CTP-defined collaterals is a better predictor of intracranial atherosclerotic stenosis-related large-vessel occlusion than multiphase CTA-defined collaterals.

The optimal neuroimaging modalities for differentiating intracranial atherosclerotic stenosis-related large vessel occlusion (ICAS-LVO) from embolism related LVO remain uncertain. This study aimed to address this question by directly comparing collateral circulation using either baseline CT perfusion (CTP) or multiphase CT angiogram (mCTA) to define collaterals. We retrospectively analyzed consecutive patients with acute large vessel occlusion from October 2021 to December 2023. All patients underwent CTP before endovascular therapy, and mCTA was reconstructed from CTP data. In-situ ICAS-LVO was confirmed by a neuro-interventional radiologist. Favorable collateral circulation was defined as a collateral index <0.4 on CTP or a collateral score ≥3 on mCTA. Of 377 patients, 72 (19%) had ICAS-LVO. Patients with only a collateral score ≥3 did not show significantly higher odds of ICAS-LVO (P = 0.681). In contrast, those with a collateral index <0.4 but not favorable mCTA collateral had higher odds of ICAS-LVO (OR2.69, 95%CI [1.07-7.01], P = 0.037). Subgroup analysis showed that a collateral grading scale ≥3 may not predict ICAS-LVO within 6 hours, whereas CTP's predictive performance remained consistently strong in both early and late windows. CTP defined favorable collaterals of collateral index <0.4 demonstrate greater predictive value for ICAS-LVO compared to mCTA, especially within an early time window.