External validation of six prediction scores for differentiating intracranial atherosclerotic disease-related large vessel occlusion from embolic large vessel occlusion.

Purpose- Intracranial atherosclerotic disease-related large vessel occlusion (ICAD-LVO) presents distinct challenges, particularly regarding the high risk of reocclusion and the need for specific management strategies. While several prediction scores exist to differentiate ICAD-LVO from embolic LVO (EMB-LVO), their external validity remains unproven. We aimed to externally validate six established prediction scores for differentiating the two. Methods- We analyzed data from a prospectively maintained, two-center stroke registry (June 2021-March 2025). Consecutive patients who underwent mechanical thrombectomy and had complete clinical and imaging data necessary for calculating six scores (ISAT, REMIT, ABC2D, ATHE, ICAS-LVO, and Score-ICAD) were included. LVO etiology was defined based on angiographic findings during endovascular treatment. The discriminative performance of each score was assessed using the area under the receiver operating characteristic curve (AUC). Results- Of 1,288 screened admissions, 91 patients met the inclusion criteria (ICAD-LVO, n = 18; embolic occlusion, n = 73). The AUCs (95% confidence interval) for differentiating etiology were: ISAT, 0.870 (0.664-1.000; P = 0.064); REMIT, 0.793 (0.676-0.911; P < 0.001); Score-ICAD, 0.707 (0.582-0.833; P = 0.013); ABC2D, 0.627 (0.504-0.751; P = 0.095); ATHE, 0.542 (0.396-0.687; P = 0.617); and ICAS-LVO, 0.465 (0.301-0.630; P = 0.650).

Conclusion

In this external validation cohort, the REMIT scale showed statistically significant discrimination in the full LVO population, although estimates for several scales were imprecise because of the limited number of ICAD-LVO events and subgroup restrictions. Imaging markers, particularly non-culprit stenosis, appeared to provide clinically useful information for identifying ICAD-LVO.