Construction And Analysis Of MiRNA Prognostic Risk Model In Breast Cancer Patients

ObjectiveTo explore and establish a multi-gene prognostic model for predicting the prognosis of breast cancer patients,The Cancer Genome Atlas(TCGA)and Gene Expression Omnibus(GEO)databases were used.Besides,we analyzed the relationship between the model and clinicopathological characteristics of breast cancer,which guides the clinical individualized treatment.MethodsThe miRNA data and clinical information of breast cancer cohort were downloaded from TCGA database.The mRNA sequencing data(GSE139038)of breast cancer and adjacent tissues were downloaded from GEO database.Differentially expressed genes of miRNAs was identified by edgeR package in RStudio software.Dividing patients into different groups by using caret package in RStudio software.The training cohort was used to construct a prediction model and the validation dataset was used to evaluate performance of the prediction model.Univariate Cox and Lasso regression analyses were performed to identify miRNAs related to patient survival in the training group.Furthermore,a risk scoring equation was established by multivariate Cox regression analysis to construct a prognostic associated miRNA prediction model.The sensitivity and specificity of the model were evaluated by receiver operating characteristic curve(ROC).The target genes of seven miRNAs were predicted and intersected with differential expressed mRNA for enrichment analysis by Kyoto Encyclopedia of Genes and Genomes(KEGG)signaling pathway and gene ontology(GO).A protein-protein interaction network(PPI)was constructed to screen hub genes.Finally,survival analysis of hub genes was performed.ResultsThe bioinformatics analysis showed that a total of 254 differentially expressed miRNAs were screened,of which 157 miRNAs were up-regulated and 97 miRNAs were down-regulated in breast cancer tissues.Finally,multivariate Cox regression analysis obtained the predictive risk model of 7 miRNAs related survival(hsa-miR2115,hsa-miR-340,hsa-miR-3610,hsa-miR-449b,hsa-miR-556,hsa-miR-551b,hsamiR-605),with the equation as follows:Risk score=(0.565×expressionhsa-miR-340)+(0.414×expression hsa-miR-605)-(0.051×expression hsa-miR-2115)-(0.164× expressionhsamiR-3610)-(0.201×expressionhsa-miR-449b)-(0.247×expression hsa-miR-551b)-(0.465×expression hsa-miR-556).Area under curve(AUC)of the ROC curve for predicting 3-year survival in the training dataset,test dataset,and whole cohort were 0.793,0.688,and 0.740.The AUC of the ROC curve for predicting 5-year survival in the training dataset,test dataset,and whole cohort were 0.725,0.678,and 0.701.Univariate and multivariate.Cox regression analyses showed that the 7-miRNAs prognostic model could be used as an independent prognostic factor.The top 10 hub genes in the PPI network screened by target genes include IGF1、IRS1、LPL、ESR1、CYCS、SLC2A1、KIT、SKP2、FABP4、FBXO9.All of them were related to the survival prognosis of breast cancer.ConclusionIn this study,seven survival related miRNAs were identified in breast cancer patients,and we established a prognostic prediction model.The result may contribute to the individualized treatment of breast cancer patients and provide new insights for the development of novel reliable and accurate cancer treatment methods.