Prognostic implications of microRNA-21 overexpression in pancreatic ductal adenocarcinoma: an international multicenter study of 686 patients

Despite progress in genomic characterization, no single prognostic marker that can be evaluated using an easy-to-perform and relatively inexpensive method is available for pancreatic ductal adenocarcinoma (PDAC). MicroRNAs, which are stable, tumor-and tissue-specific molecules, are potentially ideal biomarkers, and we es-tablished an inter-laboratory validated method to investigate miR-21 as a prognostic biomarker in PDAC. The study samples of PDAC patients were recruited from a test cohort of Glasgow (n = 189) and three validation cohorts of Pisa (n = 69), Sydney (n = 249), and International Cancer Genome Consortium (ICGC) (n = 249). Tissue microarrays were used for miR-21 staining by chromogenic in situ hybridization (CISH). The patients were subdivided into no/low and high miR-21 staining groups using a specific histoscore. Furthermore, miR-21 staining was evaluated against clinicopathological variables and follow-up data by Fisher/log-rank test and Cox proportional models. The prognostic variables found to be significant in univariate analysis (P value < 0.10) were included in multivariate analysis in a backward-stepwise fashion. MiR-21 expression was cytoplasmic, with more consistent staining in the malignant ductal epithelium than in the stroma. The expression of miR-21 was significantly associated with tumor size and lymph node metastasis, whereas no association was observed with other clinicopathological variables. High miR-21 staining (histoscore =45 [median score]) was an independent predictor of survival in the Glasgow test cohort (HR 2.37, 95% CI: 1.42-3.96, P < 0.0001) and three validation cohorts (Pisa, HR 2.03, 95% CI: 1.21-3.39, P = 0.007; Sydney, HR 2.58, 95% CI (1.21-3.39), P < 0.0001; and ICGC, HR 3.34, 95% CI: 2.07-5.84, P = 0.002) when adjusted for clinical variables in a multivariate model. In comparison to the patients with low miR-21, the patients with high miR-21 expression had significant increase in OS as they benefit from gemcitabine-based adjuvant chemotherapy (Glasgow 16.5 months [with chemotherapy] vs 10.5 months [without chemotherapy]); Sydney 25.0 vs 10.6; ICGC 25.2 vs 11.9. These results indicated that miR-21 is a predictor of survival, prompting prospective trials. Evaluation of miR-21 offers new opportunities for the stratification of patients with PDAC and might facilitate the implementa-tion of clinical management and therapeutic interventions for this devastating disease.