Reduced representation DNA sequencing of Barrett’s Oesophagus identifies patients at risk of Oesophagus adenocarcinoma development

Introduction: Barrett's oesophagus (BO) is a precursor lesion for oesophageal adenocarcinoma (OAC). BO diagnosis significantly increases the risk of OAC diagnosis, however the absolute risk of BO to OAC progression is 0.3% per year. As a result, patients with BO diagnosis enter clinical surveillance that includes regular oesophagogastroduodenoscopy (OGD) with biopsies and pictures taken.

Objectives: Since most BO patients do not progress to cancer, additional risk stratification biomarkers of BO to OAC progression are essential. Using a retrospective cohort of BO cases, this study aims to define genetic prognostic factors showing likelihood of progression from BO to OAC.

Methodology: We collected 22 patients' histological data out of a cohort from the interdisciplinary endoscopic centre of the University Hospital Mannheim between 2004-2023. In the pilot phase of the study, we performed mutREAD DNA sequencing of 44 matched biopsies from those 22 cases. They composed of 22 control biopsies and 8 low grade dysplasia (LGD), 10 high grade dysplasia (HGD), and 4 in-situ carcinomas. The biopsies were stored in formalin-fixed paraffin embedded (FFPE) blocks for up to 15 years. Using a novel bioinformatics pipeline we identified genetic features describing each sample biopsy. Next, we developed an elastic-net regression model that stratifies HGD/carcinoma biopsies from control.

Results: Sufficient DNA was extracted out of all FFPE biopsies. After sequencing and data processing six biopsies had insufficient sequencing coverage. The elastic-net model was trained on 16 normal samples and 12 HGD/carcinoma samples using 79 genome wide variables. The model identified 13 highly informative features that could distinguish HGD/carcinoma from normal biopsies with sensitivity=1 (CI=[1, 1]), specificity=0.838 (CI=[0.667, 1]), positive predictive value of 0.794 (CI=[0.556, 1]) and area under receiver curve (AUC) value of 0.912. Application of the model on the LGD samples with sufficient data quality recovered all patients that eventually progressed to HGD or cancer.

Conclusion: mutREAD-based sequencing of FFPE material provides excellent diagnosis potential of BO biopsies. Additional analysis of method’s risk stratification potential is required due to current low case count of LGD samples. Future work involves cohort expansion to BO cases without dysplasia and analysis of additional genetic features identified in mutREAD data.

Präsentiert in der Sitzung: Therapierefraktäre Refluxkrankheit und Barrett

Donnerstag, 18. September 2025, 14:30 – 16:00, Theodor Billroth

Publication History

Article published online:
04 September 2025