Integrating plasma-derived circulating proteins with multi-omics approaches to identify potential therapeutic targets for esophageal adenocarcinoma.

Abstract

Background: Esophageal adenocarcinoma (EAC) is a complex disease often preceded by Barrett's esophagus. Objective: This study aimed to identify potential drug targets and support early prevention and treatment. Method: We integrated cis-protein quantitative trait loci (cis-pQTLs) and genome-wide association study (GWAS) data from 35 559 individuals using Mendelian randomization to infer causal relationships, with FDR correction applied. To assess tissue relevance, the source cis-pQTL instruments were examined across three GTEx v8 esophageal tissues, followed by tissue-informed single-variant Wald analyses of CD55 and NDST1. Gene ontology (GO) and KEGG pathway analyses, along with protein-protein interaction (PPI) network analysis, were used to explore biological functions and interactions. Overlapping genes were identified through differential expression analysis using the GEO database. Key feature genes were screened using XGBoost and Lasso regression, and diagnostic performance was evaluated by ROC curve AUC in both training and validation sets. Immune infiltration analysis explored the tumor microenvironment, and molecular docking assessed drug potential. Results: We identified 28 plasma proteins causally linked to EAC. KEGG analysis revealed enrichment in oxidative stress and inflammation-related pathways. Transcriptomic validation confirmed three overlapping genes. CD55 and NDST1 emerged as key markers with strong diagnostic accuracy after Lasso and XGBoost screening. Most significant esophageal SNP-tissue associations were directionally concordant with the corresponding plasma pQTL effects, and the tissue-specific Wald estimates for CD55 and NDST1 were consistent with the primary plasma-based findings. Immune infiltration analysis showed significant changes in immune cell composition in EAC, particularly natural killer cells, plasma cells, and mast cells. Drug prediction and docking supported the druggability of these targets. Conclusion: CD55 and NDST1 are promising therapeutic candidates for EAC. Drugs based on these genetic findings could improve clinical trial success rates and reduce costs.