垂体瘤发生卒中的影响因素及预测模型构建

Influencing factors for pituitary apoplexy in patients with pituitary tumor and establishment of a predictive model

摘要:

目的 探讨影响垂体瘤患者发生卒中的相关因素并构建预测模型。方法 回顾性收集哈尔滨医科大学附属第一医院神经外科2021年4月—2022年10月收治的350例垂体瘤患者临床资料。根据是否发生脑卒中将其分为卒中组（n=185）和非卒中组（n=165），构建预测模型。进一步收集2022年11月—2023年7月收治的226例垂体瘤患者作为外部时段验证队列评估模型预测效能。结果 两组间视力减退、视野缺损、呕吐、动眼神经麻痹的占比比较，差异有统计学意义（P<0.05）。多因素Logistic回归分析显示，女性（OR：2.062，95%CI：1.322～3.216），巨腺瘤（OR：13.44，95%CI：2.658～67.986）是垂体瘤患者发生卒中的影响因素。建模队列内部验证的受试者操作特征曲线下面积（AUC）为0.650，外部时段验证队列的AUC为0.713。结论 女性、巨腺瘤是垂体瘤患者发生卒中的影响因素；基于影响因素构建的垂体瘤卒中风险列线图模型具有一定的预测效能。

Abstract:

Objective To investigate the influencing factors for pituitary apoplexy in patients with pituitary tumor， and to establish a predictive model.Methods A retrospective analysis was performed for the clinical data of 350 patients with pituitary tumor who were admitted to Department of Neurosurgery， The First Affiliated Hospital of Harbin Medical University， from April 2021 to October 2022. According to the presence or absence of pituitary apoplexy， the patients were divided into stroke group with 185 patients and non-stroke group with 165 patients， and a predictive model was established. A total of 226 patients with pituitary tumor who were admitted from November 2022 to July 2023 were enrolled as an external temporal validation set to assess the predictive performance of the model.Results There were significant differences between the two groups in the proportion of patients with visual deterioration， visual field defect， vomiting， and oculomotor nerve palsy （P<0.05）. The multivariate logistic regression analysis showed that female sex （OR=2.062， 95%CI： 1.322-3.216） and macroadenoma （OR=13.44， 95%CI： 2.658-67.986） were influencing factors for pituitary apoplexy in patients with pituitary tumor. The area under the receiver operating characteristic curve was 0.650 for internal validation of the modeling cohort and 0.713 for the external temporal validation cohort.Conclusions Female sex and macroadenoma are influencing factors for pituitary apoplexy in patients with pituitary tumor， and the nomogram model established based on the influencing factors has a certain efficacy in predicting the risk of pituitary apoplexy.