OBJECTIVE: The type III arch configuration has been inconsistently reported as a stroke risk factor during carotid artery stenting. However, at least three different methods for the definition of type III arch can be identified in the literature, related to the level of the origin of the innominate artery (IA). According to Casserly's definition, a type III arch presents with an origin of the IA below the horizontal plane of the inner curvature. According to Madhwal's definition, a type III arch has a distance greater than twice the diameter of the left common carotid artery between the highest point of the arch and the origin of the IA. According to MacDonald's definition, a type III arch presents with a distance of ≥2 cm between the highest point of the arch and the origin of the IA. Our aim was to assess the level of concordance between these different methods.
METHODS: Anonymized thoracic computed tomography scans of 100 healthy patients were reviewed. Two of us independently stratified the selected cases as a type I to III arch, according to the three considered definitions. The interobserver level of concordance for each type III arch classification and level of concordance among the three definitions were assessed.
RESULTS: The 100 selected patients (64% male) were 76 ± 7 years old. For each definition, the interobserver repeatability was almost perfect for all three (Madhwal, κ = 0.81; 95% confidence interval [CI], 0.71-0.99; MacDonald, κ = 0.82; 95% CI, 0.72-0.92; Casserly, κ = 0.84; 95% CI, 0.74-0.93). The level of concordance among the different definitions was very low (Madhwal vs MacDonald, 85% [P = .002]; 33% for type III arch; Madhwal vs Casserly, 60% [P < .0001]; 12% for type III arch; MacDonald vs Casserly, 75% [P < .0001]; 12% for type III arch).
CONCLUSIONS: The three definitions of the type III arch have a very low level of concordance, which might account for the varying clinical relevance of this configuration. Our findings have relevant implications for risk prediction for carotid artery stenting based on the presence of a type III arch, for comparisons of the results from different studies, and for comparisons of different datasets from multicenter trials.