Integrative Analysis of eQTL Genes Reveals Key Biomarkers and Mechanisms for Early Diagnosis of Pancreatic Ductal Adenocarcinoma.

Background: Pancreatic ductal adenocarcinoma (PDAC) is a highly lethal malignancy with a dismal 5-year survival rate, largely due to the absence of reliable biomarkers for early detection. The molecular mechanisms underpinning PDAC pathogenesis remain incompletely understood, highlighting the urgent need for novel diagnostic strategies. Objective: This study aimed to integrate eQTL-driven Mendelian randomization (MR) with transcriptomic and genome-wide association data to identify causal PDAC-associated genes and construct a diagnostic nomogram based on 5 hub genes (CTSC, SMYD3, MFGE8, IGFBP7, POC1B) for early detection of pancreatic ductal adenocarcinoma (PDAC). Methods: Transcriptomic data from GSE62165 and GSE25471 were retrieved from the Gene Expression Omnibus (GEO) and processed for differential expression using LIMMA and GEO2R, followed by batch correction and weighted gene co-expression network analysis (WGCNA). Summary-level eQTL statistics were obtained from OpenGWAS, and GWAS data included over 5000 PDAC cases. MR analysis was performed using inverse variance weighted (IVW) as the primary approach, supplemented with MR-Egger, weighted median, weighted mode, and MR-PRESSO. Instrument strength, pleiotropy, and heterogeneity were assessed via F-statistics, Egger intercept, and Cochran's Results: Five eQTL-associated hub genes- Conclusions: This study presents a multi-omics, MR-informed framework for identifying eQTL-regulated biomarkers of PDAC. The identified hub genes offer promising avenues for early detection, while the mechanistic mapping of the PI3K-Akt pathway provides translational insights. These findings warrant further validation in clinical and experimental settings and hold potential to reshape PDAC diagnostic strategies.Pancreatic ductal adenocarcinoma (PDAC) remains a formidable clinical challenge due to its aggressive nature and lack of effective early diagnostic biomarkers. To address this, we integrated transcriptomic data, genome-wide association studies (GWAS), and expression quantitative trait loci (eQTL) information using Mendelian randomization (MR) to identify genes causally associated with PDAC risk. Differentially expressed genes were identified across 2 GEO datasets (GSE62165, GSE25471) and prioritized using weighted gene co-expression network analysis (WGCNA). MR analysis employing IVW, MR-Egger, weighted median, and MR-PRESSO identified 5 hub genes-CTSC, SMYD3, MFGE8, IGFBP7, and POC1B-as significant causal drivers of PDAC. These genes were incorporated into a diagnostic model constructed using machine learning approaches (random forest, SVM-RFE, Lasso), which achieved strong classification performance (AUC > 0.85) and excellent calibration (C-index = 0.92). Functional enrichment and protein-protein interaction analyses revealed that CTSC regulates the ECM-integrin-PI3K-Akt signaling pathway, contributing to tumor cell proliferation and survival. The findings establish a multi-omics-based biomarker panel with strong diagnostic utility and mechanistic relevance, suggesting a potential framework for future translational validation in clinical cohorts.