Abstract
BACKGROUND: Accumulating evidences indicate that the specific alternative splicing (AS) events are linked to the occurrence and prognosis of gastric cancer (GC). Nevertheless, the impact of AS is still unclear and needed to further elucidation. METHODS: The expression profile of GC and normal samples were downloaded from TCGA. AS events were achieved from SpliceSeq database. Cox regression together with LASSO analysis were employed to identify survival-associated AS events (SASEs) and calculate risk scores. PPI and pathway enrichment analysis were implemented to determine the function and pathways of these genes. Kaplan-Meier (K-M) analysis and Receiver Operating Characteristic Curves were used to evaluate the clinical significance of genes of SASEs. Q-PCR were applied to validate the hub genes on the survival prognosis in 47 GC samples. Drug sensitivity and immune cell infiltration analysis were conducted. RESULTS: In total, 48 140 AS events in 10 610 genes from 361 GC and 31 normal samples were analyzed. Through univariate Cox regression, 855 SASEs in 763 genes were screened out. Further, these SASEs were analyzed by PPI and 17 hub genes were identified. Meanwhile, using Lasso and multivariate Cox regression analysis, 135 SASEs in 132 genes related to 7 AS forms were further screened and a GC prognostic model was constructed. K-M curves indicates that high-risk group has poorer prognosis. And the nomogram analysis on the basis of the multivariate Cox analysis was disclosed the interrelationships between 7 AS forms and clinical parameters in the model. Five key genes were then screened out by PPI analysis and Differential Expression Gene analysis based on TCGA and Combined-dataset, namely STAT3, RAD51B, SOCS2, POLE2 and TSR1. The expression levels of AS in STAT3, RAD51B, SOCS2, POLE2 and TSR1 were all significantly correlated with survival by qPCR verification. Nineteen drugs were sensitized to high-risk patients and eight immune cells showed significantly different infiltration between the STAD and normal groups. CONCLUSIONS: In this research, the prognostic model constructed by SASEs can be applied to predict the prognosis of GC patients and the selected key genes are expected to become new biomarkers and therapeutical targets for GC treatment.