Journal of Cerebral Blood Flow & Metabolism (2014), 1–7 © 2014 ISCBFM All rights reserved 0271-678X/14 $32.00 www.jcbfm.com
ORIGINAL ARTICLE
Predictive value of the velocity of collateral filling in patients with acute ischemic stroke Sebastian E Beyer1,6, Louisa von Baumgarten2,6, Kolja M Thierfelder1, Marietta Rottenkolber3, Hendrik Janssen4, Martin Dichgans5, Thorsten RC Johnson1, Andreas Straube2, Birgit Ertl-Wagner1, Maximilian F Reiser1 and Wieland H Sommer1 The velocity of collateral filling can be assessed in dynamic time-resolved computed tomography (CT) angiographies and may predict initial CT perfusion (CTP) and follow-up lesion size. We included all patients with an M1 ± internal carotid artery (ICA) occlusion and follow-up imaging from an existing cohort of 1791 consecutive patients who underwent multimodal CT for suspected stroke. The velocity of collateral filling was quantified using the delay of time-to-peak (TTP) enhancement of the M2 segment distal to the occlusion. Cerebral blood volume (CBV) and mean transit time (MTT)-CBV mismatch were assessed in initial CTP. Follow-up lesion size was assessed by magnetic resonance imaging (MRI) or non-enhanced CT (NECT). Multivariate analyses were performed to adjust for extent of collateralization and type of treatment. Our study comprised 116 patients. Multivariate analysis showed a short collateral blood flow delay to be an independent predictor of a small CBV lesion (P o 0.001) and a large relative mismatch (P o 0.001) on initial CTP, of a small follow-up lesion (P o 0.001), and of a small difference between initial CBV and follow-up lesion size (P = 0.024). Other independent predictors of a small lesion on follow-up were a high morphologic collateral grade (P = 0.001), lack of an additional ICA occlusion (P = 0.009), and intravenous thrombolysis (P = 0.022). Fast filling of collaterals predicts initial CTP and follow-up lesion size and is independent of extent of collateralization. Journal of Cerebral Blood Flow & Metabolism advance online publication, 5 November 2014; doi:10.1038/jcbfm.2014.182 Keywords: brain imaging; brain ischemia; cerebral blood flow measurement; imaging; neuroradiobrain ischemia
INTRODUCTION In acute ischemic stroke, leptomeningeal collateral vessels provide blood flow to the vascular territory of the occluded artery.1,2 They are considered to help sustain brain viability and reduce ischemic core size.3 Recent studies showed that robust collateral vessels on baseline study to be an independent predictor of a small followup lesion4 and good clinical outcome5 in patients with acute ischemic stroke. Conventional angiography is the acknowledged reference standard for assessing leptomeningeal collaterals,6 but requires an invasive procedure and is not routinely performed during initial workup. Noninvasive assessment of collateralization can be performed using either magnetic resonance angiography3,7 or computed tomography (CT) angiography (CTA).4,6,8–10 The latter offers the advantages of short scan times and wide availability. In conventional CTA, however, the assessment of collaterals is heavily dependent on the timing of image acquisition. Particularly, delayed filling of collaterals distal to an arterial occlusion may impair their assessment on conventional CTA.11 To overcome this limitation, multiple acquisition time points can improve assessment by capturing optimal enhancement of collateral vessels and quantifying the delay of enhancement. Technical developments now allow dynamic CTAs to be reconstructed from whole-brain CT perfusion (WB-CTP) raw data
sets. Smit et al.12 have shown the value of this technique for assessing collateral vessel enhancement showed in temporal maximum intensity projections (tMIPs). Using this approach, a comprehensive assessment of the watershed of collaterals over the entire scan time has been shown to predict clinical outcome.13–15 However, little is yet known about the value of analyzing the temporal aspects of the filling of leptomeningeal collateral vessels. Benefits of a dynamic analysis include identification of the origin of the dominant collaterals11 and quantification of the delay of maximum enhancement.16 Calculating the delay for a standardized distance as a marker of filling velocity may indicate functionality of the collateral vessels and may therefore be an important predictor of outcome. The aim of this study was to test the association of filling velocity of collateral vessels during dynamic CTA with initial CT perfusion, follow-up infarct size and the difference between CTP and follow-up. MATERIALS AND METHODS Study Design and Study Population The institutional review board of the Ludwig-Maximilians-University Hospital, Munich approved the study and waived requirement for
1 Institute for Clinical Radiology, Ludwig-Maximilians-University Hospital Munich, Munich, Germany; 2Department of Neurology, Ludwig-Maximilians-University Hospital Munich, Munich, Germany; 3Department of Medical Informatics, Biometry and Epidemiology, Ludwig-Maximilians-University Munich, Munich, Germany; 4Department of Neuroradiology, LudwigMaximilians-University Hospital Munich, Munich, Germany and 5Institute for Stroke and Dementia Research, Klinikum der Universität München, Ludwig-Maximilians-Universität, Munich, Germany. Correspondence: Dr WH Sommer, Institute for Clinical Radiology, Ludwig-Maximilians-University Hospital Munich, Marchioninistrasse 15, Munich 81377, Germany. E-mail: [email protected] The authors thank Allan Paris, DO, retired (Battle Creek, Mich), for his contribution and editing of the manuscript. 6 These authors contributed equally to this work. Received 9 April 2014; revised 27 September 2014; accepted 29 September 2014
Predictive value of dynCTA time delay in stroke SE Beyer et al
2
Figure 1. Collateral blood flow delay. (A) Temporal maximum intensity projection (tMIP) (30 mm) shows an M1 occlusion (white arrow). Measurements were performed in the M2 segment distal to the occlusion (red box) and in the M2 segment of the contralateral unaffected hemisphere (blue box). Crosses indicate positions of regions of interest (ROIs). (B) The averaged time attenuation curves (TACs) of the stroke hemisphere (red) and the unaffected contralateral hemisphere (blue) are shown.
informed consent. Analysis of patient data was performed in accordance with the Declaration of Helsinki. Our initial cohort consisted of 1791 consecutive patients who underwent WB-CTP for suspected acute ischemic stroke at our institution between April 2009 and January 2014. We included all patients with complete occlusion of a middle cerebral artery (MCA) in the M1 segment, with or without an occlusion of the internal carotid artery (ICA), on conventional and dynamic CTA. We excluded patients with (1) missing WB-CTP raw data sets, (2) missing follow-up non-enhanced CT (NECT) or magnetic resonance imaging (MRI) acquired at least 1 day after initial CTP imaging,17 (3) nondiagnostic quality of conventional CTA or dynamic CTA, (4) insufficient scanning time of dynamic CTA to analyze the delay of collateral blood flow, and (5) lack of collateral reconstitution of the M2 segment distal to the occlusion. Data on time of symptom onset, treatment with intravenous (IV) thrombolysis, and mechanical thrombectomy were collected by review of the medical chart.
Computed Tomography Examination Protocol The multimodal CT protocol consisted of an NECT to exclude intracranial hemorrhage, a supraaortic CTA, and a WB-CTP, all performed using one of the following CT scanners: Somatom Definition AS+, a 128 slice CT scanner; Somatom Definition Flash, a 128 slice dual source CT scanner; and Somatom Definition Edge, a 128 slice CT scanner (all by Siemens Healthcare, Erlangen, Germany). Images of WB-CTP were obtained with 0.6 mm collimation and 100 mm scan coverage in the z axis using a toggling table technique. One scan was acquired every 1.5 seconds. Tube voltage and current were 80 kV and 200 mAs. CT dose index was 276 mGy. A total of 35 mL of iodinated contrast agent was administered at a flow rate of 5 mL/s followed by a saline flush of 40 mL at 5 mL/s. Thirty-one axial slices were reconstructed per view with a thickness of 10 mm and an increment of 3 mm.
segment as a surrogate marker for the filling velocity because it represents the terminus of the collateral circulation in M1 occlusions, resulting in a maximally delayed enhancement, and because it can be reliably identified. Using an axial tMIP of the entire scan time, regions of interest (ROIs) were placed proximally and distally in all visible M2 trunks, yielding four ROIs in the M2 segment if two M2 trunks were visible (Figure 1). The time attenuation curves (TACs) of all ROIs were interpolated using cubic spline interpolation to improve temporal resolution as practiced with CTP TACs.18 The TTP defined the TTP enhancement of the interpolated TAC. Delay was defined as the mean difference of all M2 TTP measurements to the averaged TTP measurements of the contralateral M2 segment. We averaged all M2 TTP measurements to account for differences of collateral reconstitution in different areas of the MCA territory. The M1 segment proximal to the occlusion was assessed for an early temporal branch because this finding has been associated with a better outcome19 and could affect collateral blood flow delay. Origin of dominant collateralization. The origin of the dominant collateralization was identified with a volume rendering technique reconstruction. Collaterals from the anterior circulation (ACA-MCA) and collaterals from the posterior circulation (PCA-MCA) were compared using a combination of the three criteria adopted from Menon et al.:11 (1) anatomic extent, (2) prominence of arteries compared with similar vessels in the opposite MCA territory, and (3) retrograde filling time. The origin was then rated as (1) anterior dominant or (2) posterior dominant. For those cases that could not be differentiated we introduced a third category (3) indeterminate. To evaluate the origin of the dominant collateralization, we also examined the circle of Willis for the anterior communicating artery (ACoA) and the posterior communicating artery (PCoA) using tMIPs of the entire scan time.
