Identification and validation of relevant diagnostic biomarkers for osteoporosis by Weighted Gene Co-expression Network Analysis and machine learning

Introduction: Osteoporosis (OP) is a systemic metabolic bone disease characterized by complex pathogenesis and high prevalence. Current diagnostic and therapeutic approaches have limited effectiveness, and new biomarkers are needed to improve the treatment and diagnosis of OP. Objective: The present study aimed to identify novel diagnostic biomarkers for OP through integrated bioinformatics analysis. Methods: We performed an integrative bioinformatics analysis combining Weighted Gene Co-expression Network Analysis and machine learning on two Gene Expression Omnibus datasets (GSE35958, GSE35956). Differentially expressed genes (DEGs) were identified using “limma” package of R software, followed by module construction and key gene screening via Least Absolute Shrinkage and Selection Operator (LASSO) regression. Functional enrichment, immune infiltration, and drug prediction analyses were conducted to explore biological mechanisms and therapeutic potential. Results: Differential expression analysis identified 1,020 DEGs, from which 10 co-expression modules were constructed. The blue module demonstrated the strongest correlation with OP (r = 0.99, p<0.0001). LASSO regression analysis prioritized seven candidate genes (LOC286177, nucleobindin 1 [NUCB1], peroxisomal biogenesis factor 19 [PEX19], metastasis associated 1 [MTA1], DRA aassociated protein 1 [DRAP1], protocadherin gamma A1 [PCDHGA1], and pre-mRNA processing factor 39 [PRPF39]), with subsequent validation confirming NUCB1, PEX19, MTA1, DRAP1, and PCDHGA1 as robust diagnostic biomarkers (Area under the curve > 0.85). Functional enrichment implicated these genes in endoplasmic reticulum stress, Wnt/β-catenin signaling, and immune regulatory pathways. Immune profiling further revealed significant perturbations in T-cell and macrophage populations in OP. The Coremine Medical database was leveraged to predict potential therapeutic agents, including both small-molecule and phytochemical candidates. Conclusion: The present study identified NUCB1, PEX19, MTA1, DRAP1, and PCDHGA1 as promising OP diagnostic markers and explored their roles in bone metabolism. The findings offer insights for early diagnosis and targeted therapy but require further clinical validation.