Identification of Cystatin 3 as a Potential Diagnostic Biomarker for Osteoporosis Using Machine Learning.

Background: Osteoporosis (OP) is one of the most common systemic bone metabolic diseases, but its specific pathogenesis remains unclear. Cystatin 3 (CST3) is a cysteine protease inhibitor involved in various physiological and pathological processes, yet its role in osteoporosis has not been clarified. This study aims to explore the diagnostic value and potential mechanism of CST3 in OP. Methods: Transcriptome data of OP patients and healthy individuals were obtained from the Gene Expression Omnibus (GEO) database. After normalization and batch effect correction, differentially expressed genes (DEGs) were screened. Gene Ontology (GO) enrichment analysis, enrichment term interaction analysis, and protein-protein interaction (PPI) analysis were performed on these DEGs. Characteristic genes were screened using Least Absolute Shrinkage and Selection Operator (LASSO) regression and Random Forest (RF) algorithms, and their diagnostic efficacy was evaluated by combining with ROC curves. A rat model of OP was constructed, and the expression of characteristic genes in bone marrow mesenchymal stem cells (BMSCs) was verified by quantitative real-time polymerase chain reaction (qRT-PCR). The functions of genes related to the characteristic genes and their potential association with immune infiltration were analyzed through co-expression analysis, PPI network, and CIBERSORT algorithm. Results: A total of 178 DEGs were screened, which were enriched in pathways such as extracellular matrix regulation and collagen metabolism. Machine learning algorithms identified CST3 and FLJ36848 as characteristic genes, with the area under the ROC curve (AUC) of both exceeding 0.9, showing excellent diagnostic efficacy. Moreover, the diagnostic efficacy of CST3 in the validation set was superior to that of FLJ36848. Animal experiments confirmed that the expression of CST3 was upregulated in BMSCs of OP rats, while the expressions of ALP and OCN were downregulated. The PPI network showed that CST3 interacted with 178 node genes. Immune infiltration analysis revealed that the infiltration proportions of M2-type macrophages, NK cells, etc. were significantly increased in the CST3 high-expression group, suggesting that CST3 may be involved in the progression of OP by regulating the immune microenvironment. Conclusion: This study found that CST3 is related to the pathogenesis of osteoporosis and may represent a promising biomarker associated with osteoporosis progression, which could be explored as a potential therapeutic target in future studies. Its potential mechanisms involve the association of CST3 with the regulation of extracellular matrix decomposition, collagen metabolism, calcium ion transmembrane transport, as well as immune cell infiltration and its function in osteoporosis. It should be clearly stated that this study still lacks direct functional evidence and verification with large-scale clinical samples. Therefore, these findings still need to be verified by more animal experiments and clinical trials, and the specific molecular mechanisms require further in-depth research.