Comprehensive Analysis of Alternative Splicing Signature in Gastric Cancer Prognosis Based on The Cancer Genome Atlas (TCGA) and SpliceSeq Databases

Abstract

BACKGROUND Increasing evidence suggests that the alternative splicing (AS) signature plays a role in the carcinogenesis and prognosis of various cancers. However, the prognostic role of AS in gastric cancer is not clear and needs to be clarified. MATERIAL AND METHODS To identify the differentially expressed AS (DEAS) events, we performed a differential expression analysis between normal and tumor tissue. The DEAS event was further applied to construct a prognostic signature by performing univariate Cox regression analysis and least absolute shrinkage and selection operator (LASSO) analysis. The Kaplan-Meier curve analysis and receiver operating characteristic curve (ROC) analysis were used to evaluate the prognostic value of the AS signature. In addition, the network of the splicing events with splicing factors was constructed using the Cytoscape software. RESULTS A total of 30 005 alternative splicing (AS) events with 372 patients were retrieved from the SpliceSeq database and TCGA database. By performing differential expression analysis, a total of 419 alternative splicing events were screened out, including 56 upregulated and 363 downregulated. We further constructed an AS-related prognostic signature by conducting a series bioinformatics analyses. Moreover, we identified that the AS signature could serve as an independent predictor for the prognosis of GC. We also found that AS signature had a more robust and precise efficacy for prognostic prediction in GC patients. Interestingly, the areas under 3- and 5-year survival curves are similar, both of which are greater than 1-year survival curve, suggesting that the long-term predictive accuracy of our prognostic model built upon AS signature is superior. CONCLUSIONS We performed a comprehensive analysis of overall prognostic-associated AS events concerning GC and constructed a prognostic model to predict the long-term prognostic survival outcomes in GC patients. We also developed a network of splicing events with splicing factors to reveal new potential molecular diagnostic biomarkers and therapeutic targets for GC patients.

Figure 1

The flowchart to illustrate the workflow and the novel findings of this study.

Figure 2

The landscape of aberrant alternative splicing events in GC cohort. (A) Upset plot of ASEs for the 7 different patterns, including AA as alternate acceptor site, AT as alternate terminator, ES as exon skip, AD as alternate donor site, AP as alternate promoter, ME as mutually exclusive exons, and RI as retained intron in the GC. (B) The expression heatmap of differentially expressed alternative splicing events (DEAS). (C) Volcano plots of the distribution of DEAS in the GC dataset. Red dots represent upregulated alternative splicing events whereas green dots represent downregulated ones.

Figure 3

Forest plots for univariate Cox analysis of the survival associated AS events in GC cohort with hazard ratios and 95% confidence intervals. The color scale beside indicates the P values.

Figure 4

Identification of prognostic model based on the AS events in GC. (A) Evaluation of OS-SEs with the coefficients calculated in the LASSO regression. (B) Plots of the cross-validation error rates. The dashes signify the value of the minimal error and greater λ value. (C) Survival curve of prognostic model in patients with GC for 2 groups (high-risk group vs. low-risk group) based on median-cut of risk score calculated by evaluation of OS-SEs with the coefficients calculated in LASSO regression.

Figure 5

mRNA signature risk score distribution, heatmap of the mRNA expression profiles. Rows represent mRNAs, and columns represent patients. (A) The risk score curve, red dots show high-risk samples and while green dots show low-risk samples. (B) Distribution of patients’ survival status and overall survival (OS) times classified with risk scores, red dots indicate dead while green dots indicate alive. (C) Heatmap displays splicing pattern of the mRNA signatures. Color transition from green to red indicates the increasing PSI score of corresponding genes expression from low to high.

Figure 6

Evaluation of the prognostic and clinical factors in GC. (A) Receiver operating characteristic (ROC) analysis of the sensitivity and specificity of the survival time by the risk model in 1 year, 3 years, and 5 years. The black line evaluates whether the difference of 2 estimated AUCs at each timepoint is statistically significant. (B) Receiver operating characteristic (ROC) analysis of the sensitivity and specificity of the risk model and clinical traits, including age, sex, grade, and TNM stage. The univariate (C) and multivariate (D) Cox regression analysis for the risk model score and clinical traits, including age, sex, grade, and TNM stage.

Figure 7

Receiver operating characteristic (ROC) analysis of the sensitivity and specificity of the risk model and prognostic DEAS expression.

Figure 8

Correlation network between expression of survival AS factors and PSI values of AS genes generated using Cytoscape.

Figure 9

Survival analysis of the prognostic AS event associated with GC.

Figure 10

Gene set enrichment analysis for the prognostic AS event with gene sets in “C2: Canonical pathways” access from MSigDB database. Significant enrichment results include “autophagy”, “cellular response to DNA damage stimulus”, “DNA repair”, “regulation of cell morphogenesis”, and “response to hypoxia”. AS, alternative splicing; MSigDB, Molecular Signatures Database.