Integrative genomic analysis and diagnostic modeling of osteoporosis: unraveling the interplay of autophagy, osteogenesis, adipogenesis, and immune infiltration

Front Med (Lausanne). 2025 Apr 17;12:1544390. doi: 10.3389/fmed.2025.1544390. eCollection 2025.

ABSTRACT

BACKGROUND: Osteoporosis (OP), marked by reduced bone density and structural decay, poses a heightened risk of fractures. Our study formulates a predictive diagnostic model for OP by analyzing differential gene expression, thereby improving early diagnosis and therapeutic approaches.

METHODS: Using GSE62402, GSE56815, and GSE35958 datasets from the Gene Expression Omnibus (GEO) database, we identified differentially expressed genes (DEGs) via R packages, and evaluated the underlying molecular mechanisms by network analysis. Immune checkpoint and drug sensitivity were analyzed to construct and validate diagnostic models. The single-sample gene-set enrichment analysis (ssGSEA) was used to assess immune cell infiltration; the CIBERSORT algorithm was used to evaluate immune cells within the different subtypes of OP.

RESULTS: The study identified 1,297 DEGs, with 14 DEGs related to autophagy, osteogenesis, and adipogenesis (AP&OG&AGRDEGs) showing significant expression differences between OP and control groups, including seven upregulated and seven downregulated genes (p-value < 0.05). The analysis results from gene ontology (GO), gene set enrichment analysis (GSEA), and the Kyoto encyclopedia of genes and genomes (KEGG) indicated that oxidative stress and inflammation-related signaling pathways are closely connected to OP. Immune checkpoint analysis identified differential expression of eight genes between OP patients and controls (p-value < 0.05). The ssGSEA findings showed significant variations in immune cell infiltration levels, particularly of natural killer cells, Th2 cells, mast cells, and plasmacytoid dendritic cells (p-value < 0.05). The diagnostic model, developed utilizing logistic regression, support vector machine (SVM), and the least absolute shrinkage and selection operator (LASSO), pinpointed nine pivotal genes-AKT1, NFKB1, TNF, CTNNB1, LMNA, BHLHE40, BMP4, WNT1, and COPS3-and confirmed their diagnostic efficacy through validation. In further subgroup analysis, eight types of immune cells were found to be differentially expressed across various risk groups. Subtype analysis based on ConsensusClusterPlus revealed differential expression of six key genes in distinct subtypes of OP.

CONCLUSION: This comprehensive study established a network of OP-associated genes, and provides insights into the molecular mechanisms involving immune responses in OP. It identified key diagnostic genes and analyzed immune cell infiltration to better understand OP pathogenesis. The study underscores the importance of personalized treatment and the potential role of immune modulation in managing OP.

PMID:40313558 | PMC:PMC12043663 | DOI:10.3389/fmed.2025.1544390