摘要： 目的 筛查抗N-甲基-D-天冬氨酸受体（NMDAR）脑炎合并急性症状性癫癎发作的危险因素，并基于该危险因素建立风险预测列线图（Nomogram）模型。方法 纳入2012年5月至2020年10月首都医科大学附属北京天坛医院收治的84例抗NMDAR脑炎患者，采用单因素和多因素Logistic回归分析筛查抗NMDAR脑炎合并急性症状性癫癎发作的危险因素，并构建Nomogram模型，绘制该模型受试者工作特征（ROC）曲线和校准曲线，且行Hosmer-Lemeshow拟合优度检验。结果 84例抗NMDAR脑炎患者中有63例（75%）病程中发生急性症状性癫癎发作，其中35例（41.67%）以急性症状性癫癎发作首发。Logistic回归分析显示，男性（OR=7.680，95% CI：1.811~32.562；P=0.006）、精神行为异常（OR=6.486，95% CI：1.818~23.141；P=0.004）和脑脊液抗NMDAR抗体滴度 ≥ 1：32（OR=9.322，95% CI：2.132~40.766；P=0.003）是抗NMDAR脑炎合并急性症状性癫癎发作的危险因素，高龄是合并急性症状性癫癎发作的保护因素（OR=0.942，95% CI：0.903~0.983；P=0.006）。根据这4项危险因素建立Nomogram模型，ROC曲线下面积为0.862（95% CI：0.776~0.949，P=0.000），校准曲线显示该模型预测急性症状性癫癎发作的概率与实际概率之间一致性良好，Hosmer-Lemeshow拟合优度检验显示差异无统计学意义（χ²=4.318，P=0.827），表明Nomogram模型的区分度、校准度和稳定性均良好。结论 男性、低龄、精神行为异常和脑脊液抗NMDAR抗体滴度 ≥ 1：32的抗NMDAR脑炎患者更易并发急性症状性癫癎发作，据此构建的Nomogram模型可简便、准确地预测急性症状性癫癎发作的风险，为临床诊断与治疗提供参考。

关键词: 列线图, 抗N-甲基-D-门冬氨酸受体脑炎, 癫痫, 危险因素, Logistic模型

Abstract: Objective To investigate the risk factors of anti-N-methyl-D-aspartate receptor (NMDAR) encephalitis with acute symptomatic seizure (ASS), and establish a Nomogram model based on clinical indicators. Methods A retrospective analysis was performed on 84 patients with anti-NMDAR encephalitis who were diagnosed in Beijing Tiantan Hospital, Capital Medical University from May 2012 to October 2020. Univariate and multivariate Logistic regression analyses were used to determine the risk factors of anti-NMDAR encephalitis with ASS, and a Nomogram model was constructed. The receiver operating characteristic (ROC) curve and calibration curve of the model were plotted, and Hosmer-Lemeshow goodness of fit test was performed. Results In all 84 patients, 63 cases (75%) presented with ASS, and 35 cases (41.67%) had ASS as the initial symptom. Logistic regression analysis showed male (OR=7.680, 95%CI:1.811-32.562; P=0.006), psychosis and abnormal behavior (OR=6.486, 95%CI:1.818-23.141; P=0.004), anti-NMDAR antibody titer in cerebrospinal fluid (CSF) ≥ 1:32 (OR=9.322, 95%CI:2.132-40.766; P=0.003) were the risk factors of anti-NMDAR encephalitis with ASS, older age was the protective factor (OR=0.942, 95%CI:0.903-0.983; P=0.006). A Nomogram model was established based on the 4 risk factors, and the area under curve (AUC) of ROC was 0.862 (95%CI:0.776-0.949, P=0.000). Calibration curve showed a good agreement between the predicted probability and the actual probability of ASS occurrence, and Hosmer-Lemeshow test showed there was no statistical difference (χ²=4.318, P=0.827). Conclusions Male, young age, psychosis and abnormal behavior, and anti-NMDAR antibody titer in CSF ≥ 1:32 are more likely to have ASS in patients with anti-NMDAR encephalitis. This Nomogram model constructed based on the above 4 indicators may accurately and conveniently predict ASS and provide a reference for clinical diagnosis and treatment.

Key words: Nomograms, Anti-N-methyl-D-aspartate receptor encephalitis, Epilepsy, Risk factors, Logistic models