Abstract 87: Predictive Scoring System for Pre-Thrombectomy Diagnosis of Intracranial Atherosclerotic Disease LVOS

Abstract

Introduction: Early identification of intracranial atherosclerotic disease (ICAD) underlying large vessel occlusion strokes (LVOS) may allow for mechanical thrombectomy (MT) optimization. We sought to create a pre-procedural scoring system to predict ICAD. Methods: Retrospective analysis of prospectively collected MT databases from 2 comprehensive stroke centers (derivation and validation) including patients with anterior circulation LVOS. ICAD cases were matched for age / sex (1:1) to non-ICAD controls and stepwise logistic regression utilized for various clinical and radiological (NCCT, CTA and CTP) variables. Calibration slope / intercept as well as the area under curve (auROC) were assessed. Results: Of 2870 MTs within the study period, 174 anterior circulation ICAD (6.6%) were matched with 174 controls (n=348) in the derivation cohort. Multivariable analysis β coefficients lead to a 20 point-scale: absence of AF (5), vascular risk factor burden (1 for each: hypertension, diabetes, smoking, hyperlipidemia), CTA multifocal ICAD (3), absence of cortical infarcts (3), presence of borderzone infarct (3), calcium at the siphon (2). The validation cohort comprised 75 ICAD patients (6.9% of 1359 MTs) and 75 controls. AuROC for the derivation cohort was 0.88(0.84-0.91) while for the validation cohort 0.82(0.73-0.89). Calibration slope and intercept showed a good fit for the development cohort although with overestimated risk for the validation cohort. After intercept adjustment, the overestimation was corrected (intercept 0; 95%CI -0.5-0.5 / slope 0.8 95%CI 0.5-1.1). In the full cohort (n=498), ≥ 11 points showed the best performance for distinguishing ICAD from non-ICAD patients, with a sensitivity 0.71 (0.65-0.78) and specificity 0.82 (0.77-0.87), a positive likelihood ratio of 3.92 (2.92, 5.28) and a negative LR of 0.35 (0.28, 0.44). Scores ≥ 12 showed 90% specificity, although sensitivity of 63% (55%-69%). Conclusion: We developed a predictive scoring system for pre-procedural diagnosis of ICAD LVOS with satisfactory discrimination and calibration based on clinical and non-invasive radiological data.