Divergent lineage trajectories and genetic landscapes in human gastric intestinal metaplasia organoids associated with early neoplastic progression

Abstract

Background Gastric intestinal metaplasia (IM) is a precancerous stage spanning a morphological spectrum that is poorly represented by human cell line models. Objective We aim to establish and characterise human IM cell models to better understand IM progression along the cancer spectrum. Design A large human gastric IM organoid (IMO) cohort (n=28), their clonal derivatives and normal gastric organoids (n=42) for comparison were established. Comprehensive multi-omics profiling and functional characterisation were performed. Results Single-cell transcriptomes revealed IMO cells spanning a spectrum from hybrid gastric/intestinal to advanced intestinal differentiation. Their lineage trajectories connected different cycling and quiescent stem and progenitors, highlighting differences in gastric to IM transition and the potential origin of IM from STMN1 cycling isthmus stem cells. Hybrid IMOs showed impaired differentiation potential, high lineage plasticity beyond gastric or intestinal fates and reactivation of a fetal gene programme. Cell populations in gastric IM and cancer tissues were highly similar to those derived from IMOs and exhibited a fetal signature. Genomically, IMOs showed elevated mutation burden, frequent chromosome 20 gain and epigenetic deregulation of many intestinal and gastric genes. Functionally, IMOs were FGF10 independent and showed downregulated FGFR2. Several IMOs exhibited a cell-matrix adhesion independent subpopulation that displayed chromosome 20 gain but lacked key cancer driver mutations, potentially representing the earliest neoplastic precursor of IM-induced gastric cancer. Conclusions Overall, our IMO biobank captured the heterogeneous nature of IM, revealing mechanistic insights on IM pathogenesis and progression, offering an ideal platform for studying early gastric neoplastic transformation and chemoprevention. Data are available in a public, open access repository. The WES, RNAseq, scRNAseq and scCNV data were deposited into the European Genome-Phenome Archive (EGAS00001007899). RNAseq, scRNAseq and methylation data were deposited to GEO, GSE210995.