Abstract
BACKGROUND: Osteoporosis is a serious global public health concern. The present study identified specific osteoporosis biomarkers which were used to generate a predictive prognostic model for patients with osteoporosis. METHODS: Datasets from the Gene Expression Omnibus (GEO) database were analyzed. Differentially expressed genes (DEGs) between osteoporosis and normal controls were identified by integrated microarray analysis of the GEO database. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses of differentially expressed gene function were performed. The best diagnostic gene biomarkers for Osteoporosis were identified using Weighted gene co-expression network analysis (WGCNA) and protein-protein interaction (PPIs) networks. Classification models including support vector machines (SVM) was developed to test the diagnostic value of the identified gene biomarkers for osteoporosis. Integrated microarray analysis of the GEO database (GSE62402, GSE7158 and GSE13850) was used for validation. RESULTS: A total of 1,589 DEGs related to osteoporosis were identified in the GSE35959 dataset. These DEGs were enriched in various pathways including the negative regulation of the calcium ion transmembrane transport pathway. WGCNA identified 16 models, with the blue module showing a strong positive association with osteoporosis, and the turquoise module showing a considerable negative association with osteoporosis. Six hub genes were identified and used as features to build a predictive prognostic model for osteoporosis using the GSE35959 dataset. The model was verified using the GSE62402, GSE7158, and GSE13850 datasets. CONCLUSIONS: These findings provide crucial insights into the mechanisms of the occurrence and progression of osteoporosis. Furthermore, the identification of novel potential biomarkers may contribute to the early diagnosis, prevention, and treatment of osteoporosis.