Construction of colon cancer prognostic model based on TCGA SpliceSeq alternative splicing events

doi:10.3969/j.issn.1673-8640.2021.01.013

Abstract

Abstract:

Objective To establish a genome-wide alternative splicing event risk model to predict the prognosis of colon adenocarcinoma（COAD）. Methods COAD RNA-Seq data and clinical information were downloaded from the Cancer Genome Atlas（TCGA）,and alternative splicing events were obtained from TCGA SpliceSeq database. Splice factors were downloaded from the SpliceAid 2 database. Univariate Cox regression analysis was employed to determine prognostic related alternative splicing events（PASE）. Lasso regression was used to screenvariables. Multivariate Cox regression was performed to calculate risk score and constructmodels. Cytoscape Reactome FI plug-in was used to build an interactive network and find core nodes. GO and KEGG were implemented for gene function annotation and pathway analysis. Kaplan-Meier and receiver operating characteristic （ROC）curve were used for the evaluation of PASE risk model. PASE of SF and other genes were constructed to form prognosis-related interactive network. Results A total of 35 391 mRNA alternative splicing events occurred in 9 085 genes from 398 COAD patients,and 2 015 surviving PASEs occurred in 1 811 genes. In the risk model made by 8 PASE,patients were divided into high-risk group and low-risk group according to the optimal cut-off value of 0.919,and there were statistical difference between groups （P<0.001）. The area under the ROC curve of the risk model was 0.860 (one year survival rate). In univariate Cox regression analysis,tumor invasion,lymph node metastasis,distant metastasis,clinical stages and risk model were significantly negatively correlated with overall survival time（P<0.001）. After adjusting for multiple factors,the risk model still showed a significant negative correlation with the overall survival time of patients（P<0.001）. In the risk model,there was no correlation between 8 PASE and mRNA expressions of their corresponding genes（P>0.05）. Conclusion The effect of PASE on COAD prognosis is investigated by TCGA-COAD genome-wide analysis and a risk model has been constructed that can be used to predict clinical prognosis.

Key words: Colon adenocarcinoma, Alternative splicing event, Splicing factor, Prognosis

CLC Number:

R446.1

LEI Ming, GUO Mengyue, WANG Ruoying, NI Xiaomei, SHI Qiong. Construction of colon cancer prognostic model based on TCGA SpliceSeq alternative splicing events[J]. Laboratory Medicine, 2021, 36(1): 60-68.

Figures/Tables 9

References 10

[1]	BRAY F,FERLAY J,SOERJOMATARAM I,et al.Global cancer statistics 2018:GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries[J]. CA Cancer J Clin,2018,68(6):394-424.
[2]	LEE Y,RIO D C.Mechanisms and regulation of alternative Pre-mRNA splicing[J]. Annu Rev Biochem,2015,84:291-323.
[3]	BARALLE F E,GIUDICE J.Alternative splicing as a regulator of development and tissue identity[J]. Nat Rev Mol Cell Biol,2017,18(7):437-451.
[4]	YANG X,COULOMBE-HUNTINGTON J,KANG S,et al.Widespread expansion of potein interaction capabilities by alternative splicing[J]. Cell,2016,164(4):805-817.
[5]	RYAN M,WONG W C,BROWN R,et al.TCGA SpliceSeq a compendium of alternative mRNA splicing in cancer[J]. Nucleic Acids Res,2016,44(D1):D1018-D1022.
[6]	PIVA F,GIULIETTI M,BURINI A B,et al.SpliceAid 2:a database of human splicing factors expression data and RNA target motifs[J]. Hum Mutat,2012,33(1):81-85.
[7]	KAHLES A,LEHMANN K V,TOUSSAINT N C,et al. Comprehensive analysis of alternative splicing across tumors from 8 705 patients[J]. Cancer Cell,2018,34(2):211-224.e6.
[8]	WU H Y,PENG Z G,HE R Q,et al.Prognostic index of aberrant mRNA splicing profiling acts as a predictive indicator for hepatocellula r carcinoma based on TCGA SpliceSeq data[J]. Int J Oncol,2019,55(2):425-438.
[9]	YOSHIDA T,KIM J H,CARVER K,et al.CLK2 is an oncogenic kinase and splicing regulator in breast cancer[J]. Cancer Res,2015,75(7):1516-1526.
[10]	GE Y,SCHUSTER M B,PUNDHIR S,et al.The splicing factor RBM25 controls MYC activity in acute myeloid leukemia[J]. Nat Commun,2019,10(1):172.