Computed tomography perfusion raw data sets were reconstructed as dynamic angiographies using the syngo.CT Dynamic Angio module, syngo via, VA 20 (Siemens Healthcare). The image processing included motion correction and automated bone removal. Dynamic angiographies were then represented as source images, tMIPs, and volume rendering technique reformations. Computation of tMIPs included a 4D noise reduction, previously described.12
Morphologic extent of collaterals. To analyze whether collateral blood flow delay or origin of the dominant collateralization are independent predictors of infarct size, we also assessed the maximal morphologic extent of leptomeningeal collaterals using tMIPs (20 mm slice thickness) of the entire scan time. We analyzed the collateral extent using a previously published score,13,14 which grades collaterals as 0 (the absence of vessels), 1 (collateral vessels visible in less than or equal to 50% of the occluded territory), 2 (collateral vessels visible in more than 50%, but less than 100% of the occluded territory), and 3 (collateral vessels visible in 100% of the occluded territory). Image analysis was performed by two readers: one board-certified radiologist (WHS) with over 10 years of experience in neuroimaging, and one reader (SEB) with 2 years of experience in acute stroke imaging. Quantitative measurements (e.g., drawing of ROIs) were performed by the less experienced reader. The readers were blinded to clinical data, CTA, CT perfusion, and follow-up imaging.
Dynamic Computed Tomography Angiography Image Analysis
Computed Tomography Perfusion Image Processing and Analysis
Velocity of collateral filling. The velocity of collateral filling was quantified using the delay of time-to-peak (TTP) enhancement of the M2 segment distal to the occlusion. We used the delay of enhancement of the M2
We analyzed the cerebral blood volume (CBV) and the mean transit time (MTT) to assess the association of collateralization with CTP as these parameters have been used to make inference about initial infarct size.20
Dynamic Computed Tomography Angiography Image Processing
Journal of Cerebral Blood Flow & Metabolism (2014), 1 – 7
© 2014 ISCBFM
Predictive value of dynCTA time delay in stroke SE Beyer et al
3 The relative mismatch was defined as the percentage of the MTT lesion without the corresponding CBV lesion. The axial CTP images were transferred to a workstation (Syngo MMWP, VA 21A; Siemens Healthcare) and perfusion analysis was performed with the software provided by the vendor, Syngo Volume Perfusion CT Neuro, using a semiautomated deconvolution algorithm (Auto Stroke MTT), previously described,21 the same as used in clinical practice. OsiriX version 4.0 imaging software (http://www.osirix-viewer.com) was used for 3D assessment of CBV and MTT lesion size as previously described.21 We avoided using rigid quantitative perfusion thresholds for the definition of the ischemic core since postprocessing methods vary widely among manufactures,22 and since there are currently no operationally defined and universally accepted thresholds.17 We therefore manually segmented the perfusion deficit on every axial slice using the OsiriX closed polygon tool to create an ROI; thus, the extent of the perfusion deficit could be evaluated by comparison with the contralateral hemisphere. The infarct volume was then calculated using the OsiriX volume calculation tool.
Normal distribution was assessed using the Kolmogorov-Smirnov test. In case of not-normally distributed variables, a square root transformation was performed and the Kolmogorov-Smirnov test was repeated. Univariate linear regression analysis was used to test the association between predictors and outcome variables (MTT lesion size, CBV lesion size, relative mismatch, follow-up lesion size, CBV–follow-up difference). These variables were included as predictors: age, sex, additional ICA occlusion, collateral blood flow delay, origin of dominant collateralization, morphologic collateral extent, visible ACoA, visible PCoA, visible early temporal branch, IV thrombolysis, and mechanical thrombectomy. Multivariate linear regression was used for adjusted analyses. Variables significantly associated with a favorable outcome (Po0.2) in the univariate regression were included in the multivariate models. P values below 0.05 were considered to indicate statistical significance. To provide interpretable results in the multivariate analysis in cases of not-normally distributed variables and square root transformation, analyses were performed for the original variables and the square root transformed variables.23
Follow-Up Image Analysis
RESULTS Study Population Among our cohort of 1791 patients, 1,447 had WB-CTP raw data sets available. Of these patients, 698 had an acute ischemic lesion on CT perfusion, and of those, 147 patients showed a complete M1 occlusion on conventional and dynamic CTA and were included for further analysis. Out of these 147 patients, we excluded 21 due to missing follow-up imaging, 4 due to nondiagnostic quality of dynamic CTA, 2 due to insufficient temporal scan coverage of dynamic CTA, and 4 due to lack of collateral reconstitution of the M2 segment distal to the occlusion. The remaining 116 patients constituted the study cohort (Figure 2).
An MRI with a diffusion-weighted sequence or NECT acquired at least 1 day after the initial CTP was used to assess follow-up lesion size. On MRI, lesions were assessed in the diffusion-weighted sequence with a b value of 1000 s/mm2. Volumetric analysis was performed using OsiriX version 4.0 as described for CTP analysis. To assess the association of collateralization with the discrepancy between initial CTP and follow-up, we calculated the difference between the follow-up lesion size and the CBV lesion size on initial CT perfusion. Cerebral blood volume was chosen as it has been used as a surrogate marker of initial ischemic core.20
Statistical Analysis We performed all statistical analyses using SPSS (Version 21.0; IBM, Armonk, NY, USA). All metric and normally distributed variables are reported as mean ± standard deviation; non-normally distributed variables are presented as median (interquartile range). Categorcial variables are presented as frequency and percentage.
Figure 2.
Baseline Characteristics The mean age was 70 ± 14 years. Fourty-seven (40.5%) patients were male. In the majority of patients, the dominant
Inclusion and exclusion flow chart.
© 2014 ISCBFM
Journal of Cerebral Blood Flow & Metabolism (2014), 1 – 7
Predictive value of dynCTA time delay in stroke SE Beyer et al
4 Table 1.
Patient characteristics All patients (N = 116)
Age—yearsa Male sex—no. (%) Time to CTP imagingb,—minutesc Time to follow-up imaging—daysc Intravenous thrombolysis—no. (%) Mechanical thrombectomy—no. (%) MTT lesion size—mLc CBV lesion size—mLc Relative mismatch—% c Follow-up lesion sized—mLc Additional ICA occlusion—no. (%) Collateral blood flow delay—secondsa
70.2 ± 14.3 47 (40.5) 148 (88–219) 3 (2–5) 79 (68.1) 63 (54.3) 185.3 (148.9–226.6) 36.25(17.6–97.1) 78.4 (55.5–87.0) 71.9 (16.6–217.5) 54 (46.6) 7.96 ± 4.02
Origin of dominant collateralization Anterior cerebral artery—no. (%) Posterior cerebral artery—no. (%) indifferent—no. (%)
15 (12.9) 74 (63.8) 27 (23.3)
Morphologic collateral extent Grade 0—no. (%) Grade 1—no. (%) Grade 2—no. (%) Grade 3—no. (%) Visible ACoA—no. (%) Visible PCoA—no. (%) Visible early temporal branch—no. (%)
0 (0) 27 (23.3) 44 (37.9) 45 (38.8) 71 (61.2) 56 (48.3) 23 (19.8)
ACoA, anterior communicating artery; CBV, cerebral blood volume; CTP, computed tomography perfusion; DWI, diffusion-weighted sequence; ICA, internal carotid artery; IQR, interquartile range; MRI, magnetic resonance imaging; MTT, mean transit time; PCoA, posterior communicating artery. aMean ± s.d. bDocumented for 71 patients. cMedian (IQR). d As measured by DWI-MRI (b1000) or non-enhanced CT.
