Identification Of Biomarkers And Construction Of Clinical Prediction Models In Osteosarcoma

Background:Osteosarcoma(OS)is a common primary malignant bone tumor that is more common in children and adolescents.For a long time,there has been no significant change in the treatment plan for OS.At present,the 5-year overall survival rate of patients with OS is about 65%and the long-term prognosis of patients is still not ideal.With the rapid development of sequencing technology,RNA-seq has been increasingly used in the diagnosis and treatmentof clinical diseases.In order to deeply study the pathogenesis of OS and screen biological markers,this study uses the OS gene expression profile data and clinical information in the GEO and TARGET databases to deeply explore the risk factors related to core genes in OS and OS metastasis.Purposes:Bioinformatics method was used to screen biomarkers of OS and construct clinical prediction models.Methods:Download the OS sequencing data sets GSE87624 and GSE126209 from the GEO database,and perform differential expression analysis on the twodata sets in the R software and obtain the intersection of the differential genes.Based on the clusterProfile software package in R,the GO and KEGG enrichment analysis was performed on the intersecting differential genes and the PPInetwork was constructed on the STRING website.Based on the plug-in in Cytoscape,the core genes in the PPI network are selected,and the expression levels of the core genes are further analyzed in the three chip data sets of GSE21257,GSE39055,and GSE33382.In addition,gene expression data and clinicalinformation of OS patients were obtained from the TARGET database.OS samples were divided into metastatic and non-metastatic groups based on clinical information and differentially expressed genes were screened in both groups.Subsequently,the expression data of all differential genes and their corresponding clinical data were integrated,and Cox regression analysis was used to perform single factor survival analysis on the differential genes.Differential genes with P<0.05 in the analysis results were used to further construct a Cox multivariate regression model and screenfor OS prognostic related genes through stepwise regression.Finally,the entire clinical data is divided into at raining set and a validation set.The training set data is used to build a clinical prediction model.The reliability of the model is analyzed based on the C index,calibration curve and validation data.Results:1.Screened 3512 and 4683 differential genes from GSE87624 and GSE126209,respectively.There are 1299 overlapping genes in GSE87624 and GSE126209.2.The results of GO enrichment analysis showed that the differentialgene intersections were significantly enriched in biological processes such as extracellular matrix tissues and extracellular structural tissues.The results of KEGG metabolic pathway enrichment analysis showed that the metabolic pathways that were significantly enriched by differential gene intersection were cell cycle,DNA replication,small cell lung cancer,PI3K-Akt signaling pathway,and so on.3.(1)Inthe PPI network constructed by the intersection of differential genes,six core genes were selected,including MELK,EXO1,CDC45,CDK1,CDC6,and KIF2C.(2)Analysis of core genes in GSE21257 showed that the expression of EXO1 inthe meta static group was significantly higher than that in the non-metastatic group within 5 years of OS patients,and the expression of MELK,CDC45,CDK1,CDC6,and KIF2C was not significantly different between the two groups.(3)Analysis of core genes in GSE39055 showed that the expression of CDK1 in the relapse group of osteosarcoma patients was significantly lower than that in the non-relapse group,and the expression levels of MELK,CDC45,EXO1,CDC6,and KIF2C were not significantly different between the two groups.(4)Analysis of core genes in GSE33382 showed that compared with normal osteoblasts,the expression of CDC45,EXO1,CDC6,and KIF2C was significantly up-regulated in high-grade osteosarcoma tissues,and the expression of CDK1 and MELK in normal osteoblasts and high There were no significant differences between grade osteosarcoma tissues.4.(1)A total of92 differential genes were screened from the TARGET database.Based on univariate Cox regression analysis,13 genes with P<0.05 were selected from 96 differential genes and 5 genes were finally selected through stepwise regression method,including MAGEA11,TCF24,MYC,HERC5 and GZMB.MAGEA11,TCF24,HERC5,and GZMB are protective factors for prognosis,and MYC is a risk factor for prognosis.(2)Based on all clinical data,the four variables of age,gender,primary tumor site and whether metastasisoccurred were included in the Cox modeling analysis.The model is evaluated through the C index and calibration curve,and the results show that our model has good discrimination and prediction a ccuracy.Conclusions:6 core genes of osteosarcoma,MELK,EXO1,CDC45,CDK1,CDC6 and KIF2C were screened from the gene expression data set of OS in the GEO database.Combining the gene expression data and clinical information of OS in the TARGET database,5 genes related to the prognosis of OS including MAGEA11,TCF24,MYC,HERC5 and GZMB were screened,and a clinical prediction model for OS was constructed based on the clinical data.Our work mayprovide references for future OS mechanism research and clinical diagnosis and treatment.