Leveraging Cancer Drug Metabolites for Precision Drug Repurposing
Albert A. Antolín, Luca Ruvo, Leticia Manen-Freixa, Salud Santos, Jose Manuel Brea, Marı́a Isabel Loza, Laia Josa‐Culleré, Carmen Serra, Amadeu Llebaria · 2025
Precision oncology benefits only a subset of patients, with approved drugs targeting less than 4% of the human proteome. 1 Drug metabolites—often overlooked—can reach therapeutically relevant concentrations in human plasma, and exhibit distinct in vivo activities from their parent drugs, offering untapped repurposing potential. We recently demonstrated in a proof-of-concept study that we can leverage machine learning (ML) to identify unique targets for the cancer drug rucaparib’s major metabolite, revealing anti-Parkinson activity in iPSC neurons. 2 To explore this systematically, we curated 18 drug-metabolite pairs from 224 FDA-approved anticancer drugs. These metabolites, present in plasma at >10% parent drug levels, were predicted to have distinct off-target profiles using multiple ML methods. We synthesized the metabolites, validated target binding in vitro, and submitted them to EU-OPENSCREEN for broader phenotypic screening. We also assessed their antiproliferative effects across a cancer cell line panel aligned with their parent drug indications. Our findings confirm that over 25% of tested metabolites engage different targets than their parent drugs. Notable examples include a metabolite 5–10× more potent in prostate cancer models where the parent drug failed clinically; a metabolite with immunomodulatory effects absent in the parent drug; and metabolites with improved selectivity or antibiotic activity. Overall, we show that the comprehensive characterization of drug metabolites can: (i) explain clinical efficacy/toxicity discrepancies, (ii) reveal new repurposing opportunities in diseases of high unmet medical need such as oncology, neurodegeneration, and infection, and (iii) expand the targetable proteome in precision medicine—unlocking more effective, tailored therapies for patient benefit.