A microRNA-based liquid biopsy signature for the early detection of esophageal squamous cell carcinoma: a retrospective, prospective and multicenter study

Abstract

Background: Currently, there is no clinically relevant non-invasive biomarker for early detection of esophageal squamous cell carcinoma (ESCC). Herein, we established and evaluated a circulating microRNA (miRNA)-based signature for the early detection of ESCC using a systematic genome-wide miRNA expression profiling analysis. Methods: We performed miRNA candidate discovery using three ESCC tissue miRNA datasets (n = 108, 238, and 216) and the candidate miRNAs were confirmed in tissue specimens (n = 64) by qRT-PCR. Using a serum training cohort (n = 408), we conducted multivariate logistic regression analysis to develop an ESCC circulating miRNA signature and the signature was subsequently validated in two independent retrospective and two prospective cohorts. Results: We identified eighteen initial miRNA candidates from three miRNA expression datasets (n = 108, 238, and 216) and subsequently validated their expression in ESCC tissues. We thereafter confirmed the overexpression of 8 miRNAs (miR-103, miR-106b, miR-151, miR-17, miR-181a, miR-21, miR-25, and miR-93) in serum specimens. Using a serum training cohort, we developed a circulating miRNA signature (AUC:0.83 [95%CI:0.79–0.87]) and the diagnostic performance of the miRNA signature was confirmed in two independent validation cohorts (n = 126, AUC:0.80 [95%CI:0.69–0.91]; and n = 165, AUC:0.89 [95%CI:0.83–0.94]). Finally, we demonstrated the diagnostic performance of the 8-miRNA signature in two prospective cohorts (n = 185, AUC:0.92, [95%CI:0.87–0.96]); and (n = 188, AUC:0.93, [95%CI:0.88–0.97]). Importantly, the 8-miRNA signature was superior to current clinical serological markers in discriminating early stage ESCC patients from healthy controls (p < 0.001). Conclusions: We have developed a novel and robust circulating miRNA-based signature for early detection of ESCC, which was successfully validated in multiple retrospective and prospective multinational, multicenter cohorts.