Diagnostic lncRNA Biomarkers and Immune-Related ceRNA Networks for Osteonecrosis of the Femoral Head in Metabolic Syndrome Identified by Plasma RNA Sequencing and Machine Learning

Haoyan SunMeng XuDianlong MiQingyu LiHaipeng SunYang Song^*

• The Second Hospital of Jilin University, Changchun, China

Osteonecrosis of the femoral head (ONFH) is a disabling orthopedic condition that remains challenging to diagnose at an early stage. Recent evidence suggests that immune dysregulation plays a central role in the development of both ONFH and metabolic syndrome (MetS), a cluster of metabolic abnormalities associated with increased ONFH risk. However, reliable noninvasive diagnostic biomarkers for ONFH, particularly in high-risk MetS populations, are still lacking. This study aimed to identify key diagnostic long non-coding RNAs (lncRNAs) in ONFH patients with MetS and to construct an immune-related competitive endogenous RNA (ceRNA) network. Plasma lncRNA and mRNA expression profiles from 9 ONFH patients and 6 healthy controls were analyzed to identify differentially expressed lncRNAs (DElncRNAs) and mRNAs (DEmRNAs), followed by ceRNA network construction. The MetS dataset from the Gene Expression Omnibus (GEO) was integrated, and weighted gene co-expression network analysis (WGCNA), functional enrichment, protein-protein interaction (PPI) network analysis, MCODE, CytoHubba-MCC, and random forest (RF) algorithms were employed to identify hub mRNAs and their associated lncRNAs. A nomogram model was developed, and diagnostic potential was evaluated using receiver operating characteristic (ROC) analysis and validation in an independent cohort (45 ONFH and 45 control samples). A total of 424 DElncRNAs and 1,431 DEmRNAs were identified, and a ceRNA network involving 7 lncRNAs, 24 miRNAs, and 683 mRNAs was constructed. Integration with the MetS dataset yielded 506 intersecting mRNAs, from which 11 hub mRNAs and 6 related lncRNAs were screened. Five key lncRNAs were selected by RF analysis to construct a diagnostic model with strong predictive performance (AUC > 0.7 in both RNA-seq and qRT-PCR validation). The immune-related ceRNA network also demonstrated significant associations with immune cell infiltration patterns. In conclusion, five candidate lncRNAs (MRPS30-DT, LINC01106, MIR100HG, WDR11-AS1, and PELATON) were identified as promising noninvasive diagnostic biomarkers for ONFH in MetS populations. These findings offer novel insights into immune-related regulatory mechanisms and may support early diagnosis using peripheral blood.