妊娠期系统性红斑狼疮患者不良妊娠结局的影响因素及列线图预测模型的构建

Influencing factors of adverse pregnancy outcomes in patients with systemic lupus erythematosus during pregnancy and construction of a nomogram model for predicting

目的探讨妊娠期系统性红斑狼疮(SLE)患者不良妊娠结局的影响因素,并构建其不良妊娠结局的列线图预测模型。方法选择2016年9月至2021年9月万宁市妇幼保健院收治的妊娠期SLE患者362例,采用随机抽样法按照7∶3的比例,将所有患者随机分配至独立训练集(n=254)与独立测试集(n=108)。将独立训练集中的254例妊娠期SLE患者分为训练集妊娠结局不良组(n=56)与训练集妊娠结局良好组(n=198),对其进行单因素分析。采用Logistic回归分析妊娠期SLE患者发生不良妊娠结局的危险因素,并构建列线图模型。在独立训练集与独立测试集中,采用受试者工作特征(ROC)曲线、校准曲线、决策曲线对列线图进行交叉验证。采用Pearson相关系数研究列线图评分与各临床危险因素的相关性。结果训练集妊娠结局不良组有肺动脉高压、凝血功能异常、肝功能不全、狼疮性肾炎、心功能不全、子痫前期、肾功能不全、白细胞减少、淋巴细胞减少、血小板减少、血红蛋白下降、补体下降、蛋白尿>0.5g/24h、抗dsDNA抗体(+)、抗SSB抗体(+)、抗SSA抗体(+)、抗Sm抗体(+)、β2糖蛋白(+)、ANA(+)、妊娠期间治疗(使用羟氯喹、环孢素、免疫抑制剂)、雷诺现象的占比,以及SLEDAI评分均高于训练集妊娠结局良好组(χ^(2)=4.196~88.082、t=11.619,P<0.05)。经Logistic回归分析显示,狼疮性肾炎、子痫前期、血小板减少、补体下降、雷诺现象、SLEDAI评分均是影响妊娠期SLE患者不良妊娠结局的危险因素(Waldχ^(2)=4.173~27.411,P<0.05)。在独立训练集中,列线图模型达到了较高的区分度,曲线下面积(AUC)为0.887(95%CI:0.835~0.939);在独立测试集中,列线图模型同样具有较高的区分度,AUC为0.846(95%CI:0.812~0.875),列线图模型具有较好的预测准确度。Hosmer-Lemeshow拟合优度检验χ^(2)=7.178,P=0.518。列线图评分与SLEDAI评分、狼疮性肾炎、血小板减少、补体下降、子痫前期、雷诺现象均呈显著正相关(r分别为0.479、0.312、0.263、0.214,0.201、0.190,P<0.05)。结论狼疮性肾炎、子痫前期、血小板减少、补体下降、雷诺现象、SLEDAI评分均是影响妊娠期SLE患者不良妊娠结局的危险因素;据此构建高精度的列线图模型具有较好的预测准确度,为妊娠期SLE患者预防该类风险提供了一项临床可用的预测工具。

Objective To investigate the influencing factors of adverse pregnancy outcome in patients with systemic lupus erythematosus(SLE)during pregnancy,and to construct a nomogram model for predicting adverse pregnancy outcomes.Methods 362 patients with SLE during pregnancy admitted to Wanning Maternal and Child Health Hospital from September 2016 to September 2021 were selected.All patients were randomly assigned to the independent training set(n=254)and the independent test set(n=108)using a random sampling method in a ratio of 7:3.The 254 patients in the independent training set were divided into the poor pregnancy outcome group(n=56)and the good pregnancy outcome group(n=198)for univariate analysis.Logistic regression was used to analyze the risk factors for adverse pregnancy outcomes in SLE patients during pregnancy,and a nomogram model was constructed.In the independent training set and independent test set,the receiver operating characteristic(ROC)curve,calibrationcurve and decision curve were used to cross-validate the nomogram model.Pearson correlation coefficient was used to investigate the correlation between the nomogram model scores and clinical risk factors.Results The rate of pulmonary hypertension,abnormal coagulation,hepatic insufficiency,lupus nephritis,cardiac insufficiency,preeclampsia,renal insufficiency,leukopenia,lymphopenia,thrombocytopenia,decreased hemoglobin,decreased complement,proteinuria>0.5 g/24 h,anti-dsDNA antibody(+),anti-SSB antibody(+),anti-SSA antibody(+),anti-Sm antibody(+),β2 glycoprotein(+),ANA(+),treatment during pregnancy(use of hydroxychloroquine,cyclosporine and immunosuppressant),and Raynaud′s phenomenon,and score of SLEDAI in the training set with poor pregnancy outcomes were higher than those in the training group with good pregnancy outcomes(χ^(2)=4.196-88.082,t=11.619,P<0.05).Logistic regression analysis showed that lupus nephritis,preeclampsia,thrombocytopenia,decreased complement,Raynaud′s phenomenon,and SLEDAI score were risk factors for poor pregnancy outcomes in patients with SLE during pregnancy(Waldχ^(2)=4.173-27.411,P<0.05).In the independent training set,the nomogram achieved high discrimination with an area under the curve(AUC)of 0.887(95%CI:0.835-0.939).In the independent test set,the nomogram also had high discrimination with an AUC of 0.846(95%CI:0.812-0.875).The nomogram had good predictive accuracy.The Hosmer-Lemeshow goodness-of-fit test showedχ^(2)=7.178 and P=0.518.The nomogram model score were significantly and positively correlated with SLEDAI scores,lupus nephritis,thrombocytopenia,decreased complement,pre-eclampsia,and Raynaud′s phenomenon(r=0.479,0.312,0.263,0.214,0.201 and 0.190,P<0.05).Conclusion Lupus nephritis,preeclampsia,thrombocytopenia,decreased complement,Raynaud′s phenomenon and SLEDAI score are risk factors for adverse pregnancy outcomes in patients with SLE during pregnancy.Therefore,a high-precision nomogram model is built with good prediction accuracy,which provides a clinically available prediction tool for preventing such risks in patients with SLE during pregnancy.