Autoimmune gastritis diverges from a shared corpus origin: molecular reconstruction of the adenocarcinoma axis and clinical anchoring of the neuroendocrine axis.

Autoimmune gastritis (AIG) arises from a T-cell-mediated immune attack on the gastric parietal cell and carries a dual neoplastic risk: gastric adenocarcinoma through spasmolytic polypeptide-expressing metaplasia (SPEM) and intestinal metaplasia, and type-1 gastric neuroendocrine tumour (NET) through hypergastrinaemia-driven enterochromaffin-like (ECL) cell hyperplasia. This "one origin, two fates" framework is conceptually established but has never been given a molecular trajectory from an AIG origin, and molecular data for the neuroendocrine endpoint are essentially absent. We integrated public single-cell transcriptomes of the human AIG corpus (GSE271866; three samples, 14,323 cells) and of a histologically staged premalignant cascade (GSE134520; 44,012 cells) with a cross-sectional cohort of approximately 203 AIG patients, treating the two axes asymmetrically: the adenocarcinoma axis was reconstructed at the molecular level, whereas the neuroendocrine axis was clinically and literature anchored. The AIG corpus showed near-absent parietal cells (0.45%) with extensive SPEM (18.6%) and intestinal metaplasia (16.3%). Cascade pseudotime increased monotonically with histological stage (Spearman's ρ = 0.496; unchanged on the unbridged graph) and defined a gastric-to-intestinal progression signature that, transferred to the AIG corpus, tracked the AIG-intrinsic pseudotime (ρ = 0.673). Most AIG metaplastic cells were projected to the strong-metaplasia stage (90.9%; bootstrap 95% CI 84-98%; n = 55), a suggestive rather than definitive placement that is nonetheless consistent with AIG occupying the atrophy-to-metaplasia segment of the adenocarcinoma axis. In the cohort, hypergastrinaemia scaled with corpus atrophy (gastrin-17 versus pepsinogen I/II ratio ρ = -0.413, p = 9.9 × 10-8), and single-cell analysis confirmed a gastrin-responsive ECL population (HDC + 93%, CCKBR + 76%) whose neuroendocrine programme was orthogonal to the intestinal programme of the adenocarcinoma arm. Because AIG is a prototypic autoimmune disease, we frame how the autoimmune infiltrate steers each fate as the central open immunological question. AIG thus enters the metaplasia-to-cancer axis from a corpus, parietal-cell-loss origin and diverges, through a serum-gastrin-ECL bridge, towards the neuroendocrine fate; we provide an interpretable progression signature and a severity-anchored, hypothesis-generating framework-not an outcome predictor, since the cohort is cross-sectional with essentially no neoplastic endpoints-and report the molecular data gap for the neuroendocrine endpoint as a field-level priority.