Exploring the association between circadian rhythms and osteoporosis: new diagnostic and therapeutic targets identified via machine learning

TL;DR

This study uses machine learning to find new biomarkers linking circadian rhythms and osteoporosis, offering potential for early diagnosis and treatment.

Contribution

Novel circadian rhythm-related biomarkers for osteoporosis identified using machine learning and bioinformatics.

Findings

140 circadian rhythm-related differentially expressed genes were identified in osteoporosis.

Five key genes (ECE1, FLT3, APPL1, RAB5C, FCGR2A) showed high diagnostic performance with AUC of 0.904 and 0.887.

Immune cell infiltration analysis revealed altered immune profiles in osteoporosis patients.

Abstract

Osteoporosis (OP) is a systemic metabolic bone disease that may increase the risk of disability or death. Increasing evidence suggests that circadian rhythms play an important role in OP, yet the specific mechanisms remain unclear. Therefore, this study aims to utilize bioinformatics and machine learning algorithms to identify novel diagnostic biomarkers related to the circadian rhythm in OP, providing new targets for early diagnosis and treatment of OP.

The OP dataset GSE56815 was downloaded from the GEO database, differential expression analysis was performed to identify differentially expressed genes (DEGs) between OP and control samples. DEGs were intersected with circadian rhythm-related genes (CRRGs) to obtain circadian rhythm-related differentially expressed genes (CRRDEGs), which were subjected to Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses. Four machine learning algorithms were applied to identify key genes for constructing a diagnostic model. The diagnostic performance of the model was validated by plotting receiver operating characteristic (ROC) curves using the GSE7158 dataset. Gene set enrichment analysis (GSEA) was performed on the key genes. Single-sample gene set enrichment analysis (ssGSEA) was used to analyze immune cell infiltration and explore the correlation between key genes and immune cells. Drug-gene interaction networks and competitive endogenous RNA (ceRNA) networks were constructed using the key genes.

A total of 140 CRRDEGs were identified. By comparing four machine learning algorithms, the top five genes from the SVM algorithm (ECE1, FLT3, APPL1, RAB5C and FCGR2A) were determined as key genes for OP. The diagnostic model based on these five key genes demonstrated high diagnostic performance, with AUC of 0.904 for the training set and 0.887 for the validation set. Immune cell infiltration analysis revealed that Type 2 T helper cells and CD56dim natural killer cells were significantly upregulated in the OP group, while activated dendritic cells were significantly downregulated. The drug-gene interaction network and ceRNA network constructed based on the key genes revealed potential therapeutic targets for OP.

This study identified ECE1, FLT3, APPL1, RAB5C and FCGR2A as circadian rhythm-related novel diagnostic biomarkers for OP, providing new insights for further understanding the early diagnosis and treatment of OP.

Figures