AI-derived Carotid Elongation Ratio may predict procedural delay but offer limited prognostic utility in mechanical thrombectomy.
Authors
Affiliations (2)
Affiliations (2)
- From the Univ Brest (J.O.), LATIM, INSERM UMR1101, CHU Brest, Brest, France; Department of Radiology (E.A.B., G.E.H., R.K., D.F.K.), Neurologic Surgery (R.K., W.B.), Mayo Clinic Rochester, Minnesota, USA; Stroke Unit, Neurology (P.C., J.L., M.R.-G., J.M., M.R.), Hospital Vall d'Hebron, Barcelona, Spain; Department of Radiology (P.C.), Stanford University, Stanford, CA, USA; Departament de Medicina (M.R.), Universitat Autònoma de Barcelona, Barcelona, Spain. [email protected].
- From the Univ Brest (J.O.), LATIM, INSERM UMR1101, CHU Brest, Brest, France; Department of Radiology (E.A.B., G.E.H., R.K., D.F.K.), Neurologic Surgery (R.K., W.B.), Mayo Clinic Rochester, Minnesota, USA; Stroke Unit, Neurology (P.C., J.L., M.R.-G., J.M., M.R.), Hospital Vall d'Hebron, Barcelona, Spain; Department of Radiology (P.C.), Stanford University, Stanford, CA, USA; Departament de Medicina (M.R.), Universitat Autònoma de Barcelona, Barcelona, Spain.
Abstract
Carotid artery tortuosity, quantified by the carotid elongation ratio (CER), may hinder mechanical thrombectomy (MT). We evaluated whether CER independently predicts procedural efficiency and 90-day functional outcome. We retrospectively analyzed data from 412 patients from Vall d'Hebron University Hospital prospectively maintained registry, who underwent MT for anterior circulation acute ischemic stroke between 2017 and 2023. CER was computed from pre-treatment CT angiography using an AI-based centerline algorithm. Associations with procedural time (groin-to-reperfusion) and 90-day modified Rankin Scale (mRS) were assessed via correlation, quartile comparison, and multivariable logistic regression adjusted for confounders. Mediation analysis tested whether CER mediates the effect of age on procedural time. Mean CER was 1.33 ± 0.15. CER correlated with longer procedure time (ρ = 0.11, p = 0.049), but not with final reperfusion success or number of passes. In univariate analysis, higher CER was associated with worse 90-day outcome (mRS 0-2 vs. 3-6: 1.30 vs. 1.35, p = 0.001), and quartile analysis showed a significant trend (p = 0.004). However, CER was not independently predictive after adjustment (adjusted OR per 0.1 CER = 0.97, 95% CI 0.80-1.18, p = 0.77). Mediation analysis showed CER significantly mediated the relationship between age and procedural time (indirect effect = +0.13 min/year, Sobel p = 0.047). While CER is associated with procedural delay and outcome in unadjusted analyses, it is not an independent predictor of 90-day mRS. Its clinical relevance lies in anticipating technical difficulty, especially in older patients.