Artificial Intelligence-Enabled Precision Medicine Reveals Prognostic Impact of TGF-Beta Pathway Alterations in FOLFOX-Treated Early-Onset Colorectal Cancer Among Disproportionately Affected Populations

Abstract

Early-onset colorectal cancer (EOCRC; <50 years) incidence is increasing most rapidly among Hispanic/Latino (H/L) populations. While the transforming growth factor-beta (TGF-{beta}) pathway influences colorectal cancer (CRC) progression, its prognostic role in FOLFOX-treated EOCRC, particularly in H/L patients, is unclear. We analyzed 2,515 CRC cases (H/L = 266; NHW = 2,249) stratified by ancestry, age at onset, and FOLFOX treatment using Fishers exact, chi-square, and Kaplan-Meier analyses. We then applied AI-HOPE and AI-HOPE-TGF{beta}, conversational artificial intelligence (AI) platforms that integrate clinical, genomic, and treatment data, to perform complex, natural language-driven queries requiring multi-parameter integration. TGF-{beta} pathway alterations occurred in 28-39% of H/L and 23-31% of NHW patients, with SMAD4 as the predominant driver. BMPR1A mutations were enriched in FOLFOX-treated EO H/L patients (5.5% vs. 1.1% EO NHW; p = 0.0272), while late-onset NHW non-FOLFOX cases had higher SMAD2/TGFBR2 mutation rates. In FOLFOX-treated EO H/L patients, TGF-{beta} pathway alterations predicted poorer survival (p = 0.029); no survival impact was seen in other groups. SMAD4 mutations were less frequent in EO H/L than EO NHW receiving FOLFOX (2.74% vs. 13.87%; p = 0.013). TGF-{beta} pathway alterations may serve as an ancestry- and treatment-specific biomarker of poor prognosis in FOLFOX-treated EO H/L CRC. AI-enabled integration accelerated biomarker discovery, supporting precision medicine.