Artificial intelligence-driven precision medicine identifies prognostic WNT pathway alterations in African American colorectal cancer patients treated with FOLFOX

Background: African Americans (AA) experience disproportionate burden of colorectal cancer. Dysregulation of the Wingless-related integration site (WNT) and transforming growth factor-beta (TGF-beta) pathways contributes to tumor progression, yet their prognostic roles in FOLFOX-treated CRC among AA patients remain understudied. Methods: We analyzed 2,562 colorectal cancer cases stratified by ancestry, age at onset, and FOLFOX treatment using Fisher's exact, chi-square, and Kaplan-Meier analyses from AACR Project GENIE and cBioPortal databases. To enhance data integration and interpretation, we applied AI-HOPE and AI-HOPE-WNT/TGFbeta, conversational artificial intelligence (AI) platforms designed to integrate clinical, genomic, and treatment data through natural language-driven queries. Results: Overall survival analyses showed that early-onset AA patients treated with FOLFOX who had WNT pathway alterations experienced significantly better survival (p = 0.035). WNT pathway alterations were less frequent in late-onset AA patients treated with FOLFOX compared to those not treated (80% vs. 92%; p = 0.05). Similarly, TGF-beta pathway alterations were reduced in late-onset non-Hispanic White (NHW) patients receiving FOLFOX compared to untreated cases (23% vs. 31%; p = 0.0005). Conclusions: Chemotherapy exposure may influence pathway-specific mutation frequencies across ancestry and disease stage. AI-enabled integrative analyses highlight the potential of conversational AI platforms to accelerate biomarker discovery and reveal ancestry- and treatment-specific vulnerabilities in colorectal cancer.