Search

AT-rich interactive domain 1A (ARID1A), which is a tumor suppressor gene, is frequently mutated in Epstein-Barr virus-positive gastric cancer [EBV (+) GC]. While most ARID1A mutations in GC are truncating mutations, leading to loss of ARID1A protein expression, epigenetic modifications appear to contribute to ARID1A deficiency in EBV (+) GC harboring wild-type ARID1A. Based on the significant role of epigenetic modifications in EBV (+) GC that contributes to ARID1A deficiency, the methylation status of ARID1A was evaluated in EBV-infected cells and GC patients using a publicly available microarray and the Cancer Genome Atlas (TCGA) database. EBV-encoded miRNAs that potentially target ARID1A were identified as an additional epigenetic modulator by computational prediction. In vitro experiments were conducted to evaluate how EBV-encoded miRNAs affected ARID1A mRNA and protein levels. In clinical GC samples, the expression of predicted miRNAs and ARID1A and the mutation status of ARID1A was evaluated. As results, ARID1A was not hypermethylated in EBV (+) GC samples or EBV-infected GC cells. EBV infection did not alter ARID1A mRNA levels, suggesting that ARID1A protein deficiency was caused by post-transcriptional gene silencing in ARID1A-WT EBV (+) GC. Overexpression of miR-BART11-3p and miR-BART12, which were identified as miRNAs that potentially bind ARID1A, suppressed ARID1A protein expression in MKN7 and NCI-N87 cells. Highly expressed miR-BART11-3p and miR-BART12 were correlated with decreased ARID1A levels in GC tumors which did not harbor ARID1A mutations. The present findings revealed that ARID1A expression was epigenetically regulated by miR-BART11-3p and miR-BART12 in EBV (+) GC.

Adjuvant chemotherapy is considered for patients with stage II colorectal cancer (CRC) characterized by poor prognostic clinicopathological features; however, current stratification algorithms remain inadequate for identifying high-risk patients. To develop prognostic assays, we conducted a step-wise screening and validation strategy using nine cohorts of stage II patients based on multiple platforms, including microarray, RNA-sequencing (RNA-seq) and immunohistochemistry (IHC) on formalin-fixed paraffin-embedded (FFPE) tissues. Four microarray datasets (total n = 458) were used as the discovery set to screen for single genes associated with postoperative recurrence. Prognostic values of candidate genes were evaluated in three independent microarray/RNA-seq validation cohorts (n = 89, n = 93 and n = 183, respectively), and then IHC for KRT17 was conducted in two independent FFPE series (n = 110 and n = 44, respectively). We found that high levels of KRT17 transcript expression were significantly associated with poor relapse-free survival (RFS) not only in the discovery set, but also in three validation cohorts, and its prognostic impact was independent of conventional factors by multivariate analyses. Positive staining of KRT17 protein was significantly associated with poor RFS in two independent FFPE cohorts. KRT17 protein expression had independent prognostic impact on RFS in a multivariate model adjusted for conventional variables, including high-risk clinicopathological features. In conclusion, using nine independent cohorts consisting of 997 stage II patients, we identified and validated the expression of KRT17 transcript and KRT17 protein as a robust prognostic biomarker that can discriminate postoperative stage II patients who are at high probability of disease recurrence, providing additional prognostic stratification beyond the currently available high-risk factors.