Purpose Multiple scores have been described for the assessment of collateralization in acute ischemic stroke. Currently, there is no gold standard for collateral assessment by CT angiography (CTA). This study compared four frequently used collateral scores with regard to their correlation with early infarct core and mismatch ratio.
Methods 30 consecutive patients with acute occlusion of the M1 segment or terminal carotid artery were reviewed retrospectively. Collaterals were assessed using dynamic and also single-phase CTA according to grading systems by the American Society of Interventional and Therapeutic Neuroradiology/Society of Interventional Radiology (ASITN/SIR), Alberta Stroke Program Early CT Score (ASPECTS) (on collaterals), Christoforidis et al and Miteff et al. The Christoforidis and ASITN/SIR scores, which were initially designed for conventional angiography, were adapted to be applicable to CTA. The scores were compared with respect to early infarct core and mismatch ratio in perfusion CT estimated by RAPID software using Spearman correlation.
Results ASITN/SIR and ASPECTS collateral scores showed good correlation with early infarct core (rho=−0.696, p<0.001 and rho=−0.677, p<0.001) and mismatch ratio (rho=0.609, p<0.001 and rho=0.581, p<0.001). In contrast, the Christoforidis and Miteff scores correlated less well with infarct core (rho=0.245, p=0.191 and rho=−0.272, p=0.145, respectively) and mismatch ratio (rho=−0.329, p=0.075 and rho=0.279, p=0.135, respectively). ASPECTS and ASITN/SIR showed excellent cross-correlation (rho=0.901, p<0.001).
Conclusions Compared with the Christoforidis and Miteff scores, the modified ASITN/SIR and ASPECTS collateral scores showed consistently higher correlation with the extent of early infarct core and mismatch volume. This is probably because these scores evaluate the extent and delay of vascular enhancement in the affected territory rather than the backflow of contrast medium to the occlusion.
- CT Angiography
- CT perfusion
Statistics from Altmetric.com
Recent trials such as REVASCAT,1 SWIFT-PRIME,2 MR CLEAN,3 EXTEND-IA,4 and ESCAPE5 showed improved outcome after mechanical recanalization in proximal artery occlusion compared with intravenous thrombolysis alone.6 ,7 To achieve high rates of good outcome, advanced imaging selection of patients for mechanical recanalization is crucial.8–10 In the last few years, many studies have emphasized the relevance of collateral assessment on preinterventional imaging.11–18 However, different collateral grading systems based on different diagnostic modalities such as CT, MRI, and conventional angiography are reported in the literature.19–21 The diagnostic and predictive accuracy with respect to postinterventional outcome and interobserver and intraobserver reliability are still insufficiently investigated.
Most commonly, the collateral score of the Society of NeuroInterventional Surgery (formerly the American Society of Interventional and Therapeutic Neuroradiology)/Society of Interventional Radiology (ASITN/SIR) which is based on conventional angiography,22 the Alberta Stroke Program Early CT Score (ASPECTS) on collaterals (not to be confused with ASPECTS on ischemic parenchymal changes in non-enhanced CT),23 and the scores of Christoforidis et al24 ,25 and Miteff et al26 are generally used for collateral assessment. To our knowledge, there is no direct comparison of these collateral scores. Since multimodal CT is gaining increasing importance in acute stroke imaging,8 ,20 this study aimed to compare the abovementioned collateral scores applied to single or multiphase CT angiography (CTA) with regard to their correlation with early infarct core and mismatch ratio as evaluated by CT perfusion imaging.
Between June 2014 and January 2015 all patients with occlusion of the M1 segment or terminal carotid artery diagnosed in multimodal CT imaging (non-enhanced CT, volume perfusion CT, and single-phase CTA) were retrospectively selected. The study was approved by our institutional review board.
Imaging and post-processing
Imaging was performed using a 64-multislice CT (SOMATOM Definition AS, Siemens, Erlangen, Germany). Scanning order was (1) non-enhanced CT, (2) perfusion CT, and (3) CTA. The detector enables 8 cm z-axis coverage for CT perfusion. Acquisition parameters for CT perfusion were 180 kV and 80 mAs. Total perfusion CT acquisition duration was 60 s. A single contrast bolus of 36 mL Xenetix 350 (Guerbet, Sulzbach, Germany) was applied at a flow rate of 6.0 mL/s, followed by a saline flush of 20 mL and a flow rate of 6.0 mL/s. CT perfusion was routinely reconstructed with a slice width of 5 mm every 3 mm. Acquisition parameters for single-phase CTA were 120 kV and 20 mAs. A single contrast bolus of 65 mL Xenetix 350 was given with a flow rate of 4.0 mL/s, followed by a saline flush of 20 mL and a flow rate of 4.0 mL/s. CTA was routinely reconstructed in axial, sagittal, and coronal planes with a slice thickness of 1 mm every 1 mm in maximum intensity projection.
Dynamic angiographies were manually post-processed using Siemens Dynamic Angio as previously published by Frölich et al.27 ,28 RAPID software was used for automated estimation of infarct core and mismatch ratio.29–31
The reviewer was blinded to the results of infarct core and mismatch ratio, which were determined by RAPID software. Collaterals were assessed primarily in dynamic CTA and additionally in single-phase CTA where 8 cm z-axis coverage for CT perfusion did not entirely cover the basal cerebral arteries. Conventional angiography was not used for collateral assessment.