collateralization originated from a posterior cerebral artery (PCA) (74 subjects, 63.8%). Fifteen patients (12.9%) had dominant collateralization originating from the anterior cerebral artery (ACA). In 27 patients (23.3%), the origin was indeterminate due to evenly balanced anterior and posterior contributions. The mean collateral blood flow delay was 7.96 ± 4.02 seconds. Follow-up was with MRI in 53 patients (45.7%) and with CT in 63 patients (54.3%). Patient characteristics are given in Table 1. Association of Dynamic Computed Tomography Angiography with Initial Computed Tomography Perfusion In the univariate analysis, a short collateral blood flow delay and a high morphologic collateral grade were significantly associated with a small MTT lesion (P o 0.001), a small CBV lesion (P o0.001), and a large relative mismatch (P o0.001). Figure 3A shows the correlation between the collateral blood flow delay and CBV lesion size. Lack of an additional ICA occlusion and a visible early temporal branch were also significantly associated with a small MTT lesion (P = 0.003 and P = 0.006, respectively), a small CBV lesion (P o0.001 and P = 0.020, respectively), and a large relative mismatch (P o0.001 and P = 0.035, respectively). Female sex was significantly associated with a small MTT lesion (P = 0.006). No statistical significance was evident for the origin of the dominant collateralization (Table 2). In the multivariate analysis, a short collateral blood flow delay was significantly associated with a small CBV lesion (P o 0.001) and a large relative mismatch (P o 0.001). A high morphologic collateral grade was significantly associated with a small MTT lesion (P = 0.002), a small CBV lesion (P o 0.001), and a large relative mismatch (P o0.001). Lack of an additional ICA occlusion was significantly associated with a small CBV lesion (P = 0.018) and Journal of Cerebral Blood Flow & Metabolism (2014), 1 – 7
a large relative mismatch (P = 0.046). A visible early temporal branch and female sex were significantly associated with a small MTT lesion (P = 0.033 and P = 0.029, respectively). The findings indicate these predictors operate independently and provide complementary information. The adjusted coefficients of determination were 0.283 for MTT lesion size, 0.751 for CBV lesion size, and 0.717 for relative mismatch. CBV lesion size was not normally distributed and its consistency was verified after square root transformation. Association of Dynamic Computed Tomography Angiography with Follow-Up Lesion Size In the univariate analysis, a short collateral blood flow delay (P o 0.001), a high morphologic collateral grade (P o 0.001), lack of an additional ICA occlusion (P o0.001), a visible early temporal branch (P = 0.034), IV thrombolysis (P = 0.001), and mechanical thrombectomy (P o0.001) were significantly associated with a small follow-up lesion (Table 2). Figure 3B shows the correlation between collateral blood flow delay and follow-up lesion size; examples are presented in Figure 4. In the multivariate analysis, a short collateral blood flow delay (P o 0.001), a high morphologic collateral grade (P = 0.001), lack of an additional ICA occlusion (P = 0.009), IV thrombolysis (P = 0.022), and mechanical thrombectomy (P = 0.048) were significantly associated with a small follow-up lesion. The adjusted coefficient of determination was 0.560. The variable follow-up lesion size was not normally distributed; no more statistical significance was evident for mechanical thrombectomy after square root transformation (P = 0.152). Statistical significance was not affected by square root transformation for the other variables. Association of Dynamic Computed Tomography Angiography with Computed Tomography Perfusion–Follow-Up Difference In the univariate analysis, a short collateral blood flow delay (P o 0.001), a high morphologic collateral grade (P o 0.001), lack of an additional ICA occlusion (Po 0.001), IV thrombolysis (P = 0.001), and mechanical thrombectomy (P o0.001) were significantly associated with a small CBV–follow-up difference. In the multivariate analysis, only a short collateral blood flow delay (P = 0.024), IV thrombolysis (P = 0.044), and mechanical thrombectomy (P = 0.015) were significantly associated with a small CBV–follow-up difference. The adjusted coefficient of determination was 0.287. The CBV–follow-up difference was not normally distributed. Consistency of statistical significance was verified after square root transformation of this variable. DISCUSSION In this exploratory study on collateral circulation in acute ischemic stroke, we showed that dynamic CTA provides useful information about collateral blood flow. We found that the morphologic extent of the collaterals and the collateral blood flow delay were independent predictors of CTP and follow-up lesion size. Our findings are consistent with and extend prior reports showing robust collateralization in dynamic CTA to predict a small follow-up lesion.14 In these studies, the collateralization status was assessed based on the maximal morphologic extent of collateral vessels over the entire scan time. While important, no information about the time point of collateral reconstitution was provided with this method. Other studies showed that dynamic CTA can be used to determine the origin of the collaterals and to measure the time point of enhancement in these vessels;11,16 however, the predictive value of this information had not been assessed. In our study, we © 2014 ISCBFM
Predictive value of dynCTA time delay in stroke SE Beyer et al
5
Figure 3. Correlation between collateral blood flow delay and infarct size. Longer collateral blood flow delay is associated with (A) increased cerebral blood volume (CBV) lesion size in initial computed tomography (CT) perfusion (CTP) and (B) increased follow-up lesion size.
Table 2.
Univariate linear regression for initial CTP and follow-up lesion size MTT lesion size
CBV lesion size
Mismatch %
Follow-up lesion size
β = 0.056; P = 0.899 β = 34.678; P = 0.006 — — β = 37.003: P = 0.003 β = 6.665; P o0.001
β = 0.006; P = 0.987 β = 14.332; P = 0.168 — — β = 50.771; P o0.001 β = 9.878; P o0.001
β = − 0.036; P = 0.805 β = − 2.781; P = 0.509 — — β = − 20.027; Po0.001 β = − 3.976; Po0.001
β = − 0.802; P = 0.301 β = 31.887; P = 0.156 β = − 75.784; P = 0.001 β = − 94.655; Po0.001 β = 113.514; Po 0.001 β = 18.754; Po 0.001
Origin of dominant collateralization Anterior cerebral artery Reference Posterior cerebral artery β = − 16.888; P = 0.335 Indifferent β = − 5.883; P = 0.780
Reference β = − 18.249; P = 0.197 β = − 25.941; P = 0.129
Reference β = 5.841; P = 0.306 β = 10.823; P = 0.116
Reference β = − 20.630; P = 0.503 β = 33.957; P = 0.306
Morphologic collateral extent Grade 0 Grade 1 Grade 2 Grade 3 Visible ACoA Visible PCoA Visible early branch
— Reference β = − 79.287; Po 0.001 β = − 115.463; Po 0.001 β = − 0.094; P = 0.993 β = − 2.245; P = 0.827 β = − 29.606; P = 0.020
— Reference β = 30.363; P o0.001 β = 45.023; P o0.001 β = 0.537; P = 0.899 β = 0.094 P = 0.982 β = 10.844; P = 0.035
— Reference β = − 139.297; Po0.001 β = − 189.260; Po0.001 β = 3.293; P = 0.885 β = − 17.557; P = 0.428 β = − 58.399; P = 0.034
Age Male sex Intravenous thrombolysis Mechanical thrombectomy Additional ICA occlusion Collateral blood flow delay
— Reference β = − 41.559; P = 0.005 β = − 84.172; Po 0.001 β = 5.203; P = 0.688 β = − 0.187; P = 0.988 β = − 42.448; P = 0.006
ACoA, anterior communicating artery; CBV, cerebral blood volume; CTP, computed tomography perfusion; ICA, internal carotid artery; MTT, mean transit time; PCoA, posterior communicating artery; β, coefficient. Bold entries indicate significant results with Po0.05.
