PAIRWISE: Deep Learning based Prediction of Effective Personalized Drug Combinations in Cancer.
Overview
abstract
Combination therapies offer promise for improving cancer treatment efficacy and preventing recurrence. Preclinical screening strategies can prioritize synergistic drug combinations. However, identifying optimal combinations tailored to specific cancer subtypes and patients is challenging due to the vast number of possibilities and tumor heterogeneity. To address this gap, we developed PAIRWISE which explicitly modeled synergistic effects of drug combinations in cancer cell lines or tumor samples. PAIRWISE outperformed competing models with an AUROC of 0.847 on held-out cancer cell lines. When applied to an independent dataset with Bruton Tyrosine Kinase inhibitors in Diffuse Large B Cell Lymphoma, PAIRWISE accurately predicted synergistic combinations with an AUROC of 0.720. We validated 30 approved or investigational agents for DLBCL treatment and their synergy with BTKi in HTS. PAIRWISE showed strong concordance with in vitro screening results. These findings demonstrate the ability of PAIRWISE to nominate effective personalized drug combinations, accelerating precision oncology development.
