A Novel Diagnostic and Subtype Classification Model Based on RNA N6-Methyladenosine Regulators for Behçet’s Uveitis

To investigate the role of RNA N6-methyladenosine (m6A) regulators in the diagnosis and subtype classification of Behçet’s uveitis, aiming to establish a novel predictive model and explore distinct molecular patterns for personalized therapeutic approaches.
Data from the Gene Expression Omnibus GSE209567 dataset comprising 22 Behçet’s uveitis patients and 15 controls were analyzed for m6A regulator expression. Differentially expressed genes were identified using the “limma” R package, followed by random forest (RF) and support vector machine (SVM) model construction to select critical m6A regulators. A nomogram model was developed for prediction, and consensus clustering identified distinct m6A and gene-regulating patterns. Immune cell infiltration analysis was conducted using ssGSEA, and m6A scores were computed to quantify molecular patterns.
Eleven m6A regulators were significantly differentially expressed. The top four candidate m6A regulators (FTO, YTHDF2, CBLL1, and METTL14) were identified to predict the risk of Behçet’s uveitis. A nomogram was constructed based on the four candidate m6A regulators to visualize the association between the expression levels of the candidate with the risk of onset of Behçet’s uveitis. Two distinct m6A patterns and gene patterns were identified, validated by consensus clustering. High m6A scores were associated with more severe disease stages, with differential immune cell infiltration observed between subtypes. Immune-related genes, such as LRRN3 and DAAM2, were identified as key in differentiating m6A patterns.
M6A modification plays an important role in the occurrence of uveitis. Distinct m6A patterns and gene clusters highlight their potential for early diagnosis and personalized treatment.

为探究RNA N6-甲基腺嘌呤（RNA N6-methyladenosine, m6A）调控因子在白塞病葡萄膜炎（Behçet’s uveitis）的诊断与亚型分型中的作用，本研究旨在构建新型预测模型，并探索可用于个性化治疗的差异化分子模式。
本研究分析了基因表达综合数据库（Gene Expression Omnibus, GEO）GSE209567数据集的数据，该数据集纳入22例白塞病葡萄膜炎患者与15例健康对照，用于分析m6A调控因子的表达水平。采用"limma" R包筛选差异表达基因，随后构建随机森林（random forest, RF）与支持向量机（support vector machine, SVM）模型以筛选关键m6A调控因子。构建用于预测的列线图（nomogram）模型，并通过一致性聚类识别差异化的m6A与基因调控模式。采用单样本基因集富集分析（single-sample gene set enrichment analysis, ssGSEA）开展免疫细胞浸润分析，同时计算m6A评分以量化分子模式。
共筛选出11个存在显著差异表达的m6A调控因子。鉴定出排名前四的候选m6A调控因子（FTO、YTHDF2、CBLL1及METTL14）可用于预测白塞病葡萄膜炎的发病风险。基于这四个候选m6A调控因子构建列线图，以可视化展示候选因子表达水平与白塞病葡萄膜炎发病风险之间的关联。经一致性聚类验证，共识别出两种差异化的m6A模式与基因模式。高m6A评分与更严重的疾病分期相关，且不同亚型间的免疫细胞浸润存在差异。鉴定出LRRN3、DAAM2等免疫相关基因可作为区分m6A模式的关键标志物。
m6A修饰在葡萄膜炎的发生发展中发挥重要作用。差异化的m6A模式与基因簇凸显了其在早期诊断与个性化治疗中的应用潜力。