中国麻风皮肤病杂志 ›› 2023, Vol. 39 ›› Issue (7): 473-478.doi: 10.12144/zgmfskin202307473

• 论著 • 上一篇    下一篇

CLE进展为SLE的影响因素分析及预测模型建立和评价

汪婷1,2,姚田华3,陶康1,黄金羽3,李时飞1,王红迁4,伍亚舟3,翟志芳1   

  1. 1陆军军医大学第一附属医院皮肤科，重庆，400038；2重庆市垫江县人民医院皮肤科，重庆垫江，408300；3陆军军医大学军事预防医学系军队卫生统计学教研室，重庆，400038；4陆军军医大学第一附属医院医学大数据与人工智能中心，重庆，400038
  • 出版日期:2023-07-15 发布日期:2023-07-05

Analysis of risk factors of progression from cutaneous lupus erythematosus to systemic lupus erythematosus and the establishment of predictive models

WANG Ting1,2, YAO Tianhua3, TAO Kang1, HUANG Jinyu3, LI Shifei1, WANG Hongqian4, WU Yazhou3, ZHAI Zhifang1   

  1. 1 Department of Dermatology and Venerology, the First Affiliated Hospital of Army Medical University，Chongqing  400038, China;
    2 Department of Dermatology，Dianjiang People's Hospital，Dianjiang  408300, Chongqing, China;
    3 Department of Military Preventive Medicine，Army Medical University, Chongqing 400038, China;
    4 Center for Medical Data and Artificial Intelligencethe, First Affiliated Hospital of Army Medical University, Chongqing  400038, China
  • Online:2023-07-15 Published:2023-07-05

PDF (PC)

87

摘要： 目的：分析皮肤型红斑狼疮（cutaneous lupus erythematosus, CLE）患者临床进展为系统性红斑狼疮（systemic lupus erythematosus, SLE）的影响因素，并建立预测模型。方法：收集2010年1月至2019年12月首次就诊于我院且明确诊断为CLE的患者，统计其初诊时及随访3年以上复诊时临床及实验室相关指标，根据其是否进展为SLE分为两组，比较分析两组患者首诊时的临床及实验室指标，对存在组间差异的指标进行相关性分析，并进行预测建模评价。结果：最终共纳入CLE患者183例，其中未进展组120例，进展组63例。单因素回归分析结果表明，性别、白细胞数目、红细胞数目、血红蛋白、血小板数目、补体C3、补体C4、抗核抗体（ANA）、抗U1RNP抗体、抗Ro-52抗体、脱发是与狼疮进展相关的影响因素（均P＜0.05）。根据上述特征建立KNN、Logistic、Bagging、BP、SVM五种预测模型，各模型的AUC值依次为0.7200、0.7861、0.6035、0.8191、0.6591。结论：女性患者、脱发以及血常规、补体、ANA、抗U1RNP抗体、抗Ro-52抗体等实验室指标均为CLE向SLE进展的影响因素，据此构建的疾病转归预测模型证实具有较高的准确度与区分度。

关键词: 皮肤型红斑狼疮, 系统性红斑狼疮, 疾病转归, 影响因素, 预测模型

Abstract: Objective: To analyze the risk factors of clinical progression to systemic lupus erythematosus (systemiclupuserythematosus, SLE) in patients with cutaneous lupus erythematosus (cutaneouslupuserythematosus, CLE) and establish predictive models. Methods: The patients who were first diagnosed as CLE in our hospital from January 2010 to December 2019 were collected, and the clinical and laboratory data were statistically analyzed. They were divided into two groups according to their progression to SLE or not. The clinical and laboratory data of the two groups at the first visit were compared, and they were incorporated into five predictive modeling to verify the effectiveness. Results: A total of 183 patients with CLE were analyzed, including non-progressive group (n=120) and progressive group (n=63). Univariate regression analysis showed that sex, WBC, RBC, HGB, PLT , complement 3, complement C4, ANA, anti-U1RNP antibody, anti-Ro-52 antibody and alopecia were related to the progression of lupus erythematosus (Ps＜0.05). According to the above characteristics, the prediction models were established, and the test results showed that the areas under the curve (AUC) of KNN, logistic, Bagging, BP, SVM  models were 0.7200, 0.7861, 0.6035, 0.8191, 0.6591, respectively. Conclusion: The risk factors of the progression from CLE to SLE include female patients, alopecia, and laboratory data such as blood routine, complement and ANA, anti-U1RNP antibody and anti-Ro-52 antibody et al. The lupus  progression prediction models we constructed have high accuracy and differentiation.

Key words: cutaneous lupus erythematosus, systemic lupus erythematosus, disease outcome, risk factors, predictive model