br Table br Clinical characteristics of gastric
Clinical characteristics of gastric cancer patients.
Variables Case, n (%)
Age at diagnosis
NA, non available.
derived from regression results. Diﬀerences with a P-value < 0.05 were considered statistically significant.
3.1. Identification of diﬀerentially expressed miRNAs in GC
This study included 364 selected samples as mentioned previously. As shown in Table 1, the clinical characteristics of these samples con-tained information regarding gender, age, histological type, T stage, clinical stage, lymph node status and metastasis. Based on the cut-oﬀ criteria (P < 0.05 and log2FC > 2.0), 129 diﬀerentially expressed miRNAs, including 116 up-regulated and 13 down-regulated miRNAs were identified in these sample groups (Fig. 1A). The diﬀerentially expressed miRNA patterns were screened based on cancerous tissue and normal tissue.
3.2. Identification of five miRNAs associated with overall survival (OS) in GC
To test the potential association of miRNAs with the survival per-formance of patients with GC, the log-rank test and Kaplan-Meier curve were used for the assessment of this association. Results showed a po-sitive correlation between the OS and four miRNAs (miR-106b, miR-141, miR-145 and miR-20a), as well as a negative correlation between OS and one miRNA (miR-135b) (Fig. 1B–F). Furthermore, the associa-tions between clinical characteristics of samples and these five miRNAs were evaluated (Table 2). Results showed association of miR-20a with histology, miR-106b with histology, T stage, lymph node status, and Gene 699 (2019) 125–134
age, miR-135b with age, metastasis and histology, miR-141 with his-tology, T stage, clinical stage, lymph node status, metastasis and age, and a significant association of miR-145 with histology.
3.3. Prognostic value of the five-miRNAs signature risk score in GC
A prognostic signature was established based on the integration of these five miRNA AR-13324 profiles, as well as their corresponding regression coeﬃcients. Subsequently, patients were ranked by risk score calculated using the prognostic signature. Based on the median risk scores, these patients were divided into two groups: low risk (n = 154) and high risk (n = 153) (Fig. 1G). Univariate and multi-variate analyses were conducted to identify OS-correlated character-istics. In univariate analysis, metastasis (HR = 2.673, P = 0.015), lymph node status (HR = 1.632, P = 0.0), clinical stage (HR = 1.454, P = 0.118), and T stage (HR = 2.639, P < 0.001), were associated with the OS of patients with GC and the prognostic signature of these five miRNAs (HR = 2.762, P < 0.001). In contrast, according to the results of the multivariate analysis, the prognostic signature of these five miRNAs (HR = 2.430, P = 0.001) was demonstrated as an in-dependent prognostic factor for patients with GC (Table 3).
3.4. Target prediction and function analysis
Online analytical instruments, including miWalk, PicTar, miRDB and TargetScan were used to predict the targets of these five miRNAs (miR-106b, miR-135b, miR-141, miR-145 and miR-20a). The number of identified overlapping target genes of miR-20a, miR-145, miR-141, miR-135b and miR-106b were respectively 15, 18, 49, 13 and 176, respectively (Fig. 2A). Next, enrichment analysis was performed to in-vestigate the biological functions of these target genes. Significant en-richment of Kyoto Encyclopedia of Genes and Genomes (KEGG) path-ways was observed, for example, the Hippo, Wnt, TGF-β, cell cycle, AMPK and MAPK signaling pathways (Fig. 2B). The KEGG disease terms were primarily enriched in cancers, cardiovascular diseases and urinary system diseases (Fig. 2C). In addition, the enriched gene ontology (GO) biological process (BP) terms were mainly associated with cell mor-phogenesis during diﬀerentiation, regulation of positive epithelial cell migration and positive apoptotic process (Fig. 3A).