Physiologically relevant Multi-Omics Flux Analysis Reveals Metabolic Mechanisms of drug-Induced Liver Toxicity

Adverse drug reactions remain a major challenge in drug development and clinical practice. Biomarkers for drug-induced liver injury reflect evolving injury rather than enabling early detection or mechanistic insight before leakage occurs. Physiologically relevant assays that also capture dynamic human drug exposure may close that gap. Here, we present a systems pharmacology framework integrating physiologically based pharmacokinetic modeling with multi-omics profiling of primary human liver spheroids exposed to clinically relevant drug concentration trajectories. Using the anti-tuberculosis drug isoniazid, we combined time-resolved transcriptomics, proteomics, metabolite exchange, and viability measurements with context-specific genome-scale metabolic modeling to resolve pathway-level metabolic function. Despite limited transcriptional changes, proteomic alterations more closely reflected metabolic flux adaptations, particularly in lipid metabolism. Flux analysis revealed early increases in amino acid utilization preceding apoptosis, indicating stress-induced metabolic reprogramming prior to cell death. Toxic drug exposure induced a lipid-dominated metabolic state characterized by increased lipid turnover, elevated redox demand, and enhanced amino acid utilization, despite limited lipid availability from cell death. In contrast, therapeutic exposure led to a more balanced metabolic state, including bile acid synthesis. Together, this PBPK-informed, multi-omics framework enables mechanistic interpretation of drug-induced metabolic stress and provides a causality-driven strategy for improving early detection and safety assessment.