Establishment of a nomogram prediction model for the etiological type of large vessel occlusive stroke based on clinical and imaging parameters
10.3760/cma.j.issn.1673-4165.2023.06.002
- VernacularTitle:基于临床和影像学的大血管闭塞性卒中病因学分型的列线图预测模型的建立
- Author:
Ling LI
1
;
Ruoyao CAO
;
Yao LU
;
Yun JIANG
;
Peng QI
;
Guoxuan WANG
;
Kezhen YU
;
Juan CHEN
Author Information
1. 首都医科大学附属北京世纪坛医院放射科，北京 100038
- Keywords:
Ischemic stroke;
Intracranial arteriosclerosis;
Intracranial embolism;
Tomography, X-ray computed;
Computed tomography angiography;
Perfusion imaging;
Multi
- From:
International Journal of Cerebrovascular Diseases
2023;31(6):409-417
- CountryChina
- Language:Chinese
-
Abstract:
Objective:To develop a nomogram model based on clinical and imaging parameters to predict the etiological type of acute ischemic stroke (AIS).Methods:Patients with AIS received endovascular treatment in Beijing Hospital from March 2016 to December 2021 were retrospectively included. According to the etiological type, they were divided into large artery atherosclerosis (LAA) and cardioembolism (CE). The clinical and imaging parameters mostly relevant to the etiological type were selected by LASSO regression, and a nomogram model for predicting the etiological type of AIS was established by multifactorial logistic regression to investigate the predictive value of relevant clinical imaging parameters. In addition, the diagnostic efficacy of the prediction model was assessed by receiver operator characteristic (ROC) curves, calibration curves, and clinical decision curves. Results:A total of 136 AIS patients with anterior circulation large vessel occlusion received endovascular treatment were included, including 62 patients with CE (45.6%) and 74 with LAA (54.4%). Variables with P<0.10 in the univariate analysis were included in LASSO regression to screen for relevant variables. The gender, baseline National Institute of Health Stroke Scale (NIHSS) score, penumbra to ischemic core ratio, brain natriuretic peptide (BNP), and platelet (PLT) count were included into the multivariate logistic regression model. The results revealed that gender (odds ratio [ OR] 2.632, 95% confidence interval [ CI] 1.048-6.607; P=0.039), baseline NIHSS score ( OR 1.078, 95% CI 1.002-1.160; P=0.043), BNP ( OR 1.004, 95% CI 1.002-1.007. P<0.001), PLT ( OR 0.991, 95% CI 0.982-0.999; P=0.031) as the predictors to distinguish LAA from CE. In addition, the penumbra to infarct core ratio ( OR 0.886, 95% CI 0.785-1.000; P=0.050) also played an important role in predicting the model. The diagnostic efficacy of this predictive model was analyzed by the ROC curves, with an area under the curve of 0.881 (95% CI 0.815-0.930, P<0.001). Bootstrap internal validation showed that the good compliance with a mean absolute error of 0.027 for true versus predicted value compliance. Calibration curves, clinical decision curves, and Hosmer-Lemeshow test ( P=0.562) showed good agreement between the predicted and actual values of the model. Conclusion:Patients with CE are more common in women, have higher NIHSS scores and BNP, and have lower PLT and penumbra to ischemic core ratio. The nomogram model combining the above indicators can better identify LAA and CE, and maybe helpful in clinical decision making.
