Professor University of California, Irvine, United States
Introduction:: RA's pathophysiology involves persistent synovial membrane inflammation, which can lead to the destruction of articular cartilage and adjacent bone. Recent findings on biological pathways have enhanced the comprehension of rheumatoid inflammation and its effects. In the era of big data, managing vast quantities of information has become a prominent subject, supported by the continuous advancement of machine learning algorithms. This study aims to serve as a reference for RA diagnosis and treatment by identifying key feature genes using machine learning and exploring their association with immune infiltration, thereby revealing RA's molecular-level pathogenesis.
Materials and Methods:: Gene Expression Omnibus (GEO) database data, specifically GSE12021 and GSE55235, were used as training sets, while GSE55457 served as a validation set. Differential gene expressions in the training sets were analyzed, followed by the development of LASSO regression and support vector machine models using machine learning techniques. Intersection genes were identified as feature genes, and a receiver operator characteristic (ROC) curve was generated. The validation set was then employed to verify the results. Moreover, the RA expression profile was examined through immune cell infiltration, and the co-expression relationship between feature genes and immune cells was established.
Results, Conclusions, and Discussions:: Rheumatoid arthritis (RA) is a chronic, systemic autoimmune disease characterized by inflammation and progressive destruction of joints, leading to disability and reduced quality of life. Early diagnosis and treatment are crucial for improving patient outcomes, as they can prevent joint damage, reduce inflammation, and improve long-term function. Currently, the diagnosis of RA is based on clinical features, laboratory tests, and imaging findings[26]. Treatment options include nonsteroidal anti-inflammatory drugs (NSAIDs), corticosteroids, conventional synthetic disease-modifying antirheumatic drugs (csDMARDs), and targeted biologic agents, such as tumor necrosis factor (TNF) inhibitors. The importance of early diagnosis and treatment is emphasized by evidence demonstrating improved prognosis with timely interventions. The significance of this paper lies in its identification of PNRC1 and DENND1B as characteristic genes for RA, as well as their association with immune cell infiltration. These findings contribute to the understanding of RA pathogenesis and may have implications for the development of novel diagnostic and therapeutic strategies. Also, based on prior research that low difference threshold and excessive differential genes have impacted the accuracy of enrichment analysis and machine learning algorithms, in this case, we enlarged the threshold for differential analysis, setting the Log2FC to 2, which was intended to be more precise and pertinent. However, some limitations of this study include the need for larger patient cohorts and functional studies to further elucidate the mechanistic roles of PNRC1 and DENND1B in RA. Additionally, the role of other genes and pathways contributing to the complex etiology of RA warrants further investigation.
