Selection and Application of Tissue microRNAs for Nonendoscopic Diagnosis of Barrett's Esophagus.

Abstract

Background & Aims: MicroRNA (miRNA) is highly stable in biospecimens and provides tissue-specific profiles, making it a useful biomarker of carcinogenesis. We aimed to discover a set of miRNAs that could accurately discriminate Barrett’s esophagus (BE) from normal esophageal tissue and to test its diagnostic accuracy when applied to samples collected by a non-invasive esophageal cell sampling device. Methods: We analyzed miRNA expression profiles of 2 independent sets of esophageal biopsy tissues collected during endoscopy from 38 patients with BE tissues and 90 patients with non-BE esophagus (controls) using Agilent microarray and Nanostring counter assays. Consistently upregulated miRNAs were quantified by real-time PCR in esophageal tissues collected by Cytosponge from patients with BE or without BE. miRNAs from plasmids and anti-sense oligonucleotides were expressed in NES normal esophageal squamous cells and effects on proliferation and gene expression patterns were analyzed. Results: We identified 15 miRNAs that were significantly upregulated in BE vs control tissues. Of these, 11 (MIR215, MIR194, MIR 192, MIR196a, MIR199b, MIR10a, MIR145, MIR181a, MIR30a, MIR7, MIR199a) were validated in Cytosponge samples. The miRNAs with the greatest increases in BE tissues (7.9-fold increase inexpression or more, P<0.0001: MIR196a, MIR192, MIR194, and MIR215) each identified BE vs control tissues with area under the curve (AUC) values of 0.82 or more. We developed an optimized multivariable logistic regression model based on expression levels of 6 miRNAs (MIR7, MIR30a, MIR181a, MIR192, MIR196a, and MIR199a) that identified patients with BE with an AUC value of 0.89, 86.2% sensitivity, and 91.6% specificity. Expression level of MIR192, MIR196a, MIR199a, combined with Trefoil Factor 3 (TFF3), identified patients with BE with an AUC of 0.93, 93.1% sensitivity, and 93.7% specificity. Hypo-methylation was observed in the promoter region of the highly upregulated cluster MIR192-194. Overexpression of these miRNAs in NES cells increased their proliferation, via GRHL3 and PTEN signaling. Conclusions: In analyses of miRNA expression patterns of BE vs non-BE tissues, we identified a profile that can identify Cytosponge samples from patients with BE with an AUC of 0.93. Expression of MIR194 is increased in BE samples via epigenetic mechanisms that might be involved in BE pathogenesis.