Identification And Verification Of A Novel Combined Ferroptosis-and Iron Metabolism-related 8-gene Signature Of Myocardial Infarction By Machine Learning

Objective:Myocardial infarction(MI)is a serious threat to human health and the main cause of death from cardiovascular diseases worldwide.MI is a polygenic disease,mainly determined by the joint action of genetic and environmental factors.Research shows that some MI patients lack traditional risk factors related to the disease,which may be related to gene susceptibility.Ferroptosis is a new type of programmed cell death mode that is iron ion-dependent and different from necrosis,apoptosis and autophagy.It plays an important role in the development of many diseases.Recent studies have shown that ferroptosis plays an increasingly important role in the development of cardiovascular diseases,especially in myocardial infarction.This study screened and evaluated the diagnostic titer level of ferroptosis related genes in myocardial infarction through machine learning model,which may help to understand the pathological mechanism of MI and provide potential theoretical basis for its early diagnosis and comprehensive treatment.Methods:In this study,the GSE59867 data set in the Gene expression omnibus(GEO)was used as the development set.We performed differential expression analysis and volcanic map analysis on the microarray data of MI samples at four time points after the occurrence of MI(admission,discharge,1 month after discharge,6 months after discharge)Differential expressed genes(DEGs)at discharge were analyzed for function and PPI network;DEGs at four time points were intersected with ferroptosis and iron metabolism-related genes to obtain four gene sets.The minimum absolute shrinkage and selection operator(LASSO)and random forest machine learning model are used to analyze the intersection gene set at the time of admission,and the genes screened by the two algorithms are intersected to obtain the target gene.The expression of the selected target genes at four time points after MI was analyzed and analyzed by Friend analysis,correlation analysis,receiver operating characteristic(ROC)and nomogram analysis.After that,samples from other GEO MI data sets GSE66360 and GSE123342 were used as validation sets for ROC analysis to further clarify the diagnostic value of the selected genes.Result:1.Differential expression analysis showed that ferroptosis and iron metabolism-related DEGs were the most at admission;The functional analysis of DEGs at admission showed that it was related to "Regulation of reactive oxygen species metabolic process";The analysis of human disease-related genes and mutations of DEGs at the time of admission showed that they were related to "Acute myocardial infarction".2.22 genes were obtained from LASSO analysis,and 15 genes were obtained from random forest model analysis.Eight genes,AMD,PPARG,SOCS3,TSPO,ASGR2,FAM20 C,ST14 and TCN2,were obtained from the intersection of the two.3.According to the T test of non-paired samples,the expression of 8genes in MI group increased significantly at the time of admission and discharge,while the expression of TSPO,ADM and ASGR2 was higher than that in the control group at 1 month after discharge,the difference was statistically significant;At 6 months after discharge,the expression of TSPO and ASGR2 was significantly higher than that of the control group.Friend analysis results show that ADM has the strongest correlation with other DEGs.Pearson correlation analysis showed that these 8 genes were strongly correlated with most ferroptosis related genes.4.The ROC analysis results show that the AUC value of ST14 is 0.881(CI: 0.83-0.932),the AUC value of TCN2 is 0.883(CI: 0.24-0.942),the AUC value of FAM20 C is 0.865(CI: 0.8-0.929),the AUC value of SOCS3 is 0.914(CI: 0867-0.961,the AUC value of PPARG is 0.911(CI: 0.64-0.958),the AUC value of TSPO is 0.902(CI: 0.52-0.952),the AUC value of ADM is 0.92(CI: 0874-0.967),and the AUC value of ASGR2 is 0.928(CI: 0.888-0.968).The results of the nomogram show that PPARG and ASGR2 have the greatest impact on the outcome events.5.The validation set data shows that the total AUC value of GSE66360 is 0.875(CI: 0.804-0.945),and the total AUC value of GSE123342 is 0.859(CI: 0.773-0.944).Conclusion:Eight genes related to ferroptosis and iron metabolism screened by machine learning model have high diagnostic titer for MI.