S1333 Integrative Analyses Reveal Bile Acid Receptors Are Novel Prognostic Markers for Gastric Cancer

Abstract

INTRODUCTION: Bile acid receptors (BARs) are increasingly being spotlighted as potential drivers of gastrointestinal cancers. Few studies have explored the role BARs play in gastric cancer (GC) tumorigenesis and potential clinical implications, especially as it relates to duodenogastric biliary reflux (DGR). We aggregated transcriptomic data to identify functional and prognostic roles for BARs as novel biomarkers in GC management. METHODS: We collected transcriptomic data from The Cancer Genome Atlas, Genome Tissue Expression, and Gene Expression Omnibus repositories. We leveraged unique aspects of each dataset to analyze associations between BARs and GC prognosis, subtype, and clinicopathology. We also utilized ingenuity pathway analysis to assess BAR associated functions and upstream regulators as it relates to GC. RESULTS: BARs showed differential distribution in GC; membrane BARs (GPBAR1, S1PR2, and CHRM2) were enriched in diffuse, genome stable, and mesenchymal-type tumors while nuclear BARs (PXR, CAR, and FXR) expression was mainly unaltered only being elevated metabolic subtypes. Every BAR (except PXR and FXR), was significantly associated with poor prognosis in GC. Membrane, but not nuclear, BARs were also associated with advanced stage, grade, tumor size, treatment failure, and leukocyte infiltration. We found additional associations with H. pylori infection and membrane BARs, which were geographically distributed according to GC mortality rates; enriched in Eastern European and Asian samples. Last, we found genes associated with GC tumorigenesis including EGFR, TGFB1, SMARCA4, SMAD4, NORAD, TEAD1-4, and P38-MAPK as common activated and miR-218, miR-146a-5p, miR-34a-5p, and ETV6-RUNX1 as common inhibited upstream regulators of membrane BARs. CONCLUSION: GC is the fifth most common and third deadliest cancer worldwide, carrying a median survival of less than a year. H. pylori infection remains the top risk factor for developing GC and is a known cause of DGR. However, little is known about the role components of DGR, namely bile acids, plays in GC. In a comprehensive transcriptomic analysis of more than 1500 samples, we identified BARs as prognostically relevant, novel biomarkers associated with unfavorable clinicopathologic features of GC. Specifically, we find compelling evidence that membrane BARs are directly implicated in diffuse GC tumorigenesis, primarily through bile acid induced TGFB1 signaling and downstream EFGR activation as result of DGR.Figure 1.: BARs are differentially expressed and are novel prognosticators of GC. Comparative meta-analysis of BAR expression patterns between (A) normal and gastric tumor tissue and (B) between diffuse and intestinal Laurén GC subytpes. Each heat map panel represents the mean difference (MD) of Log2-transformed values. Overall MD was based on inverse variance and random effects weighting. (D) Plot showing median time to first progression based off median low and high BAR expression cutoffs. (E & F) BAR hazard ratios (HRs) for (E) post-progression and (F) overall survival as determined by univariate and multivariate analyses. (G-I) Kaplan-Meier plots comparing percent survival between low and high expression cohorts in intestinal and diffuse-type GC for membrane receptors; (G) GPBAR1, (H) S1PR2, and (I) CHRM2; significance was calculated based of Cox-Mantel regression for univariate analyses, and proportional Cox regression for multivariate analyses. Statistics: *P < 0.05, **P < 0.01, ***P < 0.001, and I Z-score I > 1.96.Figure 2.: BAR expression is associated with distinct GC molecular signatures and subtypes. (A) Heat map of expression Z-scores across the four TCGA-Molecular GC subtypes and their signatures; microsatellite instability (MSI), Epstein-Barr Virus (EBV), Chromosomal instability (CIN), and Genomic stable (GS). (B-G) BAR expression stratified by Lei-Molecular subtypes; Proliferative (n = 70), Metabolic (n = 40), and Invasive (n = 51), from GSE15459 for (B) PXR, (C) CAR, (D) FXR, (E) GPBAR1, (F) S1PR2, and (G) CHRM2. Significance was computed by one-way ANOVA with multiple t-tests corrected for false discovery rate (FDR < 0.05) by Benjamini-Krieger-Yekutieli two-stage method. Statistics: *P < 0.05, **P < 0.01, ***P < 0.001, and I Z-score I > 1.96.Figure 3.: Ingenuity pathway analysis of membrane BARs in GC. (A) Analysis of commonly upregulated disease and function pathways associated with membrane BARs. (B) Common activated and inhibited upstream regulators associated with each BAR. Dark shades indicate common associations between all three membrane BARs. Lighter hues indicate common gene associations between two membrane BARs. Red indicates predicted activity; blue indicates predicted inhibition. (C) Pathway analysis of downstream effectors common between all at least two membrane BARs. Pink (EGFR) indicates common convergence between all three BARs. All disease & function associations, pathways, and upstream regulators are significant (Z-score < -1.96 or > 1.96). Data was sourced from TCGA-FH dataset.