A microRNA Signature Identifies Patients at Risk of Barrett Esophagus Progression to Dysplasia and Cancer

Background Progression of Barrett esophagus (BE) to esophageal adenocarcinoma occurs among a minority of BE patients. To date, BE behavior cannot be predicted on the basis of histologic features. Aims We compared BE samples that did not develop dysplasia or carcinoma upon follow-up of ≥ 7 years (BE nonprogressed [BEN]) with BE samples that developed carcinoma upon follow-up of 3 to 4 years (BE progressed [BEP]). Methods The NanoString nCounter miRNA assay was used to profile 24 biopsy samples of BE, including 13 BENs and 11 BEPs. Fifteen samples were randomly selected for miRNA prediction model training; nine were randomly selected for miRNA validation. Results Unpaired t tests with Welch’s correction were performed on 800 measured miRNAs to identify the most differentially expressed miRNAs for cases of BEN and BEP. The top 12 miRNAs ( P < .003) were selected for principal component analyses: miR-1278, miR-1301, miR-1304-5p, miR-517b-3p, miR-584-5p, miR-599, miR-103a-3p, miR-1197, miR-1256, miR-509–3-5p, miR-544b, miR-802. The 12-miRNA signature was first self-validated on the training dataset, resulting in 7 out of the 7 BEP samples being classified as BEP (100% sensitivity) and 7 out of the 8 BEN samples being classified as BEN (87.5% specificity). Upon validation, 4 out of the 4 BEP samples were classified as BEP (100% sensitivity) and 4 out of the 5 BEN samples were classified as BEN (80% specificity). Twenty-four samples were evaluated, and 22 cases were correctly classified. Overall accuracy was 91.67%. Conclusion Using miRNA profiling, we have identified a 12-miRNA signature able to reliably differentiate cases of BEN from BEP.