Four different collateral grading systems were applied: the ASPECTS on collaterals (table 1),23 which is designed for dynamic CTA, and the Miteff26 score for CTA in general (table 2). The collateral scores by ASITN/SIR22 and Christoforidis et al25 (table 3) are intended for use in conventional angiography. The Christoforidis score was therefore adapted to be applicable to single-phase CTA and temporal maximum intensity projection of dynamic CTA, just as is suggested for the use in conventional angiography. The ASITN/SIR score, which is also intended for conventional angiography and exploits information on temporal hemodynamics, was adapted to be applicable to dynamic CTA according to the criteria shown in table 4.
Statistical analysis was performed using R software environment. Spearman correlation coefficients were used to compare infarct core and mismatch ratio with collateral grades. A value of p<0.01 was considered statistically significant.
A total of 30 patients (mean age of 73±12 years, 43% women) with acute M1 segment or terminal carotid artery occlusion on CT imaging were included. The median National Institutes of Health Stroke Scale score on presentation was 17 (range 8–35) and the median premorbid Rankin Scale score was 1 (range 0–5). The mean time between symptom onset and imaging was 144 min (range 40–452 min). Eight patients received intravenous thrombolysis only, mechanical thrombectomy was performed in 17 patients, and 5 patients were not eligible for either option.
The mean infarct core was 27 mL (range 2–146 mL) and the mean mismatch ratio was 29 (range 3–92). ASPECTS and ASITN/SIR collateral scores showed a good correlation with both infarct core (rho=−0.677, p<0.001 and rho=−0.696, p<0.001, respectively) and mismatch ratio (rho=0.581, p<0.001 and rho=0.609, p<0.001, respectively). Also, ASPECTS and ASITN/SIR scores had an excellent cross-correlation (rho=0.901, p<0.001). Collateral scores according to the Christoforidis and Miteff scores showed a poor correlation with infarct core (rho=0.245, p=0.191 and rho=−0.272, p=0.145) and mismatch ratio (rho=−0.329, p=0.075 and rho=0.279, p=0.135). However, they showed good cross-correlation (rho=−0.623, p<0.001) (tables 5 and 6).
There is some evidence that collateralization has an impact on clinical outcome after thrombectomy in acute ischemic stroke.17 ,32 ,33 However, collateral assessment is inconsistent in the literature,19–21 which makes it difficult, for instance, to compare patient populations of different trials on thrombectomy. The inconsistency might also impede the integration of collateral assessment in radiology reporting. We therefore compared several frequently used collateral scores with regard to their cross-correlation and their correlation with infarct core and mismatch ratio in patients with M1 or terminal carotid artery occlusion.
Our retrospective study showed that ASITN/SIR22 and ASPECTS23 collateral scores (not to be confused with ASPECTS on early ischemic parenchymal changes in non-enhanced CT) showed good correlation with infarct core and mismatch ratio, unlike the Christoforidis25 and Miteff26 scores (table 5). This might be because ASPECTS and ASITN/SIR scores evaluate the vascular enhancement of the affected territory rather than focusing on the retrograde flow of contrast medium. A filiform backflow of contrast medium to the thrombus is enough to apply the highest grade according to the Christoforidis and Miteff system. However, this backflow is probably not enough to maintain sufficient blood flow to the affected territory. ASITN/SIR and ASPECTS collateral scores consider both the extent and the delay of collateralization, which are highly relevant pathophysiological factors not considered by the Christoforidis and Miteff scores. We also found that ASITN/SIR and ASPECTS scores have an excellent cross-correlation, which indicates that these scores are comparable.
Our study has several limitations. The ASITN/SIR and Christoforidis scores were initially published for collateral assessment on conventional angiography; however, we found them technically applicable for CTA. The modified ASITN/SIR score appears to be particularly suitable for dynamic CTA because, unlike single-phase CTA, dynamic CTA visualizes the time-dependent collateralization of the ischemic territory. Owing to the relatively small number of patients (n=30), we decided to consider p values <0.01 as statistically significant. Moreover, we did not compare our data with clinical outcomes because of the relatively small number of patients. The hypothesis that ASITN/SIR and ASPECTS collateral scores are more eligible for the estimation of collaterals in acute stroke should therefore be tested in future larger and prospective studies.
Compared with the Christoforidis and Miteff scores, the ASITN/SIR and ASPECTS collateral scores are more appropriate for the assessment of collaterals in acute anterior circulation stroke by CTA. This is probably because the ASITN/SIR and ASPECTS collateral scores evaluate the extent and delay of vascular enhancement in the affected territory rather than the backflow of contrast medium to the occlusion.
The abstract of this manuscript has been published in an abstract book.
Contributors FS conceived the study. FS, MM, MB, and MP initiated the study design. AP helped to conduct the statistical analysis. All authors contributed to the refinement of the study protocol and reviewed and approved the final manuscript.
Competing interests None declared.
Ethics approval Ethics approval was obtained from the Institutional Review Board of Heidelberg University.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.