Background: Esophageal cancer (EC) is the sixth leading cause of cancer-related deaths worldwide. Despite the fact that esophageal squamous cell carcinoma (ESCC) accounts for about 80% of all ECs, currently there are no established serological markers available for its early diagnosis. While growing evidence indicates that miRNAs are promising substrates for noninvasive cancer diagnostics, the expression levels of individual miRNAs appear to have limited clinical usefulness as disease biomarkers. In contrast to microarray-based platforms, recent advancements in whole transcriptomic RNA-Seq approaches have allowed comprehensive miRNA expression profiling - an important requisite for identification and development of novel and robust miRNA biomarkers for various diseases, including ESCC.
Aim: We aimed to develop and validate a circulating miRNA-based diagnostic panel for ESCC by systematic and comprehensive miRNA expression analysis in large, multiple, independent ESCC patient cohorts..
Methods: A RNA-Seq dataset (n=207) was used to identify initial miRNA candidates, which were subsequently validated in two large datasets (n=238 and 216). The expression of these candidates was validated by qRT-PCR in 50 paired cancer and normal tissues. The robustness of the candidates was assessed in agematched serum samples (ESCC and healthy, n=247), which led to the establishment of a circulating miRNA-panel. Two independent patient cohorts (n=553 and 178) were analyzed to assess and validate the diagnostic performance of this noninvasive miRNA panel.
Results: We initially identified a panel of 15 most differentially expressed miRNAs from the RNASeq data, of which, six were successfully validated in two independent in-silico datasets and 50 pairs of matched cancer and normal tissues. Remarkably, a combination panel of these 6 miRNAs could significantly discriminate ESCC patients from normal subjects in all cohorts (AUC=1.0). Next, we assessed the expression of these tissue-candidates in serum specimens from an ESCC patient cohort, and established and optimized a 4-miRNA panel (miR-21, -93, -106b and -17) with an impressive AUC value of 0.92. The diagnostic performance of this miRNA signature was evaluated in two large independent validation cohorts with a resulting AUC of 0.91 in both cohorts. Furthermore, our data revealed that this miRNA panel could distinguish early stage ESCC patients from healthy controls with a sensitivity of 72%, which is much superior to currently used serological tumor markers.
Conclusions: Using a comprehensive biomarker discovery and validation approach in potentially the largest cohort of ESCC patients analyzed to date (n=1078), we have developed and validated a novel and robust miRNA-based panel for the early detection of ESCC, which has the potential for noninvasive diagnostics of ESCC patients in future.