项目	P值	HR
年龄	0.116	1.663（0.896~2.727）
性别	0.546	0.851（0.505~1.435）
息肉史	0.400	1.384（0.649~2.954）
肿瘤浸润	<0.001	3.916（2.272~6.752）
淋巴结转移	<0.001	2.083（1.536~2.825）
远处转移	<0.001	5.784（3.272~10.225）
临床分期	<0.001	2.397（1.767~3.250）
预后风险模型	<0.001	3.071（1.673~5.639）

项目	P值	HR
年龄	0.116	1.663（0.896~2.727）
性别	0.546	0.851（0.505~1.435）
息肉史	0.400	1.384（0.649~2.954）
肿瘤浸润	<0.001	3.916（2.272~6.752）
淋巴结转移	<0.001	2.083（1.536~2.825）
远处转移	<0.001	5.784（3.272~10.225）
临床分期	<0.001	2.397（1.767~3.250）
预后风险模型	<0.001	3.071（1.673~5.639）

项目	P值	HR
年龄	0.027	2.330（1.100~4.934）
性别	0.592	1.204（0.610~2.376）
息肉史	0.59	1.242（0.564~2.736）
肿瘤浸润	0.021	2.610（1.153~5.907）
淋巴结转移	0.567	1.217（0.621~2.384）
远处转移	0.93	1.072（0.225~5.108）
临床分期	0.242	2.027（0.620~6.627）
预后风险模型	<0.001	2.564（1.191~5.519）

项目	P值	HR
年龄	0.027	2.330（1.100~4.934）
性别	0.592	1.204（0.610~2.376）
息肉史	0.59	1.242（0.564~2.736）
肿瘤浸润	0.021	2.610（1.153~5.907）
淋巴结转移	0.567	1.217（0.621~2.384）
远处转移	0.93	1.072（0.225~5.108）
临床分期	0.242	2.027（0.620~6.627）
预后风险模型	<0.001	2.564（1.191~5.519）

基因名称	P值	HR	基因名称	P值	HR
NRIP2	0.003	1.194（1.061~1.344）	C9orf78	0.024	1.076（1.009~1.146）
WBP11	0.038	1.046（1.003~1.091）	NUMA1	0.032	1.052（1.004~1.102）
KHSRP	0.011	1.015（1.004~1.027）	MFAP1	0.026	1.063（1.007~1.122）
PDCD7	0.036	1.086（1.005~1.172）	DHX38	0.024	1.067（1.009~1.128）
CELF4	<0.001	5.525（2.526~12.084）	ZC3H18	0.037	1.100（1.006~1.204）
RBM3	0.025	0.992（0.985~0.999）	RBM47	0.015	0.954（0.919~0.991）
CCDC130	0.015	1.077（1.014~1.144）	DDX19A	0.032	1.192（1.016~1.399）
NOVA2	0.004	2.785（1.392~5.570）	SART1	0.017	1.040（1.007~1.074）
HSPB1	0.038	1.001（1.000~1.002）	CLK2	0.019	1.061（1.010~1.114）
GRSF1	0.040	0.953（0.911~0.998）	ISY1	0.001	1.228（1.083~1.393）
RBM17	<0.001	1.095（1.038~1.155）	CWC25	0.022	1.102（1.014~1.197）