Predicting renal clearance of morphine and morphine‐6‐glucuronide using the “organ‐on‐a‐chip” technology and physiologically‐based pharmacokinetic modeling

Abstract

BackgroundRenal elimination of xenobiotics consists of dynamic sequential and parallel processes including glomerular filtration, active secretion, and active and passive reabsorption. Conventional ways to predict renal clearance involve allometric scaling and in vitro‐to‐in vivo extrapolation using cell lines. However, static predictions using these methods are often not quantitatively accurate because of the differences in transporter expression and function between species and systems, and the lack of dynamics of reabsorption and concentration gradients that exist in the kidney. We hypothesized that a 3D vascularized human proximal tubule microphysiological system (VPT‐MPS) together with physiologically‐based pharmacokinetic (PBPK) modeling could be used to accurately predict human renal clearance. The goal of this study was to test this hypothesis using morphine and its 6‐glucuronide (M6G) as model compounds and to predict their renal clearances from data collected from our VPT‐MPS 1 using our mechanistic kidney PBPK model 2.MethodsHuman proximal tubule epithelial cells (PTECs) collected from three donors were seeded in the 3D VPT‐MPS. Passive diffusion and active secretion clearances of morphine and M6G were measured using the 3D VPT‐MPS in the presence and absence of probenecid (OAT inhibitor) and tetraethyl ammonium (OCT inhibitor). Similarly, passive diffusion and active secretion clearances across PTECs for morphine and M6G were measured in a 2D transwell system. The measured clearances were used with our verified mechanistic kidney PBPK model to predict the renal clearances of morphine and M6G. The predicted renal clearances were compared to the observed in vivo renal clearances to define the accuracy of the predictions.ResultsThe clearances across PTECs in the 3D VPT‐MPS for morphine and M6G were significantly reduced in the presence of probenecid and tetraethyl ammonium. In the 3D VPT‐MPS system, the passive diffusion clearances of morphine and M6G were 0.8 (range: 0.4 – 1.2) and 0.3 (range: 0.2 – 0.7) μL/h, respectively, while the active secretion clearances of morphine and M6G were 2.6 (range: 1.4 – 7.9) and 0.9 (range: 0.04 – 9.3) μL/h, respectively. These results showed a significant active secretion for both morphine and M6G that is mediated by OAT and OCT transporters. The predicted renal clearances using these measured clearances and our mechanistic kidney PBPK model were 7.6 ± 3.0 L/h for morphine and 6.4 ± 1.5 L/h for M6G. These predicted renal clearances were within 2‐fold of the observed population mean renal clearances demonstrating successful in vitro‐to‐in vivo prediction of renal clearance using our method.ConclusionThis study shows that the 3D VPT‐MPS together with mechanistic PBPK modeling can accurately predict renal clearance of drugs that are eliminated through both passive and active processes.Support or Funding InformationNIH UG3TR002158‐02