AUTHOR=Wu Xiaofeng , Pan Zhou , Liu Wei , Zha Shiqian , Song Yan , Zhang Qingfeng , Hu Ke TITLE=The Discovery, Validation, and Function of Hypoxia-Related Gene Biomarkers for Obstructive Sleep Apnea JOURNAL=Frontiers in Medicine VOLUME=9 YEAR=2022 URL=https://www.frontiersin.org/journals/medicine/articles/10.3389/fmed.2022.813459 DOI=10.3389/fmed.2022.813459 ISSN=2296-858X ABSTRACT=

While there is emerging evidence that hypoxia critically contributes to the pathobiology of obstructive sleep apnea (OSA), the diagnostic value of measuring hypoxia or its surrogates in OSA remains unclear. Here we investigated the diagnostic value of hypoxia-related genes and explored their potential molecular mechanisms of action in OSA. Expression data from OSA and control subjects were downloaded from the Gene Expression Omnibus database. Differentially-expressed genes (DEGs) between OSA and control subjects were identified using the limma R package and their biological functions investigated with the clusterProfiler R package. Hypoxia-related DEGs in OSA were obtained by overlapping DEGs with hypoxia-related genes. The diagnostic value of hypoxia-related DEGs in OSA was evaluated by receiver operating curve (ROC) analysis. Random forest (RF) and lasso machine learning algorithms were used to construct diagnostic models to distinguish OSA from control. Geneset enrichment analysis (GSEA) was performed to explore pathways related to key hypoxia-related genes in OSA. Sixty-three genes associated with hypoxia, transcriptional regulation, and inflammation were identified as differentially expressed between OSA and control samples. By intersecting these with known hypoxia-related genes, 17 hypoxia-related DEGs related to OSA were identified. Protein-protein interaction network analysis showed that 16 hypoxia-related genes interacted, and their diagnostic value was further explored. The 16 hypoxia-related genes accurately predicted OSA with AUCs >0.7. A lasso model constructed using AREG, ATF3, ZFP36, and DUSP1 had a better performance and accuracy in classifying OSA and control samples compared with an RF model as assessed by multiple metrics. Moreover, GSEA revealed that AREG, ATF3, ZFP36, and DUSP1 may regulate OSA via inflammation and contribute to OSA-related cancer risk. Here we constructed a reliable diagnostic model for OSA based on hypoxia-related genes. Furthermore, these transcriptional changes may contribute to the etiology, pathogenesis, and sequelae of OSA.