Computational-based integrative methylome-transcriptome analysis reveals widespread enhancer hypomethylation in esophageal carcinoma.

Abstract

Background: Esophageal carcinoma (ESCA), comprising esophageal squamous cell carcinoma (ESCC) and esophageal adenocarcinoma (EAC), has a poor prognosis due to late diagnosis and incomplete molecular characterization. We sought to define the ESCA epigenomic landscape by integrating methylome and transcriptome data across both subtypes.

Research design and methods: TCGA-ESCA HumanMethylation450K and RNA-seq profiles (182 tumors: 95 ESCC and 87 EAC; 64 normal tissues) were analyzed using ELMER and Bioconductor workflows. Differentially methylated probes were annotated by genomic category, probe-gene coupling was tested by Spearman correlation, and a LASSO-penalized 12-CpG classifier was trained and externally validated in an independent GEO cohort (GSE68303; n = 43).

Results: ESCA showed pervasive enhancer hypomethylation, with 72.5% of the hypomethylated probes mapping to the distal enhancers. Methylation-expression coupling identified LINC02547 and PYGL as candidate effector genes, and enriched transcription factor motifs included AP-1 and NFIL3. The classifier achieved a cross-validated AUC of 0.996, approximately 0.99, sensitivity at 95% specificity, and reproducible performance across ESCC and EAC subtypes. Advanced disease stage was the dominant prognostic factor in multivariable survival models.

Conclusions: Enhancer hypomethylation is a hallmark of ESCA, nominating LINC02547 and PYGL as biomarker candidates with diagnostic/prognostic promise. Limitations include retrospective design and need for experimental validation.