Molecular therapy matching score correlates with clinical outcomes: a system for navigating cancer complexity

Abstract

Historically, precision medicine implies matching one biomarker to cognate monotherapies. However, next-generation precision oncology must address cancer complexity. Indeed, advanced tumors have a median of five genomic alterations; with ~700 cancer-causing genes, there are >1 trillion patterns. We describe an algorithmic Matching Score (expert-curated knowledge base within a digital reasoning framework) that evaluates cancer drugs’/drug combinations’ ability to address a patients’ entire next-generation sequencing/multi-omic-derived molecular profile. Using prospective, peer-reviewed published clinical trial and other study results analyzed retrospectively, Matching Scores were produced for 540 advanced/metastatic cancer patients (number of unique pathogenic alterations = 374); the most common cancers were breast ( n  = 98) and colorectal ( n  = 64). Matching Score results was evaluated with Kaplan–Meier, Univariate and Multivariate Cox regression analyses demonstrating that Matching Scores correlated significantly with improved progression-free survival (0% vs. 67–100% matching: 3.0 vs. 9.0 months, hazard ratio (HR): 0.34, P  < 0.01), overall survival (0% vs. 67–100% matching: 9.0 vs. 18.7 months, HR: 0.54, P  < 0.01), and clinical benefit rate (stable disease≥6 months/objective response) (0% vs. 67–100% matching: 22.1% vs. 55.9%, odds ratio: 4.59, P  < 0.01). This study underscores the clinical importance of the precision medicine approach that maximizes matching of multiple molecular alterations with cognate drugs.