br Protein protein interaction PPI network
3.5. Protein-protein interaction (PPI) network construction and module screening
To obtain the interaction between the target genes, the PPI network was explored and visualized using Cytoscape. The nodes with high in-teractions were described as the hub. In total, 170 edges and 294 nodes comprised the network. The top three significant hub proteins were CRK (interaction = 24), YWHAG (interaction = 19), and STAT3 (in-teraction = 18) (Fig. 3B). Based on the PPI network, the analysis was conducted using the plug-in MCODE. The top 3 significant modules were selected and the functional annotation of the genes involved in the modules was analyzed. Results of the enrichment analysis showed that the main associations of these three modules were with EPHA-mediated growth cone collapse, axon repulsion, clathrin-mediated and macro-autophagy (Fig. 3C).
3.6. Prognostic nomogram for OS
A nomogram was constructed by integrating of each independent factor with statistical significance. Fig. 4A shows the association of the EPZ-6438 levels of these five miRNAs with the OS of patients in the primary cohort. The C-index for OS prediction was 0.72 (95% CI, 0.64 to 0.78). According to the calibration plot, the prediction on the post-surgery three- or five-year survival probability of patients with GC provided by the nomogram was consistent with the actual observation (Fig. 4B, C). In the validation cohort, the C-index of the nomogram for
Fig. 1. MiRNAs were associated with overall survival in gastric cancer patients by using Kaplan Meier curve and Log-rank test. A. Volcano plot of diﬀerentially expressed miRNAs. The red dot represents up-regulated miRNA, and green dot represents down-regulated miRNA. B–F. The patients were stratified into high expression group and low expression group according to median of each miRNA. G. Kaplan-Meier curve for the five-miRNA signature in gastric cancer patients. The patients were stratified into high risk group and low risk group based on median. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
3.7. Comparison of predictive accuracy between the nomogram and a single independent factor
stage had the highest hazard ratio for OS. The GC prognosis prediction capability was compared between T stage and the nomogram. The OS prediction C-index of the T stage (0.59) was lower than that of the nomogram (0.72; P < 0.001).
Based on the data presented in Table 3, among all the factors, T
Table 2 Association of five miRNAs and clinical features.
Univariate and multivariate Cox regression analysis in CC patients.
HR P value
HR P value
Lymph node status
Histology type (diﬀuse vs intestinal)
Gender (male vs female)
3.8. Comparison of predictive accuracy for OS between nomogram and conventional staging systems
According to the results shown in Fig. 5, the AJCC 7th edition presented good performance in stratifying prognosis for patients with GC in stages I and II. Nevertheless, the stratification performance of the AJCC 7th edition was not that satisfactory when patients between stages II and III were considered. More accurate OS prediction in the primary cohort was presented by the proposed nomogram. The C-index by AJCC (which was 0.60) was lower than that of the proposed no-mogram (0.72, P < 0.001). Therefore, the nomogram can be used as an eﬀective tool for OS prediction in primary-cohort patients with GC.
Despite the decreased frequency of occurrence of GC in developed countries, it still persists as a common disease in many areas of the world. Eastern Asia, Eastern Europe, and South America have the highest rates of occurrence, with 42% of worldwide cases being re-ported in China (Torre et al., 2015). GC prognosis would have improved remarkably if tumor behavior could have been assessed reliably at the time of initial diagnosis. Hence, it is essential to understand the mole-cular mechanisms of the disease and identify useful biomarkers of GC