Integrated Bulk and Single-Cell Transcriptomic Analysis Followed by Clinical Validation Reveal Programmed Cell
Jihua Liu1, Zehao Hu2, Zixuan Xu1
1School of Clinical Medicine, Jiangxi University of Chinese Medicine, Nanchang 330004, China.
Abstract:
Osteoarthritis (OA) and major depressive disorder (MDD) share inflammatory and oxidative stress pathways, but the role of programmed cell death (PCD) in their comorbidity remains unclear. This study used independent OA synovial and MDD peripheral blood transcriptomic datasets-not a unified comorbid discovery cohort-to identify candidate PCD-related molecular signatures commonly dysregulated in both conditions. Transcriptomic data from OA synovium and MDD brain tissues were obtained from GEO (six training [three OA synovial and three MDD peripheral-blood], seven validation, and two single-cell RNA-seq datasets). Differentially expressed genes (DEGs) were identified, and PCD-related DEGs were screened. Machine learning (LASSO, SVM-RFE, Random Forest) was used to identify hub PCD-DEGs from the OA training set. WGCNA identified MDD-associated modules for comorbidity-gene selection. Functional enrichment, immune infiltration, scRNA-seq localization, and clinical validation (qRT-PCR/WB) were performed. From the OA cohort, four hub PCD-DEGs (CDKN1A, CX3CR1, INHBB, RHOB) showed moderate diagnostic value for OA (nomogram AUC = 0.82). Eight candidate genes (VAMP8, PDK4, P2RX4, ITM2C, IL10RA, HSP90AA1, CTSO, CRIP1) were commonly dysregulated across both OA and MDD datasets. Immune infiltration revealed upregulated B memory cells, plasma cells, Tregs, and neutrophils in OA, and neutrophils in MDD. scRNA-seq localized CDKN1A/RHOB to OA synovial cells and HSP90AA1/ITM2C to MDD neurons. Enrichment analyses highlighted TNF signaling, apoptosis, and stress responses in both diseases. An independent OA-MDD clinical cohort confirmed differential expression of CDKN1A, RHOB, ITM2C, and HSP90AA1. This study identifies four PCD-related hub genes associated with OA and eight candidate comorbidity genes showing common dysregulation across OA and MDD datasets and in an independent clinical cohort. These findings generate hypotheses about shared inflammatory pathways linking OA and MDD. As these associations derive from independent disease-specific cohorts rather than a true comorbid discovery cohort, they represent candidate signatures requiring functional validation rather than established mechanisms.
