Abstract
BACKGROUND/OBJECTIVES: Early-onset colorectal cancer (EOCRC; diagnosed before age 50) is rising at an accelerated rate, with a disproportionate impact on underserved populations. While alterations in the receptor tyrosine kinase-RAS (RTK-RAS) signaling pathway play a fundamental role in colorectal cancer (CRC) biology, their prognostic significance in the setting of FOLFOX chemotherapy-particularly across different age groups and ancestral backgrounds-remains insufficiently characterized. We sought to characterize age-, ancestry-, and treatment-specific associations between RTK-RAS alterations and clinical outcomes using an AI-enabled precision oncology framework. METHODS: We analyzed 2515 CRC cases, including 266 Hispanic/Latino (H/L) and 2249 non-Hispanic White (NHW) patients, stratified by age at onset, ancestry, and FOLFOX treatment status. Mutation frequencies were assessed using Fisher's exact and chi-square tests, while overall survival was analyzed with Kaplan-Meier methods. The AI-HOPE and AI-HOPE-RTK-RAS conversational artificial intelligence platforms were used to integrate clinical, genomic, and treatment data via multi-parameter, natural language-based queries. RESULTS: In early-onset Hispanic/Latino patients, ERBB2 and NF1 mutations occurred at significantly lower frequencies in FOLFOX-treated cases compared with untreated cases (p = 0.01 for both). In late-onset H/L patients, NTRK2 mutations were depleted in FOLFOX-treated tumors (p = 0.04). In untreated early-onset H/L patients, MAPK3 and NF1 mutations were enriched relative to NHW counterparts. Among early-onset NHW patients, IGF1R and ERRFI1 mutations were less frequent with FOLFOX exposure, while multiple RTK-RAS genes were reduced in FOLFOX-treated late-onset NHW patients. Survival analyses revealed worse overall survival in FOLFOX-untreated early-onset NHW patients with RTK-RAS alterations (p = 0.029), but improved survival in FOLFOX-treated late-onset NHW patients (p = 0.048). CONCLUSIONS: RTK-RAS pathway alterations demonstrate strong age-, ancestry-, and treatment-specific prognostic effects and may serve as precision biomarkers of differential chemotherapy response. AI-enabled analytics substantially accelerated integrative biomarker discovery, supporting their utility for advancing precision oncology in EOCRC.