AI-assisted Discovery of an Ethnicity-influenced Driver of Cell Transformation in Esophageal and Gastroesophageal Junction Adenocarcinomas

Although Barrett’s metaplasia of the esophagus (BE) is the only known precursor lesion to esophageal adenocarcinomas (EACs), drivers of the metaplasia→dysplasia→neoplasia cascade in the esophagus remains incompletely understood. Using an AI-guided network transcriptomics approach, in which EAC initiation and progression is modeled as networks to simplify complex multi-cellular processes, we first predict cellular continuum states and disease driving processes with an unprecedented degree of precision. Key AI-guided predictions are subsequently validated in a human organoid model and patient-derived biopsies of BE, a case-control study of genomics of BE progression, and in a cross-sectional study of 113 patients with BE and EACs. We find that all EACs must originate from BE, pinpoint a CXCL8/IL8↔neutrophil immune microenvironment as a driver of cellular transformation in both EACs and gastroesophageal junction-ACs. This driver is prominent in Caucasians (Cau), but notably absent in African Americans (AAs). Network-derived gene signatures, independent signatures of neutrophil processes, CXCL8/IL8, and an absolute neutrophil count (ANC) are associated with risk of progression. SNPs associated with ethnic changes in ANC modify that risk. Thus, findings define a racially influenced immunological basis for cell transformation and suggest that benign ethnic neutropenia in AAs may serve as a deterrent to BE→EAC progression. BRIEF SUMMARY Esophageal adenocarcinoma (EAC) is a highly lethal cancer among Caucasians, while African Americans are somewhat protected; what factors drive transformation with racial disparity remain unknown. AI-enabled creation of the first computational map of neoplastic progression in the esophagus built and validated using transcriptomic datasets from diverse cohorts of human samples pinpointed CXCL8↔neutrophil tumor immune-microenvironment as a racially influenced driver of EACs and GEJ-ACs. Computational tools pinpoint a racially influenced driver of cell transformation during BE→EAC progression; in doing so, it reveals new novel biology, informs disease modeling, therapeutic strategies, and biomarkers. LAY SUMMARY By modeling diseases as networks, this work unravels a fundamental race-influenced immunologic driver of cell transformation in adenocarcinomas of the esophagus and the gastroesophageal junction. ### Competing Interest Statement RY is a consultant for Medtronic (Institutional), Ironwood Pharmaceuticals (Institutional) and Phathom Pharmaceuticals, and has research support from Ironwood Pharmaceuticals. She is also on the advisory Board with Stock Options at RJS Mediagnostix. These entities had no influence on study design or conclusions. * BE : Barrett’s metaplasia of the esophagus EAC : Esophageal adenocarcinoma GEJ : Gastroesophageal junctionx Cau : Caucasian AA : African American SNP : single nucleotide polymorphism DARC : Duffy antigen/ receptor for chemokines