Abstract C006: Automated generation and multi-endpoint validation of novel prodrug candidates for pediatric cancer therapeutics using 110 chemical transformations

Abstract

Drug repurposing offers accelerated development timelines for pediatric cancers, but the use of off-patent generics presents intellectual property challenges that limit commercial investment in programs serving small patient populations. Novel prodrug modifications can create patentable compositions of matter while preserving established pharmacological activity, enabling both IP protection and pharmacokinetic improvements relevant to pediatric dosing and CNS penetration for brain tumor indications. We developed an automated computational pipeline for generating, evaluating, and characterizing novel prodrug candidates from parent drugs identified through systematic rare disease and pediatric cancer repurposing screens. The pipeline employs 110 validated chemical transformations implemented in RDKit, comprising 77 prodrug modifications including site-specific esters, amino acid conjugates, phosphate and phosphonate variants, carbamates, heterocyclic modifications, and fluorinated analogs, and 33 metabolic stability modifications including CYP450 blockers, glucuronidation blockers, esterase resistance modifications, and deuteration strategies. An intelligent combination generator produces dual and triple modifications with automated site-conflict detection, yielding 300 to 500 novel candidates per parent drug. Each candidate undergoes evaluation across 98 ADMET endpoints using ADMET-AI, molecular docking with AutoDock Vina, and patent novelty scoring via PubChem and ChEMBL similarity searches. Atom-level position tracking via maximum common substructure analysis enables position-specific patent claim generation for each novel compound. Across 25 analysis campaigns targeting rare pediatric disease and pediatric cancer drug candidates, the pipeline generated 1,172 novel prodrug structures. Automated ADMET screening identified candidates with improved oral bioavailability, enhanced blood-brain barrier penetration, or extended half-life relative to parent compounds, properties of particular relevance for pediatric CNS tumor therapeutics. This pipeline addresses the critical IP barrier that has historically deterred investment in generic drug repurposing for pediatric cancers by enabling systematic generation of patentable analogs with comprehensive computational characterization.