Physiologically based pharmacokinetic (PBPK) modeling is less well established for substrates of UDP-glucuronosyltransferases (UGT) than for cytochrome P450 (CYP) metabolized drugs and more verification of simulations is necessary to increase confidence. To address specific challenges of UGT substrates, we developed PBPK models for four drugs cleared majorly via glucuronidation (lorazepam, oxazepam, naloxone, and zidovudine). In vitro to in vivo scaling of intrinsic clearance generated with co-cultured human hepatocytes was applied for hepatic metabolism and extra-hepatic clearance was extrapolated based on relative expression of UGT isoforms in the liver, kidney, and intestine. Non-metabolic clearance and the contributions of individual UGT isoforms to glucuronidation were based on in vitro and in vivo studies taken from the literature and simulations were verified and evaluated with a broad set of clinical pharmacokinetic data. Model evaluation showed systemic clearance predictions within 1.5-fold for all drugs and all simulated parameters were within 2-fold of observed. However, during the verification step, top-down model fitting was necessary to adjust for under-prediction of zidovudine V_SS and renal clearance and over estimation of intestinal first pass for lorazepam, oxazepam, and zidovudine. The impact of UGT2B15 polymorphisms on the pharmacokinetics of oxazepam and lorazepam was simulated and glucuronide metabolites were also simulated for all four drugs. To increase confidence in predicting extra-hepatic clearance, improvement of enzyme phenotyping for UGT substrates and more quantitative tissue expression levels of UGT enzymes are both needed. Prediction of glucuronide disposition is also challenging when active transport processes play a major role.

By Luca Docci, Kenichi Umehara, Stephan Krähenbühl, Stephen Fowler, Neil Parrott