The diagnostic value of inflammation-related genes in bronchopulmonary dysplasia

Abstract

This study aims to explore the diagnostic value of inflammation-related genes in peripheral blood mononuclear cells in bronchopulmonary dysplasia (BPD). By using bioinformatics analysis, three datasets including GSE32472, GSE125873, and GSE220135, which contain whole-genome expression profile data of 251 neonates, were included. The GSE32472 dataset was used as a training dataset to detect differentially expressed genes between non-BPD and BPD neonates in peripheral blood mononuclear cells. The gene enrichment analysis (GSEA) was used to detect the pathway enrichment of up-regulated genes in BPD newborns. The main regulatory factors analysis (MRA) algorithm was used to filter the main regulatory genes in the inflammation-related pathway (GO:0006954). After obtaining the main regulatory genes, the expression of the main regulatory genes in the GSE32472, GSE125873, and GSE220135 datasets was detected. Through the logistic regression model, risk scoring was conducted for neonates, and the risk scores of non-BPD and BPD neonates were compared. Lastly, the classification performance of the model was evaluated using the area under the curve (AUC). The results showed that compared with non-BPD neonates, there were 486 up-regulated genes and 433 down-regulated genes in the peripheral blood mononuclear cells of BPD neonates. The inflammation-related pathway was highly enriched in the up-regulated genes. Ultimately, phospholipase C beta 1 (PLCB1), nidogen 1 (NID1), serum response factor binding protein 1 (SRFBP1), centrosomal protein 72 (CEP72), excision repair cross complementation group 6 like (ERCC6L), and peptidylprolyl isomerase like 1 (PPIL1) were identified as the main regulatory genes. The prediction model's calculation formula for risk score was PLCB1×0.26+NID1×0.97+SRFBP1×1.58+CEP72×(-0.36)+ERCC6L×2.14+PPIL1×0.67. The AUCs in the GSE32472 test dataset, GSE125873 dataset, and GSE220135 dataset were 0.88, 0.86, and 0.89, respectively. This prediction model could distinguish between non-BPD and BPD neonates. In conclusion, the prediction model based on inflammation-related pathway genes has a certain diagnostic value for BPD.