show that the collateral blood flow delay is an independent predictor of high relative mismatch on initial CTP and a small follow-up lesion as well as a small difference between initial CBV and follow-up lesion size. Thus, the collateral blood flow delay adds important functional information to an assessment limited to the morphologic extent of collaterals. To our knowledge, this is the first study to quantitatively show a predictive value of the collateral blood flow delay independent of the morphologic collateral grade. Qualitative assessment of hemodynamic and morphologic parameters of leptomeningeal collaterals has been used in catheter angiography grading systems and has been shown to predict a favorable outcome,24–26 showing the importance of enhancement time for collateral quality. Saito et al.27 assessed the correlation of transit time from the ICA to the M2 segment with the lesion size on follow-up NECT. However, only a few patients were included and no statistical tests were performed. Our results extend these observations because we could show that collateral blood flow delay is an independent predictor using the more feasible dynamic CTA. © 2014 ISCBFM
These findings further suggest the use of the ASITN/SIR grading system over approaches that ignore temporal information since it combines qualitative assessments of the morphologic extent and the velocity. Although dynamic CTA and the CTP are reconstructed from the same data set, they display different vessel compartments and therefore merit separate evaluation. In our study, a short collateral blood flow delay and a high morphologic collateral extent were independently associated with a small CBV lesion and a high relative mismatch on initial CT perfusion. These results show that a territory with reduced CBV represents an area receiving limited collateral blood flow thereby providing further evidence that collateralization and penumbral extent are tightly linked.3 We hypothesize that collateral blood flow delay is an indicator of collateral functionality. A short collateral blood flow delay may represent high collateral blood pressure leading to fast filling of collaterals and might increase the effectiveness of perfusion of the cerebral microcirculation represented by CTP parameters. Collateral blood flow delay may therefore potentially serve as an Journal of Cerebral Blood Flow & Metabolism (2014), 1 – 7
Predictive value of dynCTA time delay in stroke SE Beyer et al
6
Figure 4. Examples of the association between collateral blood flow delay and follow-up lesion size. (A) 59 y/o female patient with an M1 +internal carotid artery (ICA) occlusion (white arrow). Average collateral blood flow delay was 2.6 seconds. No IV thrombolysis was administered and no mechanical thrombectomy was attempted. The follow-up lesion size was 26 mL. (B) 75 y/o male patient with an M1+ICA occlusion (white arrow). An early frontal branch supplying the anterior middle cerebral artery (MCA) territory is originating from the anterior cerebral artery (ACA) (white cross). Average collateral blood flow delay was 8.4 seconds. No IV thrombolysis was administered and no mechanical thrombectomy was attempted. The follow-up lesion size was 180 mL. Both patients had collateral vessels visible in 100% of the occluded vessel territory and in both patients a follow-up angiography showed the absence of recanalization.
indicator for the selection of patients who are likely to benefit from therapies that enhance collateralization. At present benefit of therapies such as hemodilution and induced hypertension are controversial.28 With respect to patient outcome, a short collateral blood flow delay and a high morphologic extent of collaterals were also independent predictors of a small follow-up lesion. This emphasizes the importance of evaluating both parameters. The predictive value of the morphologic extent is consistent with previous reports using dynamic CTA and other imaging techniques.4,14,26 Notably, both delay and extent remained independent predictors after correcting for IV thrombolysis and mechanical thrombectomy. This suggests the beneficial effect of good collateralization in dynamic CTA, even in the absence of interventions. Previous reports indicated that the extent of collateralization in dynamic CTA predicts response to IV thrombolysis.14 The collateral blood flow delay may further improve the response to IV thrombolysis since a short delay might allow more effective delivery of recombinant tissue plasminogen activator to the distal parts of the thrombus. Further studies are needed to clarify this point. Together with IV thrombolysis and mechanical thrombectomy, a short collateral blood flow delay was an independent predictor of a small difference between initial CBV and follow-up lesion size. Notably, morphologic collateral extent was not a significant predictor in the multivariate analysis. This provides further insight into which factors determine penumbral loss. Recent studies have showed that good collateralization at baseline is an independent predictor of small penumbral loss23 and that collateral failure is Journal of Cerebral Blood Flow & Metabolism (2014), 1 – 7
associated with infarct growth.3 Our results add to that by showing an important role for collateral functionality as shown by the collateral filling velocity. These findings are in line with previous studies using Tmax to assess delay in CT perfusion. Nagakane et al.29 showed that the risk of subsequent infarction increases if Tmax increases and that this probabilistic approach allowed a better prediction of the final infarction volume than single cutoff values.29 It is likely that in patients with a longer collateral filling time there is an increased proportion of tissue-atrisk with higher Tmax values. Assuming an increased likelihood of infarction for this tissue as suggested by Nagakane et al., this may explain the observed association between collateral filling time and infarct growth. This is further emphasized by a recent study showing that a small proportion of tissue with Tmax values ≥ 10 seconds within tissue with a Tmax value of ≥ 4 seconds is associated with decreased infarct growth.30 Thus, our results may indicate the importance of the collateral filling velocity in the assessment of dynamic angiographies and may further stress probabilistic approaches taking into account the distribution of parameter values in the assessment of CT perfusion. It would, however, be desirable to investigate the influence of the morphologic extent in a larger sample and to assess the total extent of collaterals in a more quantitative way as these may be factors that have led to an underestimation of the impact of the morphologic extent in our sample. A further independent predictor of a small follow-up lesion size was the administration of IV thrombolysis. There was no evidence that the origin of dominant collateralization influenced outcome, © 2014 ISCBFM
Predictive value of dynCTA time delay in stroke SE Beyer et al
7 i.e., the extent of collateralization did not differ among groups with different origins. However, these results have to be interpreted cautiously as there were only 15 patients with an anterior dominant pattern versus 74 patients with a posterior dominant pattern. These data must be interpreted in the context of the study design. As a retrospective study, there is a potential for bias in patient selection. The use of a large cohort of consecutive patients and standardized protocols for patients with suspected stroke, however, helped to minimize this problem. Furthermore, although mechanical thrombectomy and treatment with intravenous recombinant tissue plasminogen activator were included, the state of recanalization was not assessed. However, our data apropos the follow-up infarct volume indicate that the predictive value in the initial dynamic CTA is independent of spontaneous or therapy-induced recanalization. In addition, the use of both CT and MRI as follow-up might result in heterogeneity of volume measurements. Further, measuring regional contrast enhancement CTP parameters would be a more quantitative approach but may be more heavily influenced by factors such as severe patient motion. Finally, measurement of the delay in the M2 segment and averaging of all M2 measurements is a relatively rough assessment and we acknowledge that the association between the averaged M2 delay and the lesion size might be affected by (1) further occlusions distal to the M1 segment, (2) different prominences of M2 trunks, and (3) differences in the filling time of collaterals in different areas of the MCA territory. In conclusion, we showed that assessment of the velocity of collateral filling using dynamic CTA provides an important predictor of initial CTP and follow-up lesion size and adds functional information to assess the morphologic extent of collaterals. DISCLOSURE/CONFLICT OF INTEREST The authors declare no conflict of interest.
REFERENCES 1 Liebeskind DS. Collateral circulation. Stroke 2003; 34: 2279–2284. 2 Tariq N, Khatri R. Leptomeningeal collaterals in acute ischemic stroke. J Vasc Interv Neurol 2008; 1: 91–95. 3 Campbell BC, Christensen S, Tress BM, Churilov L, Desmond PM, Parsons MW et al. Failure of collateral blood flow is associated with infarct growth in ischemic stroke. J Cereb Blood Flow Metab 2013; 33: 1168–1172. 4 Tan JC, Dillon WP, Liu S, Adler F, Smith WS, Wintermark M. Systematic comparison of perfusion-CT and CT-angiography in acute stroke patients. Ann Neurol 2007; 61: 533–543. 5 Miteff F, Levi CR, Bateman GA, Spratt N, McElduff P, Parsons MW. The independent predictive utility of computed tomography angiographic collateral status in acute ischaemic stroke. Brain 2009; 132: 2231–2238. 6 McVerry F, Liebeskind DS, Muir KW. Systematic review of methods for assessing leptomeningeal collateral flow. AJNR Am J Neuroradiol 2012; 33: 576–582. 7 Ito K, Sasaki M, Kobayashi M, Ogasawara K, Nishihara T, Takahashi T et al. Noninvasive evaluation of collateral blood flow through circle of Willis in cervical carotid stenosis using selective magnetic resonance angiography. J Stroke Cerebrovasc Dis 2013; 23: 1019–1023. 8 Menon BK, Smith EE, Modi J, Patel SK, Bhatia R, Watson TW et al. Regional leptomeningeal score on CT angiography predicts clinical and imaging outcomes in patients with acute anterior circulation occlusions. AJNR Am J Neuroradiol 2011; 32: 1640–1645. 9 Maas MB, Lev MH, Ay H, Singhal AB, Greer DM, Smith WS et al. Collateral vessels on CT angiography predict outcome in acute ischemic stroke. Stroke 2009; 40: 3001–3005.
