Automated CT Perfusion for Ischemic Core Volume Prediction in Tandem Anterior Circulation Occlusions

Abstract

Background/Aim: CT perfusion (CTP) predicts ischemic core volumes in acute ischemic stroke (AIS); however, assumptions made within the pharmacokinetic model may engender errors by the presence of tracer delay or dispersion. We aimed to evaluate the impact of hemodynamic disturbance due to extracranial anterior circulation occlusions upon the accuracy of ischemic core volume estimation with an automated perfusion analysis tool (RAPID) among AIS patients with large-vessel occlusions. Methods: A prospectively collected, interventional database was retrospectively reviewed for all cases of endovascular treatment of AIS between September 2010 and March 2015 for patients with anterior circulation occlusions with baseline CTP and full reperfusion (mTICI3). Results: Out of 685 treated patients, 114 fit the inclusion criteria. Comparison between tandem (n = 21) and nontandem groups (n = 93) revealed similar baseline ischemic core (20 ± 19 vs. 19 ± 25 cm³; p = 0.8), T_max >6 s (175 ± 109 vs. 162 ± 118 cm³; p = 0.6), T_max >10 s (90 ± 84 vs. 90 ± 91 cm³; p = 0.9), and final infarct volumes (45 ± 47 vs. 37 ± 45 cm³; p = 0.5). Baseline core volumes were found to correlate with final infarct volumes for the tandem (r = 0.49; p = 0.02) and nontandem (r = 0.44; p < 0.01) groups. The mean absolute difference between estimated core and final infarct volume was similar between patients with and those without (24 ± 41 vs. 17 ± 41 cm³; p = 0.5) tandem lesions. Conclusions: The prediction of baseline ischemic core volumes through an optimized CTP analysis employing rigorous normalization, thresholding, and voxel-wise analysis is not significantly influenced by the presence of underlying extracranial carotid steno-occlusive disease in large-vessel AIS.

CT Perfusion as a Predictor of the Final Infarct Volume in Patients with Tandem Occlusion

“…Such parameters grossly rely upon the time–density curves generated by the scanning contrast bolus passage in different arterial vessels [ 45 , 46 , 47 ]. AIF location is crucial to allow the correct comparison between perfusion parameters of the two hemispheres [ 30 , 31 ] in the RAPID software which utilizes an automated AIF algorithm optimized for detection of candidate voxels and delay; it is corrected but uninformed about site occlusion [ 18 , 24 ].…”

“…Tandem occlusions (TOs), consisting of a simultaneous extracranial internal carotid artery (ICA) and intracranial LVO, represent about 15% of all LVO strokes [ 19 , 20 , 21 ]. There are relatively few patients with TOs included in the major randomized controlled trials of EVT and a lack of robust data prevents the delineation of definitive treatment recommendations [ 22 , 23 , 24 , 25 ].…”

“…In clinical settings, both Haussen et al and Albers et al [ 9 , 24 ] found CTP reliable enough in predicting ischemic core volume in ipsilateral extracranial steno-occlusive disease, when appropriate delay correction was applied to CTP analysis, in spite of the coexistence of delay and dispersion effects due to an extracranial occlusion.…”

“…However, we noticed that in both studies the Bland–Altmann plot, employed for evaluating the degree of agreement between predicted infarct core (PIC) and final infarct volume (FIV), showed a positional bias. Namely, PIC and FIV mean differences directly correlated with PIC and FIV means, as if the accuracy in predicting FIV was dependent on final ischemic core volumes [ 9 , 24 , 32 , 33 ].…”

Automated CT Perfusion Prediction of Large Vessel Acute Stroke from Intracranial Atherosclerotic Disease

“…Patients with a technically adequate CTP and an intracranial ICA, MCA-M1 or M2 occlusion were included. Tandem extracranial carotid occlusions and intracranial occlusions were excluded to minimize the delay and dispersion effects of extracranial steno-occlusive lesions upon dynamic bolus passage [2]. …”