RESUMO
OBJECTIVES: Vascular anatomy of the aortic arch and supra-aortic arteries has been suggested as influencing the risk of carotid artery stenting (CAS). The expert opinion based Delphi anatomical risk (DAR) score was developed to predict difficulty of CAS in relation to procedural stroke risk, and thereby aid patient selection. The aim was to validate the DAR score in the context of a randomised clinical trial. METHODS: In this post hoc analysis of the International Carotid Stenting Study (ICSS), only patients treated by CAS with available pre-procedural CT angiography (CTA) were included. Patients with tortuous anatomy unsuitable for stenting were excluded from ICSS. CTA based vascular anatomy was rated by two independent observers. Every possible combination of anatomy resulted in a risk score, divided in four categories of expected risk (low, < 5.0; low-intermediate, 5.0-5.9; high-intermediate, 6.0-6.9; high, ≥ 7.0). Binomial logistic regression was used to assess the relationship between anatomical risk score and procedural risk of any stroke. Differences between predefined age groups were also assessed. RESULTS: A total of 275 patients were included. Interobserver reliability for all anatomical risk factors was high (κ = 0.76-0.84). In total, 16 strokes (6%) occurred in the procedural period. No significant relationship was observed between the DAR score and risk of procedural stroke, with the risk of stroke being 9% in the high risk vs. 4% in the low risk categories (p = .49). A higher mean DAR score was observed in patients ≥70 years compared with younger patients (4.6 ± 1.5 vs. 3.9 ± 1.4, p < .001), which was mainly explained by higher rates of arch atheroma (44% vs. 20%, p < .001). Prolonged intervention duration was significantly associated with increased stroke risk (11% vs. 4%, p < .04), but not with the DAR score. CONCLUSIONS: No statistically significant association was found between anatomical difficulty, as defined in the DAR score, and procedural stroke risk. However, the small sample size potentially rendered the study underpowered to detect group differences, and confirmation with a larger sample is essential.