© 2014 ISCBFM
10 Kim JJ, Fischbein NJ, Lu Y, Pham D, Dillon WP. Regional angiographic grading system for collateral flow: correlation with cerebral infarction in patients with middle cerebral artery occlusion. Stroke 2004; 35: 1340–1344. 11 Menon BK, O'Brien B, Bivard A, Spratt NJ, Demchuk AM, Miteff F et al. Assessment of leptomeningeal collaterals using dynamic CT angiography in patients with acute ischemic stroke. J Cereb Blood Flow Metab 2013; 33: 365–371. 12 Smit EJ, Vonken EJ, van der Schaaf IC, Mendrik AM, Dankbaar JW, Horsch AD et al. Timing-invariant reconstruction for deriving high-quality CT angiographic data from cerebral CT perfusion data. Radiology 2012; 263: 216–225. 13 Smit EJ, Vonken EJ, van Seeters T, Dankbaar JW, van der Schaaf IC, Kappelle LJ et al. Timing-invariant imaging of collateral vessels in acute ischemic stroke. Stroke 2013; 44: 2194–2199. 14 Calleja AI, Cortijo E, Garcia-Bermejo P, Gomez RD, Perez-Fernandez S, Del Monte JM et al. Collateral circulation on perfusion-computed tomography-source images predicts the response to stroke intravenous thrombolysis. Eur J Neurol 2013; 20: 795–802. 15 Frolich AM, Wolff SL, Psychogios MN, Klotz E, Schramm R, Wasser K et al. Timeresolved assessment of collateral flow using 4D CT angiography in large-vessel occlusion stroke. Eur Radiol 2014; 24: 390–396. 16 Siebert E, Diekmann S, Masuhr F, Bauknecht HC, Schreiber S, Klingebiel R et al. Measurement of cerebral circulation times using dynamic whole-brain CT-angiography: feasibility and initial experience. Neurol Sci 2012; 33: 741–747. 17 Bivard A, Levi C, Spratt N, Parsons M. Perfusion CT in acute stroke: a comprehensive analysis of infarct and penumbra. Radiology 2013; 267: 543–550. 18 Royalty K, Manhart M, Pulfer K, Deuerling-Zheng Y, Strother C, Fieselmann A et al. C-arm CT measurement of cerebral blood volume and cerebral blood flow using a novel high-speed acquisition and a single intravenous contrast injection. AJNR Am J Neuroradiol 2013; 34: 2131–2138. 19 Menon BK, Bal S, Modi J, Sohn SI, Watson TW, Hill MD et al. Anterior temporal artery sign in CT angiography predicts reduced fatal brain edema and mortality in acute M1 middle cerebral artery occlusions. J Neuroimaging 2012; 22: 145–148. 20 Schaefer PW, Barak ER, Kamalian S, Gharai LR, Schwamm L, Gonzalez RG et al. Quantitative assessment of core/penumbra mismatch in acute stroke: CT and MR perfusion imaging are strongly correlated when sufficient brain volume is imaged. Stroke 2008; 39: 2986–2992. 21 Thierfelder KM, Sommer WH, Baumann AB, Klotz E, Meinel FG, Strobl FF et al. Whole-brain CT perfusion: reliability and reproducibility of volumetric perfusion deficit assessment in patients with acute ischemic stroke. Neuroradiology 2013; 55: 827–835. 22 Kamalian S, Kamalian S, Maas MB, Goldmacher GV, Payabvash S, Akbar A et al. CT cerebral blood flow maps optimally correlate with admission diffusion-weighted imaging in acute stroke but thresholds vary by postprocessing platform. Stroke 2011; 42: 1923–1928. 23 Jung S, Gilgen M, Slotboom J, El-Koussy M, Zubler C, Kiefer C et al. Factors that determine penumbral tissue loss in acute ischaemic stroke. Brain 2013; 136: 3554–3560. 24 Sanossian N, Saver JL, Alger JR, Kim D, Duckwiler GR, Jahan R et al. Angiography reveals that fluid-attenuated inversion recovery vascular hyperintensities are due to slow flow, not thrombus. AJNR Am J Neuroradiol 2009; 30: 564–568. 25 Ovbiagele B, Saver JL, Starkman S, Kim D, Ali LK, Jahan R et al. Statin enhancement of collateralization in acute stroke. Neurology 2007; 68: 2129–2131. 26 Bang OY, Saver JL, Buck BH, Alger JR, Starkman S, Ovbiagele B et al. Impact of collateral flow on tissue fate in acute ischaemic stroke. J Neurol Neurosurg Psychiatry 2008; 79: 625–629. 27 Saito I, Segawa H, Shiokawa Y, Taniguchi M, Tsutsumi K. Middle cerebral artery occlusion: correlation of computed tomography and angiography with clinical outcome. Stroke 1987; 18: 863–868. 28 Shuaib A, Butcher K, Mohammad AA, Saqqur M, Liebeskind DS. Collateral blood vessels in acute ischaemic stroke: a potential therapeutic target. Lancet Neurol 2011; 10: 909–921. 29 Nagakane Y, Christensen S, Ogata T, Churilov L, Ma H, Parsons MW et al. Moving beyond a single perfusion threshold to define penumbra: a novel probabilistic mismatch definition. Stroke 2012; 43: 1548–1555. 30 Olivot JM, Mlynash M, Inoue M, Marks MP, Wheeler HM, Kemp S et al. Hypoperfusion intensity ratio predicts infarct progression and functional outcome in the DEFUSE 2 Cohort. Stroke 2014; 45: 1018–1023.
Journal of Cerebral Blood Flow & Metabolism (2014), 1 – 7
ORIGINAL ARTICLE
Predictive value of the velocity of collateral filling in patients with acute ischemic stroke Sebastian E Beyer1,6, Louisa von Baumgarten2,6, Kolja M Thierfelder1, Marietta Rottenkolber3, Hendrik Janssen4, Martin Dichgans5, Thorsten RC Johnson1, Andreas Straube2, Birgit Ertl-Wagner1, Maximilian F Reiser1 and Wieland H Sommer1 The velocity of collateral filling can be assessed in dynamic time-resolved computed tomography (CT) angiographies and may predict initial CT perfusion (CTP) and follow-up lesion size. We included all patients with an M1 ± internal carotid artery (ICA) occlusion and follow-up imaging from an existing cohort of 1791 consecutive patients who underwent multimodal CT for suspected stroke. The velocity of collateral filling was quantified using the delay of time-to-peak (TTP) enhancement of the M2 segment distal to the occlusion. Cerebral blood volume (CBV) and mean transit time (MTT)-CBV mismatch were assessed in initial CTP. Follow-up lesion size was assessed by magnetic resonance imaging (MRI) or non-enhanced CT (NECT). Multivariate analyses were performed to adjust for extent of collateralization and type of treatment. Our study comprised 116 patients. Multivariate analysis showed a short collateral blood flow delay to be an independent predictor of a small CBV lesion (P o 0.001) and a large relative mismatch (P o 0.001) on initial CTP, of a small follow-up lesion (P o 0.001), and of a small difference between initial CBV and follow-up lesion size (P = 0.024). Other independent predictors of a small lesion on follow-up were a high morphologic collateral grade (P = 0.001), lack of an additional ICA occlusion (P = 0.009), and intravenous thrombolysis (P = 0.022). Fast filling of collaterals predicts initial CTP and follow-up lesion size and is independent of extent of collateralization. Journal of Cerebral Blood Flow & Metabolism advance online publication, 5 November 2014; doi:10.1038/jcbfm.2014.182 Keywords: brain imaging; brain ischemia; cerebral blood flow measurement; imaging; neuroradiobrain ischemia
INTRODUCTION In acute ischemic stroke, leptomeningeal collateral vessels provide blood flow to the vascular territory of the occluded artery.1,2 They are considered to help sustain brain viability and reduce ischemic core size.3 Recent studies showed that robust collateral vessels on baseline study to be an independent predictor of a small followup lesion4 and good clinical outcome5 in patients with acute ischemic stroke. Conventional angiography is the acknowledged reference standard for assessing leptomeningeal collaterals,6 but requires an invasive procedure and is not routinely performed during initial workup. Noninvasive assessment of collateralization can be performed using either magnetic resonance angiography3,7 or computed tomography (CT) angiography (CTA).4,6,8–10 The latter offers the advantages of short scan times and wide availability. In conventional CTA, however, the assessment of collaterals is heavily dependent on the timing of image acquisition. Particularly, delayed filling of collaterals distal to an arterial occlusion may impair their assessment on conventional CTA.11 To overcome this limitation, multiple acquisition time points can improve assessment by capturing optimal enhancement of collateral vessels and quantifying the delay of enhancement. Technical developments now allow dynamic CTAs to be reconstructed from whole-brain CT perfusion (WB-CTP) raw data
sets. Smit et al.12 have shown the value of this technique for assessing collateral vessel enhancement showed in temporal maximum intensity projections (tMIPs). Using this approach, a comprehensive assessment of the watershed of collaterals over the entire scan time has been shown to predict clinical outcome.13–15 However, little is yet known about the value of analyzing the temporal aspects of the filling of leptomeningeal collateral vessels. Benefits of a dynamic analysis include identification of the origin of the dominant collaterals11 and quantification of the delay of maximum enhancement.16 Calculating the delay for a standardized distance as a marker of filling velocity may indicate functionality of the collateral vessels and may therefore be an important predictor of outcome. The aim of this study was to test the association of filling velocity of collateral vessels during dynamic CTA with initial CT perfusion, follow-up infarct size and the difference between CTP and follow-up. MATERIALS AND METHODS Study Design and Study Population The institutional review board of the Ludwig-Maximilians-University Hospital, Munich approved the study and waived requirement for
1 Institute for Clinical Radiology, Ludwig-Maximilians-University Hospital Munich, Munich, Germany; 2Department of Neurology, Ludwig-Maximilians-University Hospital Munich, Munich, Germany; 3Department of Medical Informatics, Biometry and Epidemiology, Ludwig-Maximilians-University Munich, Munich, Germany; 4Department of Neuroradiology, LudwigMaximilians-University Hospital Munich, Munich, Germany and 5Institute for Stroke and Dementia Research, Klinikum der Universität München, Ludwig-Maximilians-Universität, Munich, Germany. Correspondence: Dr WH Sommer, Institute for Clinical Radiology, Ludwig-Maximilians-University Hospital Munich, Marchioninistrasse 15, Munich 81377, Germany. E-mail: [email protected] The authors thank Allan Paris, DO, retired (Battle Creek, Mich), for his contribution and editing of the manuscript. 6 These authors contributed equally to this work. Received 9 April 2014; revised 27 September 2014; accepted 29 September 2014
Predictive value of dynCTA time delay in stroke SE Beyer et al
2
Figure 1. Collateral blood flow delay. (A) Temporal maximum intensity projection (tMIP) (30 mm) shows an M1 occlusion (white arrow). Measurements were performed in the M2 segment distal to the occlusion (red box) and in the M2 segment of the contralateral unaffected hemisphere (blue box). Crosses indicate positions of regions of interest (ROIs). (B) The averaged time attenuation curves (TACs) of the stroke hemisphere (red) and the unaffected contralateral hemisphere (blue) are shown.
