For a study, researchers sought to compare the molecular characteristics of AGEJ to those of EAC and gastric cancer. Due to morphological ambiguity and epidemiologic differences, the classification of AGEJ based on differential molecular features between EAC and gastric adenocarcinoma has been a long-standing debate. Based on differential mRNA expression of EAC (N=78) and GCFB (N=102) from The Cancer Genome Atlas (TCGA) cohort, a molecular classification model with a Bayesian compound covariate predictor was built. The genomic, transcriptomic, and proteomic features of AGEJ/cardia (N=48) in the TCGA cohort and AGEJ/upper third GC (N = 46 pairs) in the cohort were compared and grouped into the EAC-like or GCFB-like groups. According to the 400-gene classifier, AGEJ in both cohorts was classed as EAC-like (31.2%) or GCFB-like (68.8%). The phosphoinositide 3-kinase-AKT signaling in the GCFB-like group was significantly elevated, while ERBB2 expression was suppressed. Compared to the GCFB-like group, the EAC-like group had dramatically different alternative splicing, including the skipped exon of RPS24, much higher copy number amplification, ERBB2 amplification, significantly higher copy number amplification higher protein production of ERBB2 and EGFR. The EAC-like group had a considerably greater response to lapatinib than the GCFB-like group (P=0.015) in a high-throughput 3D drug test employing independent cell lines. In both the Seoul National University and TCGA cohorts, AGEJ was the merged entity of the EAC-like and GCFB-like groups, with consistently differing molecular features. Dual inhibition of ERBB2 and EGFR could successfully target the EAC-like group with a high Bayesian compound covariate prediction score.
