Abstract
BACKGROUND: Some acute ischemic stroke (AIS) patients due to large-vessel occlusion, who underwent endovascular thrombectomy (EVT), continue to experience unfavorable outcomes. Furthermore, the impact of internal carotid artery (ICA) tortuosity remains uncertain. This study aimed to determine the value of ICA tortuosity and clinical features in predicting 3-month unfavorable outcome and early neurological deterioration (END) after EVT in AIS patients through nomograms. METHODS: A total of 313 AIS patients treated with EVT at the First Affiliated Hospital of Xi'an Jiaotong University were retrospectively analyzed and randomized into two cohorts: training cohort (n=219) and validation cohort (n=94). After the selection of relevant features, nomograms for predicting the 3-month unfavorable outcome (mRS > 2) and END (an increase in NIHSS score of ≥4 within 24 hours) were established. The predictive accuracy of the nomograms was evaluated using ROC curves, calibration plots, and decision curve analysis (DCA). RESULTS: Among 313 patients, ICA tortuosity was observed in 19.50% (extracranial) and 21.10% (cavernous) of patients. Furthermore, 53.30% of patients experienced a 3-month unfavorable outcome, while END occurred in 15.70%. The independent predictors for the 3-month unfavorable outcome included age, NIHSS score, puncture-to-recanalization time, eTICI score, and blood glucose. The addition of two tortuosity features (extracranial and cavernous ICA tortuosity) resulted in a significant improvement in model differentiation. The nomogram that included ICA tortuosity achieved an AUC of 0.826 and 0.803 in the training and validation cohorts. ASPECT score, occlusion site, number of retriever passes, and blood glucose were identified as factors associated with END. The AUC was 0.770 and 0.772 in the training and validation cohorts. However, the incorporation of ICA tortuosity did not significantly enhance the model for predicting END. CONCLUSION: ICA tortuosity characteristics significantly improve the discrimination of the nomogram model in predicting the 3-month unfavorable outcome. This can be used as guidance in clinical decision-making.