Identification and validation of a natural killer cell-associated immunomodulatory gene signature for the diagnosis of osteoporosis.

INTRODUCTION

Osteoporosis is a prevalent metabolic bone disease characterized by heterogeneous molecular mechanisms. Emerging evidence implicates that natural killer cells and immunomodulator-related genes (NIRGs) were involved in its pathogenesis; however, systematic exploration of NIRGs remains limited.

OBJECTIVE

To explore the role of natural killer (NK) cells and immunomodulator-related differentially expressed genes (NIRDEGs) in osteoporosis.

METHOD: Two osteoporosis datasets (GSE56815 and GSE7429) were integrated after batch effect correction and normalization. Natural killer cell-related genes (NRGs) and immunomodulator-related genes (IRGs) were curated from GeneCards. And NIRDEGs were identified by intersecting co-differentially expressed genes (Co-DEGs) across datasets. Functional enrichment (GO, KEGG, GSEA, GSVA) and immune infiltration analyses were performed. Then, diagnostic models were developed using random forest, support vector machine (SVM), and least absolute shrinkage and selection operator (LASSO) regression, validated by receiver operating characteristic curves (ROC), and nomograms. Protein-Protein Interaction (PPI) networks and regulatory networks were generated using bioinformatic tools.

RESULTS

Twelve NIRDEGs were identified, and they were enriched in regulation of inflammatory response, myeloid cell homeostasis, negative regulation of DNA-binding transcription factor activity, homeostasis of number of cells, and I-κB kinase/NF-κB signaling. A diagnostic model comprising seven genes (IDO1, PTMA, RAB11FIP1, MYH14, HAMP, HMOX1, and ESR1) exhibited moderate accuracy (0.7 < AUC < 0.9). The protein-level expression of the six candidate genes was further validated in osteoporosis medol rats. Specifically, the IDO1 and PTMA was obviously elevated in the osteoporosis rat model, while the RAB11FIP1, MYH14, HAMP, and HMOX1 was markedly decreased. Immune infiltration analysis revealed distinct patterns of 20 immune cell types between high and low BMD groups, especially monocytes, macrophages, and eosinophils. The PPI network indicated that five key genes were interconnected (IDO1, PTMA, HAMP, HMOX1, and ESR1) and that there were 20 functionally related proteins. Regulatory networks highlighted interactions with 26 transcription factors (TFs), 28 miRNAs, 26 RBPs and 32 therapeutic compounds.

CONCLUSIONS

This study establishes NIRDEGs as critical players in osteoporosis pathogenesis and provides a clinically translatable seven-gene diagnostic model for early osteoporosis detection. The integration of multi-omic analyses uncovered key pathways, immune dynamics, and regulatory networks. These findings provide novel insights into immune-mediated mechanisms and therapeutic targets for osteoporosis management.