A microRNA-based signature predicts local-regional failure and overall survival after pancreatic cancer resection

Abstract

Resectable pancreatic adenocarcinoma (PC) is generally managed with surgery followed by chemotherapy, but the role of postoperative chemoradiation (pCRT) is controversial. We sought to identify a microRNA (miRNA) expression profile associated with higher risk for local-regional recurrence (LRR), which might help identify patients that may benefit from pCRT. Total RNA was isolated from viable tumor from 88 patients who underwent PC resection with or without chemotherapy, but did not receive radiation. Digital miRNA expression profiling was performed and risk scores were calculated based on the expression levels of the four most significantly correlated miRNAs, and dichotomized about the median to detect correlations between risk group, LRR and overall survival (OS). Two cohorts from The Cancer Genome Atlas (TCGA) and Seoul National University (SNU) were used for validation. Patients with high-risk scores had significantly worse LRR (p = 0.001) and worse OS (p = 0.034). Two-year OS rates for the high- and low-risk groups were 27.7% and 52.2%, respectively. On multivariable analysis, the risk score remained significantly associated with LRR (p = 0.018). When validated on TCGA data, a high-risk score was associated with worse OS on univariate (p = 0.03) and multivariable analysis (p = 0.017). When validated on the SNU cohort, a high-risk score was likewise associated with worse OS (p = 0.042). We have developed a 4-miRNA molecular signature that is associated with risk of LRR and OS after PC resection and validated on two separate cohorts. This signature has the potential to select patients most likely to benefit from pCRT, and should be tested further.