Qi Xuan, Zhao Doudou, Wang Naidi, Han Yipeng, Huang Bo, Feng Ruiling, Jin Yuebo, Wang Ruoyi, Lin Xiang, He Jing
Department of Rheumatology and Immunology, Peking University People's Hospital, Beijing, China.
Department of Rheumatism and Immunology, The Second Hospital of Hebei Medical University, Shijiazhuang, Hebei, China.
Clin Rheumatol. 2025 Apr 29. doi: 10.1007/s10067-025-07434-8.
This study aimed to perform a detailed stratification analysis of B lymphocyte subsets in patients with primary Sjögren's syndrome (pSS) and to investigate their associations with lymphoma risk, clinical phenotypes, and disease activity.
In this retrospective study, we analyzed data from 137 patients with pSS. We employed machine learning approaches, specifically principal component analysis (PCA) and k-means clustering, to examine B lymphocyte subset distributions from flow cytometry data and immunoglobulin IgG and complement (C3, C4) levels. The optimal cluster number was determined using the Elbow Method in R software. Based on these 10 variables, patients were categorized into distinct subgroups. We then comprehensively compared clinical characteristics, laboratory parameters, and disease activity indices among these identified subgroups.
Four distinct subgroups were identified. Cluster A exhibited a significantly higher lymphoma incidence rate of 20%, compared to 3.39% in Cluster B and 0% in Clusters C and D (p = 0.007). Cluster A also had the highest percentage of double-negative B cells (32.26 ± 17.96%) and plasma cells (2.02 ± 1.92%). ESSDAI scores indicated that disease activity was highest in Cluster A (9.00, 6.00-20.00), followed by Clusters B (7.00, 3.50-14.00), C (6.00, 1.25-17.50), and D (5.00, 1.50-9.00), respectively.
This innovative stratification method revealed the critical role of B cell subset imbalance in the pathogenesis of pSS and provided new evidence for predicting lymphoma risk and guiding personalized treatment. Key Points • Identifying a distinct patient subgroup with elevated lymphoma risk and increased disease activity could aid in risk prediction. • Applying machine learning techniques to stratify B cell populations provides insights into pSS pathogenesis. • A proposed framework for personalized treatment approaches based on B cell subset imbalances in pSS.
本研究旨在对原发性干燥综合征（pSS）患者的B淋巴细胞亚群进行详细的分层分析，并探讨它们与淋巴瘤风险、临床表型及疾病活动度的相关性。
在这项回顾性研究中，我们分析了137例pSS患者的数据。我们采用机器学习方法，特别是主成分分析（PCA）和k均值聚类，来研究流式细胞术数据中的B淋巴细胞亚群分布以及免疫球蛋白IgG和补体（C3、C4）水平。使用R软件中的肘部法则确定最佳聚类数。基于这10个变量将患者分为不同亚组。然后我们全面比较了这些已确定亚组之间的临床特征、实验室参数及疾病活动指数。
确定了四个不同的亚组。A组淋巴瘤发病率显著更高达20%，相比之下B组为3.39%，C组和D组为0%（p = 0.007）。A组双阴性B细胞百分比（32.26±17.96%）和浆细胞百分比（2.02±1.92%）也最高。ESSDAI评分表明疾病活动度在A组最高（9.00，6.00 - 20.00），其次分别为B组（7.00，3.50 - 14.00）、C组（6.00, 1.25 - 17.50）和D组（5.00，1.50 - 9.00）
这种创新的分层方法揭示了B细胞亚群失衡在pSS发病机制中的关键作用，并为预测淋巴瘤风险和指导个性化治疗提供了新证据要点：识别出淋巴瘤风险升高和疾病活动度增加的独特患者亚组有助于风险预测。应用机器学习技术对B细胞群体进行分层可深入了解pSS发病机制。提出基于pSS中B细胞亚群失衡的个性化治疗方法框架。