informed consent. Analysis of patient data was performed in accordance with the Declaration of Helsinki. Our initial cohort consisted of 1791 consecutive patients who underwent WB-CTP for suspected acute ischemic stroke at our institution between April 2009 and January 2014. We included all patients with complete occlusion of a middle cerebral artery (MCA) in the M1 segment, with or without an occlusion of the internal carotid artery (ICA), on conventional and dynamic CTA. We excluded patients with (1) missing WB-CTP raw data sets, (2) missing follow-up non-enhanced CT (NECT) or magnetic resonance imaging (MRI) acquired at least 1 day after initial CTP imaging,17 (3) nondiagnostic quality of conventional CTA or dynamic CTA, (4) insufficient scanning time of dynamic CTA to analyze the delay of collateral blood flow, and (5) lack of collateral reconstitution of the M2 segment distal to the occlusion. Data on time of symptom onset, treatment with intravenous (IV) thrombolysis, and mechanical thrombectomy were collected by review of the medical chart.
Computed Tomography Examination Protocol The multimodal CT protocol consisted of an NECT to exclude intracranial hemorrhage, a supraaortic CTA, and a WB-CTP, all performed using one of the following CT scanners: Somatom Definition AS+, a 128 slice CT scanner; Somatom Definition Flash, a 128 slice dual source CT scanner; and Somatom Definition Edge, a 128 slice CT scanner (all by Siemens Healthcare, Erlangen, Germany). Images of WB-CTP were obtained with 0.6 mm collimation and 100 mm scan coverage in the z axis using a toggling table technique. One scan was acquired every 1.5 seconds. Tube voltage and current were 80 kV and 200 mAs. CT dose index was 276 mGy. A total of 35 mL of iodinated contrast agent was administered at a flow rate of 5 mL/s followed by a saline flush of 40 mL at 5 mL/s. Thirty-one axial slices were reconstructed per view with a thickness of 10 mm and an increment of 3 mm.
segment as a surrogate marker for the filling velocity because it represents the terminus of the collateral circulation in M1 occlusions, resulting in a maximally delayed enhancement, and because it can be reliably identified. Using an axial tMIP of the entire scan time, regions of interest (ROIs) were placed proximally and distally in all visible M2 trunks, yielding four ROIs in the M2 segment if two M2 trunks were visible (Figure 1). The time attenuation curves (TACs) of all ROIs were interpolated using cubic spline interpolation to improve temporal resolution as practiced with CTP TACs.18 The TTP defined the TTP enhancement of the interpolated TAC. Delay was defined as the mean difference of all M2 TTP measurements to the averaged TTP measurements of the contralateral M2 segment. We averaged all M2 TTP measurements to account for differences of collateral reconstitution in different areas of the MCA territory. The M1 segment proximal to the occlusion was assessed for an early temporal branch because this finding has been associated with a better outcome19 and could affect collateral blood flow delay. Origin of dominant collateralization. The origin of the dominant collateralization was identified with a volume rendering technique reconstruction. Collaterals from the anterior circulation (ACA-MCA) and collaterals from the posterior circulation (PCA-MCA) were compared using a combination of the three criteria adopted from Menon et al.:11 (1) anatomic extent, (2) prominence of arteries compared with similar vessels in the opposite MCA territory, and (3) retrograde filling time. The origin was then rated as (1) anterior dominant or (2) posterior dominant. For those cases that could not be differentiated we introduced a third category (3) indeterminate. To evaluate the origin of the dominant collateralization, we also examined the circle of Willis for the anterior communicating artery (ACoA) and the posterior communicating artery (PCoA) using tMIPs of the entire scan time.
Computed tomography perfusion raw data sets were reconstructed as dynamic angiographies using the syngo.CT Dynamic Angio module, syngo via, VA 20 (Siemens Healthcare). The image processing included motion correction and automated bone removal. Dynamic angiographies were then represented as source images, tMIPs, and volume rendering technique reformations. Computation of tMIPs included a 4D noise reduction, previously described.12
Morphologic extent of collaterals. To analyze whether collateral blood flow delay or origin of the dominant collateralization are independent predictors of infarct size, we also assessed the maximal morphologic extent of leptomeningeal collaterals using tMIPs (20 mm slice thickness) of the entire scan time. We analyzed the collateral extent using a previously published score,13,14 which grades collaterals as 0 (the absence of vessels), 1 (collateral vessels visible in less than or equal to 50% of the occluded territory), 2 (collateral vessels visible in more than 50%, but less than 100% of the occluded territory), and 3 (collateral vessels visible in 100% of the occluded territory). Image analysis was performed by two readers: one board-certified radiologist (WHS) with over 10 years of experience in neuroimaging, and one reader (SEB) with 2 years of experience in acute stroke imaging. Quantitative measurements (e.g., drawing of ROIs) were performed by the less experienced reader. The readers were blinded to clinical data, CTA, CT perfusion, and follow-up imaging.
Dynamic Computed Tomography Angiography Image Analysis
Computed Tomography Perfusion Image Processing and Analysis
Velocity of collateral filling. The velocity of collateral filling was quantified using the delay of time-to-peak (TTP) enhancement of the M2 segment distal to the occlusion. We used the delay of enhancement of the M2
We analyzed the cerebral blood volume (CBV) and the mean transit time (MTT) to assess the association of collateralization with CTP as these parameters have been used to make inference about initial infarct size.20
Dynamic Computed Tomography Angiography Image Processing
Journal of Cerebral Blood Flow & Metabolism (2014), 1 – 7
© 2014 ISCBFM
Predictive value of dynCTA time delay in stroke SE Beyer et al
3 The relative mismatch was defined as the percentage of the MTT lesion without the corresponding CBV lesion. The axial CTP images were transferred to a workstation (Syngo MMWP, VA 21A; Siemens Healthcare) and perfusion analysis was performed with the software provided by the vendor, Syngo Volume Perfusion CT Neuro, using a semiautomated deconvolution algorithm (Auto Stroke MTT), previously described,21 the same as used in clinical practice. OsiriX version 4.0 imaging software (http://www.osirix-viewer.com) was used for 3D assessment of CBV and MTT lesion size as previously described.21 We avoided using rigid quantitative perfusion thresholds for the definition of the ischemic core since postprocessing methods vary widely among manufactures,22 and since there are currently no operationally defined and universally accepted thresholds.17 We therefore manually segmented the perfusion deficit on every axial slice using the OsiriX closed polygon tool to create an ROI; thus, the extent of the perfusion deficit could be evaluated by comparison with the contralateral hemisphere. The infarct volume was then calculated using the OsiriX volume calculation tool.
