Abstract
This white paper reports early findings from the Ignota Labs project Understanding and Mitigating Drug Toxicity Risks in Diverse Populations through AI-Enabled Genomics and Pharmacogenomics. Building on our previous work, in which the causal and explainable AI platform SAFEPATH predicted and experimentally validated a novel PRKD1/PRKD3-sphingolipid mechanism of gefitinib hepatotoxicity, we asked why the clinical toxicity response to gefitinib is so heterogeneous between patients. Pharmacogenomic analysis of a non-small-cell lung cancer cohort identified CYP3A4 activity as a candidate modifier, and we therefore profiled two donor-diverse, female iPSC-derived hepatocyte lines with contrasting CYP3A4 activity - i18F (low activity, predominantly Native American ancestry) and i30F (high activity, predominantly European ancestry) across five gefitinib concentrations by RNA-seq. Both lines converge on suppression of globo sphingolipid metabolism, independently reproducing the mechanism identified in our earlier case study. Beyond that shared core, the two lines diverge sharply: i18F shows early stress signalling, collapse of phase I/II biotransformation and loss of hepatocyte identity transcription factors, consistent with parent-compound accumulation; i30F mounts a high-output metabolic response with strong NRF2-ARE, oxidative stress, p53 and genotoxicity enrichment, consistent with a burden of reactive metabolites. The NRF2 signature in the higher-European-ancestry line aligns with population-level genomics, in which KEAP1-NRF2 alterations are far more frequent in Western than in East Asian NSCLC cohorts, offering a mechanistic rationale for population-specific differences in gefitinib toxicity and resistance. These findings are hypothesis-generating and define a set of specific experimental validations.
Read more:
Divergent Mechanisms of Gefitinib Hepatotoxicity in Metabolically and Genetically Diverse Hepatocyte Models can be read in full at ignotalabs.ai.
Authors: Sara Masarone, Katie V. Beckwith, Philip S. Lewis, Kate Cameron, Layla Hosseini-Gerami
Published: 17 September 2026
