BACKGROUND: The tissue-specificity and robustness of miRNAs may aid risk prediction in individuals diagnosed with Barrett's esophagus. As a initial step, we assessed whether miRNAs can positively distinguish esophageal adenocarcinoma (EA) from the precursor metaplasia Barrett's esophagus (BE).

METHODS: In a case-control study of 150 EAs frequency-matched to 148 BE cases, we quantitated expression of 800 human miRNAs in FFPE tissue RNA using NanoString miRNAv2. We tested differences in detection by case group using the chi-square test and differences in expression using the Wilcoxon rank-sum test. Bonferroni-corrected statistical significance threshold was set at P<6.25E-05. Sensitivity and specificity were assessed for the most significant miRNAs using five-fold cross-validation.

RESULTS: We observed 46 distinct miRNAs significantly increased in EA compared with BE; 35 of which remained when restricted to T1b and T2 malignancies. Three miRNAs (miR-663b, miR-421, miR-502-5p) were detected in >80% EA, but <20% of BE. Seven miRNAs (miR-4286, miR-630, miR-575, miR-494, miR-320e, miR-4488, miR-4508) exhibited the most extreme differences in expression with >5 fold-increases. Using five-fold cross validation, we repeated feature (miR) selection and case-control prediction and computed performance criteria. Each of the five folds selected the same top ten miRs which, together, provided 98% sensitivity and 95% specificity.

CONCLUSIONS: This study provides evidence that tissue miRNA profiles can discriminate EA from BE. This large analysis has identified miRNAs that merit further investigation in relation to pathogenesis and diagnosis of EA.

IMPACT: These candidate miRNAs may provide a means for improved risk stratification and more cost-effective surveillance.