Normal distribution was assessed using the Kolmogorov-Smirnov test. In case of not-normally distributed variables, a square root transformation was performed and the Kolmogorov-Smirnov test was repeated. Univariate linear regression analysis was used to test the association between predictors and outcome variables (MTT lesion size, CBV lesion size, relative mismatch, follow-up lesion size, CBV–follow-up difference). These variables were included as predictors: age, sex, additional ICA occlusion, collateral blood flow delay, origin of dominant collateralization, morphologic collateral extent, visible ACoA, visible PCoA, visible early temporal branch, IV thrombolysis, and mechanical thrombectomy. Multivariate linear regression was used for adjusted analyses. Variables significantly associated with a favorable outcome (Po0.2) in the univariate regression were included in the multivariate models. P values below 0.05 were considered to indicate statistical significance. To provide interpretable results in the multivariate analysis in cases of not-normally distributed variables and square root transformation, analyses were performed for the original variables and the square root transformed variables.23
Follow-Up Image Analysis
RESULTS Study Population Among our cohort of 1791 patients, 1,447 had WB-CTP raw data sets available. Of these patients, 698 had an acute ischemic lesion on CT perfusion, and of those, 147 patients showed a complete M1 occlusion on conventional and dynamic CTA and were included for further analysis. Out of these 147 patients, we excluded 21 due to missing follow-up imaging, 4 due to nondiagnostic quality of dynamic CTA, 2 due to insufficient temporal scan coverage of dynamic CTA, and 4 due to lack of collateral reconstitution of the M2 segment distal to the occlusion. The remaining 116 patients constituted the study cohort (Figure 2).
An MRI with a diffusion-weighted sequence or NECT acquired at least 1 day after the initial CTP was used to assess follow-up lesion size. On MRI, lesions were assessed in the diffusion-weighted sequence with a b value of 1000 s/mm2. Volumetric analysis was performed using OsiriX version 4.0 as described for CTP analysis. To assess the association of collateralization with the discrepancy between initial CTP and follow-up, we calculated the difference between the follow-up lesion size and the CBV lesion size on initial CT perfusion. Cerebral blood volume was chosen as it has been used as a surrogate marker of initial ischemic core.20
Statistical Analysis We performed all statistical analyses using SPSS (Version 21.0; IBM, Armonk, NY, USA). All metric and normally distributed variables are reported as mean ± standard deviation; non-normally distributed variables are presented as median (interquartile range). Categorcial variables are presented as frequency and percentage.
Figure 2.
Baseline Characteristics The mean age was 70 ± 14 years. Fourty-seven (40.5%) patients were male. In the majority of patients, the dominant
Inclusion and exclusion flow chart.
© 2014 ISCBFM
Journal of Cerebral Blood Flow & Metabolism (2014), 1 – 7
Predictive value of dynCTA time delay in stroke SE Beyer et al
4 Table 1.
Patient characteristics All patients (N = 116)
Age—yearsa Male sex—no. (%) Time to CTP imagingb,—minutesc Time to follow-up imaging—daysc Intravenous thrombolysis—no. (%) Mechanical thrombectomy—no. (%) MTT lesion size—mLc CBV lesion size—mLc Relative mismatch—% c Follow-up lesion sized—mLc Additional ICA occlusion—no. (%) Collateral blood flow delay—secondsa
70.2 ± 14.3 47 (40.5) 148 (88–219) 3 (2–5) 79 (68.1) 63 (54.3) 185.3 (148.9–226.6) 36.25(17.6–97.1) 78.4 (55.5–87.0) 71.9 (16.6–217.5) 54 (46.6) 7.96 ± 4.02
Origin of dominant collateralization Anterior cerebral artery—no. (%) Posterior cerebral artery—no. (%) indifferent—no. (%)
15 (12.9) 74 (63.8) 27 (23.3)
Morphologic collateral extent Grade 0—no. (%) Grade 1—no. (%) Grade 2—no. (%) Grade 3—no. (%) Visible ACoA—no. (%) Visible PCoA—no. (%) Visible early temporal branch—no. (%)
0 (0) 27 (23.3) 44 (37.9) 45 (38.8) 71 (61.2) 56 (48.3) 23 (19.8)
ACoA, anterior communicating artery; CBV, cerebral blood volume; CTP, computed tomography perfusion; DWI, diffusion-weighted sequence; ICA, internal carotid artery; IQR, interquartile range; MRI, magnetic resonance imaging; MTT, mean transit time; PCoA, posterior communicating artery. aMean ± s.d. bDocumented for 71 patients. cMedian (IQR). d As measured by DWI-MRI (b1000) or non-enhanced CT.
collateralization originated from a posterior cerebral artery (PCA) (74 subjects, 63.8%). Fifteen patients (12.9%) had dominant collateralization originating from the anterior cerebral artery (ACA). In 27 patients (23.3%), the origin was indeterminate due to evenly balanced anterior and posterior contributions. The mean collateral blood flow delay was 7.96 ± 4.02 seconds. Follow-up was with MRI in 53 patients (45.7%) and with CT in 63 patients (54.3%). Patient characteristics are given in Table 1. Association of Dynamic Computed Tomography Angiography with Initial Computed Tomography Perfusion In the univariate analysis, a short collateral blood flow delay and a high morphologic collateral grade were significantly associated with a small MTT lesion (P o 0.001), a small CBV lesion (P o0.001), and a large relative mismatch (P o0.001). Figure 3A shows the correlation between the collateral blood flow delay and CBV lesion size. Lack of an additional ICA occlusion and a visible early temporal branch were also significantly associated with a small MTT lesion (P = 0.003 and P = 0.006, respectively), a small CBV lesion (P o0.001 and P = 0.020, respectively), and a large relative mismatch (P o0.001 and P = 0.035, respectively). Female sex was significantly associated with a small MTT lesion (P = 0.006). No statistical significance was evident for the origin of the dominant collateralization (Table 2). In the multivariate analysis, a short collateral blood flow delay was significantly associated with a small CBV lesion (P o 0.001) and a large relative mismatch (P o 0.001). A high morphologic collateral grade was significantly associated with a small MTT lesion (P = 0.002), a small CBV lesion (P o 0.001), and a large relative mismatch (P o0.001). Lack of an additional ICA occlusion was significantly associated with a small CBV lesion (P = 0.018) and Journal of Cerebral Blood Flow & Metabolism (2014), 1 – 7
a large relative mismatch (P = 0.046). A visible early temporal branch and female sex were significantly associated with a small MTT lesion (P = 0.033 and P = 0.029, respectively). The findings indicate these predictors operate independently and provide complementary information. The adjusted coefficients of determination were 0.283 for MTT lesion size, 0.751 for CBV lesion size, and 0.717 for relative mismatch. CBV lesion size was not normally distributed and its consistency was verified after square root transformation. Association of Dynamic Computed Tomography Angiography with Follow-Up Lesion Size In the univariate analysis, a short collateral blood flow delay (P o 0.001), a high morphologic collateral grade (P o 0.001), lack of an additional ICA occlusion (P o0.001), a visible early temporal branch (P = 0.034), IV thrombolysis (P = 0.001), and mechanical thrombectomy (P o0.001) were significantly associated with a small follow-up lesion (Table 2). Figure 3B shows the correlation between collateral blood flow delay and follow-up lesion size; examples are presented in Figure 4. In the multivariate analysis, a short collateral blood flow delay (P o 0.001), a high morphologic collateral grade (P = 0.001), lack of an additional ICA occlusion (P = 0.009), IV thrombolysis (P = 0.022), and mechanical thrombectomy (P = 0.048) were significantly associated with a small follow-up lesion. The adjusted coefficient of determination was 0.560. The variable follow-up lesion size was not normally distributed; no more statistical significance was evident for mechanical thrombectomy after square root transformation (P = 0.152). Statistical significance was not affected by square root transformation for the other variables. Association of Dynamic Computed Tomography Angiography with Computed Tomography Perfusion–Follow-Up Difference In the univariate analysis, a short collateral blood flow delay (P o 0.001), a high morphologic collateral grade (P o 0.001), lack of an additional ICA occlusion (Po 0.001), IV thrombolysis (P = 0.001), and mechanical thrombectomy (P o0.001) were significantly associated with a small CBV–follow-up difference. In the multivariate analysis, only a short collateral blood flow delay (P = 0.024), IV thrombolysis (P = 0.044), and mechanical thrombectomy (P = 0.015) were significantly associated with a small CBV–follow-up difference. The adjusted coefficient of determination was 0.287. The CBV–follow-up difference was not normally distributed. Consistency of statistical significance was verified after square root transformation of this variable. DISCUSSION In this exploratory study on collateral circulation in acute ischemic stroke, we showed that dynamic CTA provides useful information about collateral blood flow. We found that the morphologic extent of the collaterals and the collateral blood flow delay were independent predictors of CTP and follow-up lesion size. Our findings are consistent with and extend prior reports showing robust collateralization in dynamic CTA to predict a small follow-up lesion.14 In these studies, the collateralization status was assessed based on the maximal morphologic extent of collateral vessels over the entire scan time. While important, no information about the time point of collateral reconstitution was provided with this method. Other studies showed that dynamic CTA can be used to determine the origin of the collaterals and to measure the time point of enhancement in these vessels;11,16 however, the predictive value of this information had not been assessed. In our study, we © 2014 ISCBFM
Predictive value of dynCTA time delay in stroke SE Beyer et al
5
Figure 3. Correlation between collateral blood flow delay and infarct size. Longer collateral blood flow delay is associated with (A) increased cerebral blood volume (CBV) lesion size in initial computed tomography (CT) perfusion (CTP) and (B) increased follow-up lesion size.
