Osteoarthritis (OA) is a highly prevalent and debilitating degenerative joint disorder characterized by cartilage degradation and synovial inflammation. Fibroblasts, particularly synovial fibroblasts, are increasingly recognized as pivotal drivers of this joint pathology. Therefore, this study aimed to systematically identify fibroblast-associated key genes in OA and to elucidate their potential molecular mechanisms and therapeutic target value.

To achieve this, a comprehensive integrative analytical framework was designed, combining Mendelian randomization (MR), single-cell RNA sequencing (scRNA-seq), high-dimensional weighted gene co-expression network analysis (hdWGCNA), bulk RNA-seq differential expression analysis, and in silico functional prediction. Specifically, MR analysis leveraged robust fibroblast-specific expression quantitative trait loci (eQTL) data from GTEx v10 alongside large-scale OA genome-wide association study (GWAS) summary statistics to rigorously assess genetic gene–disease causal relationships. Furthermore, scRNA-seq data were utilized for high-resolution cell-type annotation, pseudotime trajectory analysis to track cellular state transitions, and spatial mapping of hub genes. Fibroblast-specific hub genes wereultimately pinpointed by intersecting MR-supported causal genes, hdWGCNA disease-associated module genes, and bulk differentially expressed genes. Downstream functional enrichment analyses, transcription factor (TF)–miRNA regulatory network construction, virtual gene knockout analysis, and candidate drug prediction coupled with molecular docking simulations were subsequently performed.

Through this multi-layered approach, EFEMP2 (EGF Containing Fibulin Extracellular Matrix Protein 2) was consistently identified as a core fibroblast hub gene in OA. It was found to be markedly upregulated in disease-associated, pathogenic fibroblast states. Downstream functional and pathway analyses suggested that EFEMP2 heavily regulates extracellular matrix (ECM) organization, cell–matrix signaling cascades, and the dynamic remodeling of the local immune microenvironment. Moreover, candidate drug screening and structural molecular docking analyses indicated that minocycline hydrochloride, quercetin dihydrate, progesterone, and vitinoin exhibit strong binding affinities to EFEMP2. These compounds may serve as potent EFEMP2-targeting agents, providing preliminary but promising therapeutic insights.

In summary, EFEMP2 emerges as a crucial genetic mediator of fibroblast-driven tissue remodeling and immune regulation in the pathogenesis of osteoarthritis. The systematic integration of genetic causal inference with high-resolution single-cell and bulk transcriptomic evidence not only clarifies the underlying mechanistic role of EFEMP2 in OA but also successfully identifies candidate pharmacological compounds. These findings offer a robust theoretical foundation for future in vitro and in vivo functional studies, potentially accelerating clinical translation and targeted precision interventions for OA patients.