A murine ex vivo 3D kidney proximal tubule model predicts clinical drug-induced nephrotoxicity.

Abstract

Drug attrition and clinical product withdrawals due to nephrotoxicity remain major challenges for pharmaceutical drug development pipelines. Currently, no reliable high-throughput in vitro screening models are available that provide reliable, predictive toxicology data for clinical nephrotoxicity. Drug screens to predict toxicity and pharmacology assessments are compromised by standard two-dimensional (2D) cell monoculture models. Here we extend a previously reported murine three-dimensional (3D) kidney-derived intact proximal tubule model to provide ex vivo drug toxicity data that reliably compare to clinical experiences and improve nephrotoxicity predictions over current 2D cell assays. Proximal tubule cytotoxicity was monitored by ATP depletion for 12 compounds (acarbose, acetylsalicylic acid, captopril, cimetidine, cidofovir, cisplatin, doxorubicin, gentamicin, polymyxin B, polymyxin B nonapeptide, probenecid and vancomycin) in 3D proximal tubule ex vivo assays. Drug concentration-response curves (1-1000µM) and IC50, lowest effective concentration (LEC) and AUC values were compared to clinical therapeutic exposure levels (Cmax). The 100-fold Cmax threshold demonstrated best sensitivity (96.9%) and specificity (87.5%) for this assay, with high positive (93.9%) and negative (93.3%) predictive values for nephrotoxicity. IC50 values were superior to LEC. Results also support the model's capability to predict substrate-inhibitor/competitor interactions, yielding toxicity results similar to those reported in vivo. These 3D proximal tubule-based drug screens provide more reliable nephrotoxicity predictions, and more insight into complex mechanisms implicated in nephrotoxicity than current standard 2D cell assays. This new approach for rapid drug toxicity testing produces more reliable clinical comparisons than current 2D cell culture screening techniques.