AUTHOR=Zhang Zhiqiang , Zhong Xin , Jin Qunlong , Chen Zhumei , Yang Yanming , Jiang Youheng , Zhou Huaixiang , An Caiyan , Zhang Junjing , He Yulong , Fu Zhang , Wu Kaiming , Li Ningning 

TITLE=Prognostic modeling of early-onset nondistal gastric cancer identifies ARSB–PDCD1 ratio as an immune-related survival stratifier

JOURNAL=Frontiers in Immunology

VOLUME=Volume 16 - 2025

YEAR=2025

URL=https://www.frontiersin.org/journals/immunology/articles/10.3389/fimmu.2025.1655106

DOI=10.3389/fimmu.2025.1655106

ISSN=1664-3224

ABSTRACT=BackgroundDespite a global decline in gastric cancer (GC) incidence, nondistal GC (NDGC) is increasingly prevalent among younger patients, necessitating targeted investigation of early-onset NDGC (EONDGC) to identify prognostic determinants for enhanced risk stratification.MethodsEONDGC patients were identified from multiple datasets, including the Surveillance, Epidemiology, and End Results (SEER) database, the Cancer Genome Atlas (TCGA) Stomach Adenocarcinoma cohort, and the Affiliated Hospitals of Sun Yat-sen University (SYSU) as an external validation cohort. Propensity score matching was performed to reduce baseline differences between groups. A prognostic model was developed using univariate and multivariate Cox regression and LASSO analysis in a 7:3 training–validation split. The prognostic model was applied to TCGA patients to generate risk scores, and high-risk patients were selected for differentially expressed genes (DEGs) analysis. The identified genes were then analyzed using Cox regression and Kaplan-Meier methods to determine prognostic relevance. In parallel, MGC-803 and AGS cells were transiently transfected to overexpress ARSB; RT-qPCR verification, scratch and transwell migration assays quantified motility.ResultsA total of 535 EONDGC patients from SEER and 171 from SYSU were included. The prognostic model, incorporating seven clinical variables (race, pathological grade, T, N, and M stage, lymph node ratio, and chemotherapy), achieved robust performance with concordance index values of 0.758 (training), 0.718 (validation), and 0.762 (SYSU), with all AUCs > 0.75. In the TCGA patients, 73 upregulated genes were identified from high-risk patients through DEGs analysis. Among these, ARSB and PDCD1 were determined to be independent prognostic markers based on Cox and Kaplan-Meier analyses. Furthermore, a higher ARSB/PDCD1 ratio (APR) was associated with poorer overall survival (P = 0.041). In vitro, ARSB overexpression increased scratch migration area and transwell-migrated cell counts versus empty vector.ConclusionThis study developed a clinical prognostic model for EONDGC and therefore identified ARSB and PDCD1 as key molecular markers. The APR value enhances survival stratification, offering valuable insights into personalized prognosis and potential immunotherapy strategies.