Establishment Of Hub Genes Integration Analysis Model For Multiple Myeloma

Objectives To explore the expression of hub genes associated with multiple myeloma（MM）,establish and evaluate a clinical diagnostic scoring system for MM.Method 1.Data sets were obtained in Gene Expression Omnibus（GEO）public database,plasma cells of patients with MM and healthy donors were analyzed,Kyoto Encyclopedia of Genes and Genomes（KEGG）pathway,Gene Ontology（GO）annotation,Disease Ontology（DO）annotation of differentially expressed genes（DEGs）was conducted and a protein-protein interaction network（PPI）was generated by R software.2.DEGs in MM.1S cells and control HS-5 cells was verified by Microdroplet Digital PCR（dd PCR）.3.The expression level of DEGs in plasma and bone marrow was detected by ELISA.4.The hemoglobin,lactate dehydrogenase（LDH）andβ2 microglobulin（β2-MG）,serum creatinine and other clinical indexes were collected.5.SPSS 21.0 was used for statistical analysis,the measurement data were represented by（±s）,t-test was used for comparison between two groups conforming to normal distribution,variance analysis for three groups.The correlation was determined by Spearman test.The cut-off value was analyzed by ROC curve,and the independent risk factors of MM were analyzed by Logistic regression.According to theirβcoeficients,we set a clinical diagnostic scoring system for MM,ROC curve was used to evaluated the predictive value.Results 1.Two gene expression array data sets（GSE5900 and GSE6477）were downloaded.The threshold value used to define DEGs was|log2fold-change|>1.2 and adjusted P<0.05.Of 908 DEGs,416 were up-regulated and 492 down-regulated.KEGG,GO and DO annotation of DEGs were conducted and PPI network was generated,eight hub genes（CXCL2,CXCL8,CXCL12,CCL13,CCL27,CX3CL1,ELANE and LCN2）screened out by R software.In both data sets,CXCL2,CXCL8,CXCL12,CCL13 and LCN2 were lower in myeloma plasma cells than control,which were verified in MM.1S cells and control HS-5 cells.2.Expression levels of CXCL8 and CXCL12 in MM.1S were significantly lower than those in HS-5 cells,CXCL2 and LCN2were only expressed in HS-5 cells,ELANE,CCL27,CCL13 and CX3CL1 were not detected in both of cells.3.Expressions levels of eight cytokines in bone and peripheral blood plasma from patients with MM were higher than healthy controls.4.Correlation analysis with clinical indicators showed that CXCL2,CXCL8 and CXCL12 were negatively correlated with hemoglobin.CXCL2,CXCL8,CXCL12,CX3CL1,CCL13,CCL27 and LCN2 were positively correlated withβ2-MG.CXCL8,CCL13,CCL27 were positively correlated with Ca²⁺.LCN2 was correlated with serum creatinine.5.In peripheral blood plasma,the optimal cut-off values of CXCL2,CXCL8,CXCL12,CCL13,CCL27,CX3CL1,ELANE,LCN2 were 313.43ng/L,884.21ng/L,2257.12pg/m L,236.41pg/m L,189.35ng/m L,132.68ng/L,50.34（?）g/L and7.77（?）g/L,respectively.By multivariate analysis,CXCL2,CCL27 andβ2-MG were associated with the incidence of MM,combined with LDH and Ca²⁺,the five variables were included to the diagnostic scoring system.The optimal cut-off value is 1.5 points,the area under the curve（AUC）of the scoring system was 0.981[95%CI（0.957,1.000）,standard error was 0.012,P<0.001],the sensitivity was 0.906 and specificity was1.000.Conclusions:The hub genes of MM are obtained by bioinformatics and verified by experiments,with high reliability.The combination of DEGs with clinical indicators to established MM diagnostic scoring system has good clinical application value for MM Pathogenesis.