A network-based predictive gene expression signature for recurrence risks in stage Ⅱ colorectal cancer

作者：YANG Wen-jing; WANG Hai-bo; WANG Wen-da; ZHANG Gan-lin; YANG Guo-wang

来源：中国药理学与毒理学杂志, 2019, 33(10): 890.

OBJECTIVE To explore novel gene signature to predict recurrence risksand reveal underlying mechanisms in stage Ⅱ colorectal cancer( Ⅱ CRC).METHODS In this study, we performed bioinformatical analyses based on high throughput RNA sequencing of CRC from The Cancer Genome Atlas(TCGA) database to gain a panoramic view of expression patterns between recurrence and non-recurrence patients in stageⅡ CRC.Furthermore, Differentially Expressed Genes(DEGs)were used to establish a model for predict the recurrence risk by random forest sequencing, and the diagnostic effectiveness was showed by receiver operating characteristic(ROC). Then used another gene expression profiling date which were extracted from GEO(GSE12032) for further validate the robust diagnostic effectiveness of therecurrent model. In addition, gene ontology(GO) functional annotation, Kyoto Encyclopedia of Genes and Genomes(KEGG) pathway enrichment analysis, Protein Protein Interaction(PPI) network analysis and hub genesselection were adopted to jointly analyze the underlying mechanism of recurrence. And the clinical values of the recurrent model and the hub genes were further explored at the same time. RESULTS First, 124 patients gene-expression profiles dataset of stageⅡ CRC from the TCGA database were obtained to screen DEGs. 202 DEGs including 128 up-regulated and 74 down-regulated were identified in recurrence group(n=24) compared to nonrecurrence group(n=100). Furthermore, the top five(ZNF561, WFS1, SLC2 A1, MFI2, PTGR1) DEGs were identified by random forest variable hunting, and four(ZNF561, WFS1, SLC2 A1, PTGR1) of them were selected to create a four-gene recurrent model(GRM) with the area under the curve(AUC) 0.882 by ROC, and the robust diagnostic effectiveness of GRM were further validated by another gene expression profiling date from GEO(GSE12032) with the AUC of 0.943. The diagnostic effectivenessof GRM about recurrence was confirmed associated with poor disease-free survival(DFS) to all stage CRC. In addition, GO functional annotation and KEGG pathway enrichment analysissuggested 18 functions and6 pathways were enrichment. Four genes as ABCG2,CACNA1 F, CYP19 A1, TF were identified as hub genes by PPI network, which further validated they were correlated with poor pathologic stage and overall survival(OS)in all stage CRC. CONCLUSION The GRM can effectively classify stageⅡCRC into groups at high and low risks of recurrence, thereby making up for the prognostic value of the traditional clinicopathological risk factors defined by the NCCN guidelines. And the hub genes may be the useful therapeutic targets for recurrence. Thus, the GRM and hub genes potentially offer clinical values in directing individualized and precision therapeutic regimen for stageⅡ CRC.

• 单位
  山西省肿瘤医院