Predicting endovascular recanalization success in symptomatic chronic carotid occlusion: A decision tree model based on 321 cases

Background

Endovascular recanalization for chronic internal carotid artery occlusion (CICAO) remains technically challenging, with variable success rates and a lack of reliable predictive tools for patient selection. We aim to analyze risk factors associated with failed recanalization in CICAO patients and develop a decision tree model to quantify individualized recanalization potential.

Methods

We retrospectively analyzed 321 patients with symptomatic CICAO who underwent endovascular recanalization. Univariate and multivariate analyses were used to identify risk factors for recanalization failure. A decision tree model and logistic model were constructed. Models’ performance was evaluated using AUC analysis and decision curve analysis (DCA).

Results

The overall recanalization success rate was 61.7 %. Our study identified three independent risk factors for failed recanalization: occlusion length > 10 cm, type III stump morphology, and contralateral internal carotid artery stenosis. The decision-tree model demonstrated good performance (AUC 0.839 in training, 0.834 in validation) and provided clinical interpretability compared to logistic regression. DCA confirmed clinical utility across probability thresholds.

Conclusions

We developed and validated a decision tree model that effectively predicts endovascular recanalization success in CICAO patients, which may serve as a valuable tool to support clinical decision-making for patients with CICAO.