AI-Driven Repurposing of a Phase 3 Failed Compound for Pancreatic Cancer

Our Approach:

We combined deep learning, physics-based modeling, and curated biological data to discover new therapeutic potential:

Cancer Target Database: We developed a proprietary therapeutic target database for oncology, integrating genomics, proteomics, and cheminformatics to map the landscape of validated and emerging cancer targets.

AI-Driven Target Prediction: Using graph-based neural networks, the molecule was screened against this target space.

Physics-Based Screening: Structural docking and advanced molecular dynamics simulations confirmed strong predicted binding to several of the identified targets.

Biological Validation: The compound demonstrated dose-dependent inhibition of pancreatic cancer growth in multiple in vitro and in vivo models.

Combination Synergy: The molecule showed strong synergy with gemcitabine, the standard-of-care for pancreatic cancer.

Key Results:

Validated AI predictions through molecular modeling and biological assays.

Significant anti-cancer activity observed across more than one preclinical pancreatic cancer model.

Confirmed synergistic effect when combined with gemcitabine.

Repurposed molecule with clinical history, offering accelerated development potential in oncology.

Impact & Future Direction:

This work demonstrates the power of combining AI with physics-based methods to unlock hidden value in failed clinical-stage compounds. By repositioning an existing molecule into oncology specifically for pancreatic cancer, a disease with extremely poor prognosis we have identified a promising therapeutic candidate supported by mechanistic rationale and preclinical efficacy.

Disclaimer: Compound identity, molecular target and specific experimental details withheld due to confidentiality.