外侧与内侧延髓梗死临床特征分析及预后预测模型构建

Clinical feature analysis of lateral and medial medullary infarction and prognostic model construction

摘要:

目的 探讨外侧与内侧延髓梗死临床特征，并构建相关预后预测模型。方法 回顾性纳入2019年1月至2024年1月于宝鸡市陈仓医院诊治的延髓梗死患者，共251例。根据供血动脉分为外侧组（152例）和内侧组（99例）。比较两组临床特征及预后情况；采用多因素Logistic回归分析探讨外侧与内侧延髓梗死预后不良的危险因素。基于多因素Logistic回归分析的结果，构建外侧与内侧延髓梗死预后不良的多因素Logistic预测模型，采用Bootstrap内部验证（1 000次重复抽样）校正模型效能。结果 两组的头晕比例、吞咽困难比例、构音障碍比例、运动障碍比例、舌肌麻痹比例、共济失调比例、霍纳综合征比例、呃逆比例、早期神经功能恶化比例、病灶分布位置及Org 10172急性脑卒中的临床治疗试验（TOAST）分型比较，差异有统计学意义（P<0.05）。外侧组出院时和随访3个月预后不良率显著低于内侧组（P<0.05）；两组随访3个月病死率比较差异无统计学意义（P>0.05）。年龄、高血压、早期神经功能恶化均是外侧延髓梗死预后不良的危险因素（P<0.05）；年龄、早期神经功能恶化是内侧延髓梗死预后不良的危险因素（P<0.05）。基于多因素Logistic预测模型对外侧与内侧延髓梗死预后不良风险进行受试者操作特征曲线分析，曲线下面积（AUC）分别为0.967、0.921。经Bootstrap内部验证后，外侧组校正后AUC为0.872（95%CI：0.821～0.923），内侧组校正后AUC为0.845（95%CI：0.789～0.901）。结论 外侧与内侧延髓梗死患者间存在明显临床特征及预后差异。利用各危险因素构建的模型对于患者预后的预测显示出良好的效能。

Abstract:

Objective To analyze the clinical features of lateral and medial medullary infarction， and to construct prognostic models.Methods A total of 251 patients with medullary infarction admitted to and treated in Chencang Hospital， Baoji from January 2019 to January 2024 were enrolled retrospectively， and divided into a lateral group （152 cases） and a medial group （99 cases） based on feeding arteries. Clinical features and prognoses were compared between the two groups. A multivariable logistic regression analysis was performed to identify risk factors for poor prognosis in lateral and medial medullary infarction. Based on these results， logistic prediction models for unfavorable prognosis were constructed for the two groups. Bootstrap internal validation （1 000 repeated samplings） was then applied to correct for overfitting model performance.Results Statistically significant intergroup differences were observed in the incidence of concomitant dizziness， dysphagia， dysarthria， motor dysfunction， lingual paralysis， ataxia， Horner syndrome， hiccup， and early neurological deterioration （END）， as well as in lesion distribution and Trial of Org 10172 in Acute Stroke Treatment （TOAST） classification （P<0.05）. The rates of poor prognosis at discharge and at the 3-month follow-up were significantly lower in the lateral group than in the medial group （P<0.05）， whereas no significant between-group difference was found in 3-month all-cause mortality （P>0.05）. Age， hypertension， END， were independent risk factors for unfavorable prognosis in patients with lateral medullary infarction （P<0.05）； age， END， and large-artery atherosclerosis were independent risk factors for unfavorable prognosis in patients with medial medullary infarction （P<0.05）. Receiver operating characteristic curves were plotted to assess the predictive performance of the two logistic models， with areas under the curve （AUC） of 0.967 and 0.921 for lateral and medial medullary infarction， respectively. After internal validation via Bootstrap resampling with 1 000 replicates， the bias-corrected AUC was 0.872 （95%CI： 0.821-0.923） for the lateral model and 0.845 （95%CI： 0.789-0.901） for the medial model.Conclusions There are significant differences in clinical symptoms and prognosis between patients with lateral and medial medullary infarction. The prediction models constructed using these independent risk factors have good performance in predicting patient prognosis.