Alternative Splicing Events in Tumor Immune Infiltration in Colorectal Cancer

Abstract

Despite extensive research, the exact mechanisms involved in colorectal cancer (CRC) etiology and pathogenesis remain unclear. This study aimed to examine the correlation between tumor-associated alternative splicing (AS) events and tumor immune infiltration (TII) in CRC. We analyzed transcriptome profiling and clinical CRC data from The Cancer Genome Atlas (TCGA) database and lists of AS-related and immune-related signatures from the SpliceSeq and Innate databases, respectively to develop and validate a risk model of differential AS events and subsequently a TII risk model. We then conducted a two-factor survival analysis to study the association between TII and AS risk and evaluated the associations between immune signatures and six types of immune cells based on the TIMER database. Subsequently, we studied the distribution of six types of TII cells in high- and low-risk groups for seven AS events and in total. We obtained the profiles of AS events/genes for 484 patients, which included 473 CRC tumor samples and 41 corresponding normal samples, and detected 22581 AS events in 8122 genes. Exon Skip (ES) (8446) and Mutually Exclusive Exons (ME) (74) exhibited the most and fewest AS events, respectively. We then classified the 433 patients with CRC into low-risk (n = 217) and high-risk (n = 216) groups based on the median risk score in different AS events. Compared with patients with low-risk scores (mortality = 11.8%), patients with high-risk scores were associated with poor overall survival (mortality = 27.6%). The risk score, cancer stage, and pathological stage (T, M, and N) were closely correlated with prognosis in patients with CRC (P < 0.001). We identified 6479 differentially expressed genes from the transcriptome profiles of CRC and intersected 468 differential immune-related signatures. High-AS-risk and high-TII-risk predicted a poor prognosis in CRC. Different AS types were associated with different TII risk characteristics. Alternate Acceptor site (AA) and Alternate Promoter (AP) events directly affected the concentration of CD4T cells, and the level of CD8T cells was closely correlated with Alternate Terminator (AT) and Exon Skip (ES) events. Thus, the concentration of CD4T and CD8T cells in the CRC immune microenvironment was not specifically modulated by AS. However, B cell, dendritic cell, macrophage, and neutrophilic cell levels were strongly correlated with AS events. These results indicate adverse associations between AS event risk levels and immune cell infiltration density. Taken together, our findings show a clear association between tumor-associated alternative splicing and immune cell infiltration events and patient outcome and could form a basis for the identification of novel markers and therapeutic targets for CRC and other cancers in the future.

Keywords: alternative splicing; colorectal cancer; immunotherapy; prognosis; tumor immune infiltration.

Figure 1

Schematic representation of the work-flow.

Figure 2

AS event profiling in CRC. (A) Representative model of seven different AS types. (B) UpSet plot of interactions between different AS types in CRC (n = 452). (C) UpSet plot of different survival-associated AS types obtained a univariate Cox regression in CRC.

Figure 3

Construction of AS model and distributions. (A–H) represents the AS events AA, AD, AP, AT, ES, ME, RI, and total events, respectively. In each figure, the top section shows patient survival data sorted according to the risk levels, the middle section shows the risk score distribution curve, and the bottom section shows the differences in levels of identified hub immune signatures between high-risk and low-risk groups as a heatmap plot.

Figure 4

Validation of AS model with survival results. (A–H) represents the AS events AA, AD, AP, AT, ES, ME, RI, and total events, respectively. In each figure, the top section shows Kaplan-Meier curves of prognostic signatures in patients with CRC, and the bottom section shows the ROC curves of prognostic predictors in patients with CRC at 3 years.

Figure 5

Assessment of AS risk model in predicting CRC prognosis. (A–H) represents the AS events AA, AD, AP, AT, ES, ME, RI, and total events, respectively. Forest plot visualizing hazard ratios of significant survival-related clinical pathological parameters including age, gender, stage, and pathological stage T, M, and N obtained by univariate (top-green point) and multivariate (bottom-red point) Cox regression analysis.

Figure 6

Identification, construction, and validation of the TII prognostic model system in CRC. (A, B) Patient survival data in the two risk groups. (C) Heatmap plot showing differences in levels of identified hub immune signature between high-risk and low-risk groups. (D) Kaplan-Meier curves of prognostic signature for CRC patients. (E) 3-year ROC curves for prognostic predictors. Univariate (F) and multivariate (G) Cox regression analyses were conducted to assess the immune-related signatures and clinic characteristics using Forest plots.

Figure 7

Integrative analysis of TII risk level and six types of tumor-infiltrating immune cells.

Figure 8

Correlation between the TII risk and AS events. (A–H) represents the AS events AA, AD, AP, AT, ES, ME, RI, and total events, respectively. We conducted a two-factor analysis to identify the risk of different AS events with TII risk in 4 groups: Group 1 (immune risk = Low; AS risk = Low), Group 2 (immune risk = High; AS risk = Low), Group 3 (immune risk = Low; AS risk = High), and Group 4 (immune risk = High; AS risk = High).

Figure 9

Assessment of AS risk in relation to six immune infiltration cell types in CRC. After determining the correlation between low- and high-risk of different AS events in each cell type, samples with a calculated P-value < 0.05 were visualized.

Figure 10

Heat map showing differences in the infiltration of six immune cell types in the low-risk and high-risk groups of different AS events. Colors from blue to red represent high to low P-values, and yellow represents P = 0.05.