Investigating potential biomarkers associated with antineutrophil cytoplasmic antibody (ANCA)-associated vasculitis using Mendelian randomization and transcriptomic analysis.

BACKGROUND: Antineutrophil cytoplasmic antibody (ANCA)-associated vasculitis (AAV) is an autoimmune disorder characterized by multi-organ involvement. Early identification and accurate diagnosis of AAV is crucial for improving prognosis. However, research on biomarkers associated with AAV is limited. This study aimed to explore novel biomarkers for AAV through transcriptomic data analysis and Mendelian randomization (MR). METHODS: AAV-related datasets (GSE104948 and GSE108109) were analyzed. Differentially expressed genes (DEGs) between AAV and normal groups were identified in the GSE104948 dataset. MR analysis was then used to investigate the causal relationship between DEGs and AAV. Genes with a significant causal relationship were selected as candidate genes for further analysis. Machine learning algorithms, ROC curve analysis, and expression evaluation were employed to screen for biomarkers. Additionally, artificial neural networks (ANNs) were constructed, enrichment analysis and immune infiltration were performed, a molecular regulatory network was established, and potential drugs were predicted. Finally, immunofluorescence assays validated the significance of these genes in renal biopsies from patients with ANCA-associated glomerulonephritis. RESULTS: PDK4, PSMB10 (IVW, OR > 1, P < 0.05), PPARGC1A, and FN1 (IVW, OR < 1, P < 0.05) were identified as biomarkers. Specifically, PDK4 and PPARGC1A exhibited significant down-regulation in the AAV group compared to the normal group, while FN1 and PSMB10 showed an opposite pattern. The ANN created based on biomarkers exhibited a robust predictive capacity for assessing the risk of AAV. Furthermore, co-enrichment of PDK4 and PPARGC1A was observed in 'butanoate metabolism', and 'fatty acid metabolism'. Meanwhile, there was a strong positive correlation observed between naive B cells and PDK4, while a substantial negative correlation was found with PSMB10. Molecular regulatory network results demonstrated that XIST exerted regulatory effects on PDK4, FN1, and PPARGC1A through hsa-miR-103a-3p, hsa-miR-1271-5p, and hsa-miR-23a-3p simultaneously. Besides, this study revealed that 19 drugs exhibited potential targeting capabilities towards 4 biomarkers, such as dacarbazine, dichloroacetate, and bortezomib. Validation in renal biopsies from patients with ANCA-associated glomerulonephritis confirmed decreased glomerular expression of PDK4 and PPARGC1A, and increased expression of FN1 and PSMB10 compared to controls. CONCLUSION: PDK4, PPARGC1A, FN1, and PSMB10 were identified as biomarkers causally related to AAV, offering potential for both precise diagnosis and targeted treatment strategies.