A Quantitative Systems Toxicity Workflow to Model Adverse Outcome Pathways’ Biomarkers

Recent in silico advancements have introduced new approach methodologies (NAM) in drug development and risk assessment. NAMs provide various advantages, including supporting the three Rs (replace, reduce, refine), hypothesising underlying mechanisms, and facilitating system (in vitro to in vivo) and species translation of data. Here, we demonstrate a quantitative renal toxicity model using cisplatin and carboplatin as example compounds. Cisplatin is a platinum-based anticancer drug that presents significant nephrotoxicity while carboplatin, although similarly a platinum-based anticancer drug, does not present similar severities of nephrotoxicity even at a higher dose. A physiologically based pharmacokinetic (PBPK) model was developed for cisplatin and carboplatin and linked to a renal toxicity model integrating kidney transcriptomics, histopathology, and Kim-1 release to explore mechanisms for this disparity. The physiological basis of the PK model facilitated species translation that recovered known cisplatin and carboplatin PK in human, rat, and mouse. This model showed cisplatin exposure in the kidneys was significantly higher than that of carboplatin. Additionally, sensitivity analysis of the model parameters showed high cisplatin volatility and active kidney uptake as mechanisms for higher cisplatin exposure in the kidneys than carboplatin. The mechanistic structure of the PK model here can also be adapted for other platinum-based drugs, such as oxaliplatin. Linked to the physiologically relevant kidney cisplatin PK, the renal toxicity model was able to recover the dynamical changes of injury-relevant transcriptomics, necrotic cell fraction, and Kim-1 elevation observed in rat.