Kong Wei, Zhang Xin, Geng Linyu, Chen Chen, Sun Yue, Xu Xue, Zhao Shengnan, Jin Ziyi, Huang Yang, Wang Dandan, Liang Jun, Zhu Yun, Sun Lingyun
Department of Rheumatology and Immunology, Nanjing Drum Tower Hospital, Affiliated Hospital of Medical School, Nanjing University, No. 321 Zhongshan Road, Nanjing, Jiangsu, 210008, China.
Department of Clinical Nutrition, Nanjing Drum Tower Hospital, Affiliated Hospital of Medical School, Nanjing University, No. 321 Zhongshan Road, Nanjing, Jiangsu, 210008, China.
Clin Rheumatol. 2025 Apr;44(4):1729-1743. doi: 10.1007/s10067-025-07377-0. Epub 2025 Feb 28.
As systemic lupus erythematosus (SLE) primarily impacts women of childbearing age, a considerable number of patients have fertility needs. However, the risk of experiencing adverse pregnancy outcomes (APOs) was higher in these patients. Our study aimed to construct a predictive model to assess the risks for APOs of SLE.
We retrospectively analyzed the data of pregnant SLE patients hospitalized at Nanjing Drum Tower Hospital from August 2010 to April 2023. The Least Absolute Shrinkage and Selection Operator (LASSO) regression analysis was used to explore the risk factors for APOs, and a nomogram was established. Afterward, the efficacy of the nomogram was evaluated by analyzing the areas under the curves (AUCs) of Receiver Operating Characteristic (ROC), calibration curves, and Decision Curve Analysis (DCA).
Our study involved 259 pregnant patients with a median age of 29.00 years, and identified 129 cases of APOs, including preterm birth, low birth weight, congenital anomalies, stillbirth/miscarriage, and fetal distress. Through LASSO regression analysis, nine optimal features were selected as risk factors, including age, lupus nephritis, antepartum body mass index, antinuclear antibody, anti-U1RNP/Sm antibody, anti-ribosomal P protein antibody, platelet, albumin levels, SLEDAI scores, diabetes mellitus, rash, and the use of aspirin therapy. These factors were integrated into a predictive nomogram model, which showed good predictive accuracy, with AUC values of 0.870 and 0.830 in training and validation groups, respectively. The calibration curves and DCA also confirmed the good performance of the model.
We developed a tool to predict APOs in SLE patients, offering personalized risk assessments and clinical decision support. As the data used to build the predictive model was obtained from a single center, the tool is currently best suited for application within our center. Further validation in diverse populations is needed to expand its generalizability. Key Points • Our study revealed the independent predictors for APOs of SLE through LASSO regression analysis. • We developed a nomogram to predict APOs in SLE based on the results of LASSO regression analysis. • The predictive model may aid clinical decision-making, enabling timely interventions to reduce the incidence of APOs.
由于系统性红斑狼疮（SLE）主要影响育龄女性，相当一部分患者有生育需求。然而，这些患者出现不良妊娠结局（APO）的风险较高。我们的研究旨在构建一个预测模型，以评估SLE患者发生APO的风险。
我们回顾性分析了2010年8月至2023年4月在南京鼓楼医院住院的妊娠SLE患者的数据。采用最小绝对收缩和选择算子（LASSO）回归分析来探索APO的危险因素，并建立了列线图。随后，通过分析受试者操作特征（ROC）曲线下面积（AUC）、校准曲线和决策曲线分析（DCA）来评估列线图的效能。
我们的研究纳入了259例妊娠患者，中位年龄为29.00岁，其中129例发生APO，包括早产、低出生体重、先天性畸形、死产/流产和胎儿窘迫。通过LASSO回归分析，选择了9个最佳特征作为危险因素，包括年龄、狼疮性肾炎、产前体重指数、抗核抗体、抗U1RNP/Sm抗体、抗核糖体P蛋白抗体、血小板、白蛋白水平、SLE疾病活动指数（SLEDAI）评分、糖尿病、皮疹以及阿司匹林治疗的使用情况。这些因素被整合到一个预测列线图模型中，该模型显示出良好的预测准确性，训练组和验证组的AUC值分别为0.870和0.830。校准曲线和DCA也证实了该模型的良好性能。
我们开发了一种工具来预测SLE患者的APO，提供个性化的风险评估和临床决策支持。由于用于构建预测模型的数据来自单一中心，该工具目前最适合在我们中心内应用。需要在不同人群中进行进一步验证，以扩大其通用性。要点 • 我们的研究通过LASSO回归分析揭示了SLE患者发生APO的独立预测因素。 • 我们根据LASSO回归分析的结果开发了一个列线图来预测SLE患者的APO。 • 该预测模型可能有助于临床决策，能够及时进行干预以降低APO的发生率。
