Integrating multiple microarray datasets to explore the significance of ferroptosis regulators in the diagnosis and subtype classification of osteoarthritis

Abstract

Abstract Background Osteoarthritis (OA) is a prevalent chronic joint disease that reduces the quality of life. Ferroptosis plays a significant part in various biological processes. However, uncertainty surrounds the mechanism of action that underlying ferroptosis in OA. Methods In this study, we integrated seven OA synovial datasets (GSE1919, GSE12021, GSE46750, GSE55235, GSE55457, GSE82107, and GSE89408) from the Gene Expression Omnibus database to screen significant ferroptosis-related genes. Minimal residuals of Support Vector Machine (SVM) and Random Forest (RF) were compared to select the better model for subsequent analysis. Top five ferroptosis regulators in better model were used to construct nomogram models to predict the prevalence of OA patients. Consensus clustering was applied to classify OA patients into different ferroptosis pattern based on the significant ferroptosis-related genes and divide OA patients into distinct genomic subtypes based on the ferroptosis-related differentially expressed genes (DEGs) between different ferroptosis pattern. Subsequently, an immune infiltration study was performed to investigate the relationship between important ferroptosis regulators and immune cells. Single sample gene set enrichment analysis (ssGSEA) was utilized to assess the quantity of immune cells in OA samples. Finally, using principal component analysis (PCA), we calculated the ferroptosis score for each sample in both ferroptosis patterns, to quantify the patterns. Results we screened 11 significant ferroptosis-related genes in OA and five candidate ferroptosis regulators (SLC7A11, ALOX5, SLC1A5, GOT1, and GSS) were screened using the RF model to predict OA risk. The nomogram model based on these five genes proved important for assessing OA occurrence, and both the decision analysis curve and clinical impact curves indicated that the model has unique clinical diagnostic advantages. Consensus clustering analysis showed that patients with OA can be classified into two ferroptosis patterns (Clusters A and B). ssGSEA revealed that immune infiltration levels were higher in Cluster B than A and that ALOX5 expression was positively correlated with many immune cells. Two subtypes, gene Clusters A and B, were classified according to the expression of ferroptosis-related DEGs among the molecular subtypes in the ferroptosis pattern. The comparative expression of the 11 ferroptosis regulators and immune infiltration levels between gene Clusters A and B were similar to the results obtained in the ferroptosis model, validating the accuracy of the consensus clustering approach for grouping. The PCA results showed that Cluster A and gene Cluster A had a higher ferroptosis score than Cluster B or gene Cluster B, whereas Cluster B or gene Cluster B had higher expression levels of the proinflammatory cytokines interleukin (IL)-β, tumor necrosis factor, IL-6, IL-18, and IL-10. Conclusion In summary, different subtypes of ferroptosis play critical roles in OA. Furthermore, immunotherapy strategies for the treatment of OA may be guided by our study of ferroptosis patterns.