-fold of the observed data. PBPK simulations using the 4 NA population files and repaglinide as a probe compound predicted significant differences in the drug exposure for the Asians, African Americans, and White subjects. In conclusion, the developed population files can aid drug development for specific subgroups of the diverse NA population by accounting for variabilities in drug kinetics and DDI consequences.