Table 2.
Univariate linear regression for initial CTP and follow-up lesion size MTT lesion size
CBV lesion size
Mismatch %
Follow-up lesion size
β = 0.056; P = 0.899 β = 34.678; P = 0.006 — — β = 37.003: P = 0.003 β = 6.665; P o0.001
β = 0.006; P = 0.987 β = 14.332; P = 0.168 — — β = 50.771; P o0.001 β = 9.878; P o0.001
β = − 0.036; P = 0.805 β = − 2.781; P = 0.509 — — β = − 20.027; Po0.001 β = − 3.976; Po0.001
β = − 0.802; P = 0.301 β = 31.887; P = 0.156 β = − 75.784; P = 0.001 β = − 94.655; Po0.001 β = 113.514; Po 0.001 β = 18.754; Po 0.001
Origin of dominant collateralization Anterior cerebral artery Reference Posterior cerebral artery β = − 16.888; P = 0.335 Indifferent β = − 5.883; P = 0.780
Reference β = − 18.249; P = 0.197 β = − 25.941; P = 0.129
Reference β = 5.841; P = 0.306 β = 10.823; P = 0.116
Reference β = − 20.630; P = 0.503 β = 33.957; P = 0.306
Morphologic collateral extent Grade 0 Grade 1 Grade 2 Grade 3 Visible ACoA Visible PCoA Visible early branch
— Reference β = − 79.287; Po 0.001 β = − 115.463; Po 0.001 β = − 0.094; P = 0.993 β = − 2.245; P = 0.827 β = − 29.606; P = 0.020
— Reference β = 30.363; P o0.001 β = 45.023; P o0.001 β = 0.537; P = 0.899 β = 0.094 P = 0.982 β = 10.844; P = 0.035
— Reference β = − 139.297; Po0.001 β = − 189.260; Po0.001 β = 3.293; P = 0.885 β = − 17.557; P = 0.428 β = − 58.399; P = 0.034
Age Male sex Intravenous thrombolysis Mechanical thrombectomy Additional ICA occlusion Collateral blood flow delay
— Reference β = − 41.559; P = 0.005 β = − 84.172; Po 0.001 β = 5.203; P = 0.688 β = − 0.187; P = 0.988 β = − 42.448; P = 0.006
ACoA, anterior communicating artery; CBV, cerebral blood volume; CTP, computed tomography perfusion; ICA, internal carotid artery; MTT, mean transit time; PCoA, posterior communicating artery; β, coefficient. Bold entries indicate significant results with Po0.05.
show that the collateral blood flow delay is an independent predictor of high relative mismatch on initial CTP and a small follow-up lesion as well as a small difference between initial CBV and follow-up lesion size. Thus, the collateral blood flow delay adds important functional information to an assessment limited to the morphologic extent of collaterals. To our knowledge, this is the first study to quantitatively show a predictive value of the collateral blood flow delay independent of the morphologic collateral grade. Qualitative assessment of hemodynamic and morphologic parameters of leptomeningeal collaterals has been used in catheter angiography grading systems and has been shown to predict a favorable outcome,24–26 showing the importance of enhancement time for collateral quality. Saito et al.27 assessed the correlation of transit time from the ICA to the M2 segment with the lesion size on follow-up NECT. However, only a few patients were included and no statistical tests were performed. Our results extend these observations because we could show that collateral blood flow delay is an independent predictor using the more feasible dynamic CTA. © 2014 ISCBFM
These findings further suggest the use of the ASITN/SIR grading system over approaches that ignore temporal information since it combines qualitative assessments of the morphologic extent and the velocity. Although dynamic CTA and the CTP are reconstructed from the same data set, they display different vessel compartments and therefore merit separate evaluation. In our study, a short collateral blood flow delay and a high morphologic collateral extent were independently associated with a small CBV lesion and a high relative mismatch on initial CT perfusion. These results show that a territory with reduced CBV represents an area receiving limited collateral blood flow thereby providing further evidence that collateralization and penumbral extent are tightly linked.3 We hypothesize that collateral blood flow delay is an indicator of collateral functionality. A short collateral blood flow delay may represent high collateral blood pressure leading to fast filling of collaterals and might increase the effectiveness of perfusion of the cerebral microcirculation represented by CTP parameters. Collateral blood flow delay may therefore potentially serve as an Journal of Cerebral Blood Flow & Metabolism (2014), 1 – 7
Predictive value of dynCTA time delay in stroke SE Beyer et al
6
Figure 4. Examples of the association between collateral blood flow delay and follow-up lesion size. (A) 59 y/o female patient with an M1 +internal carotid artery (ICA) occlusion (white arrow). Average collateral blood flow delay was 2.6 seconds. No IV thrombolysis was administered and no mechanical thrombectomy was attempted. The follow-up lesion size was 26 mL. (B) 75 y/o male patient with an M1+ICA occlusion (white arrow). An early frontal branch supplying the anterior middle cerebral artery (MCA) territory is originating from the anterior cerebral artery (ACA) (white cross). Average collateral blood flow delay was 8.4 seconds. No IV thrombolysis was administered and no mechanical thrombectomy was attempted. The follow-up lesion size was 180 mL. Both patients had collateral vessels visible in 100% of the occluded vessel territory and in both patients a follow-up angiography showed the absence of recanalization.
indicator for the selection of patients who are likely to benefit from therapies that enhance collateralization. At present benefit of therapies such as hemodilution and induced hypertension are controversial.28 With respect to patient outcome, a short collateral blood flow delay and a high morphologic extent of collaterals were also independent predictors of a small follow-up lesion. This emphasizes the importance of evaluating both parameters. The predictive value of the morphologic extent is consistent with previous reports using dynamic CTA and other imaging techniques.4,14,26 Notably, both delay and extent remained independent predictors after correcting for IV thrombolysis and mechanical thrombectomy. This suggests the beneficial effect of good collateralization in dynamic CTA, even in the absence of interventions. Previous reports indicated that the extent of collateralization in dynamic CTA predicts response to IV thrombolysis.14 The collateral blood flow delay may further improve the response to IV thrombolysis since a short delay might allow more effective delivery of recombinant tissue plasminogen activator to the distal parts of the thrombus. Further studies are needed to clarify this point. Together with IV thrombolysis and mechanical thrombectomy, a short collateral blood flow delay was an independent predictor of a small difference between initial CBV and follow-up lesion size. Notably, morphologic collateral extent was not a significant predictor in the multivariate analysis. This provides further insight into which factors determine penumbral loss. Recent studies have showed that good collateralization at baseline is an independent predictor of small penumbral loss23 and that collateral failure is Journal of Cerebral Blood Flow & Metabolism (2014), 1 – 7
associated with infarct growth.3 Our results add to that by showing an important role for collateral functionality as shown by the collateral filling velocity. These findings are in line with previous studies using Tmax to assess delay in CT perfusion. Nagakane et al.29 showed that the risk of subsequent infarction increases if Tmax increases and that this probabilistic approach allowed a better prediction of the final infarction volume than single cutoff values.29 It is likely that in patients with a longer collateral filling time there is an increased proportion of tissue-atrisk with higher Tmax values. Assuming an increased likelihood of infarction for this tissue as suggested by Nagakane et al., this may explain the observed association between collateral filling time and infarct growth. This is further emphasized by a recent study showing that a small proportion of tissue with Tmax values ≥ 10 seconds within tissue with a Tmax value of ≥ 4 seconds is associated with decreased infarct growth.30 Thus, our results may indicate the importance of the collateral filling velocity in the assessment of dynamic angiographies and may further stress probabilistic approaches taking into account the distribution of parameter values in the assessment of CT perfusion. It would, however, be desirable to investigate the influence of the morphologic extent in a larger sample and to assess the total extent of collaterals in a more quantitative way as these may be factors that have led to an underestimation of the impact of the morphologic extent in our sample. A further independent predictor of a small follow-up lesion size was the administration of IV thrombolysis. There was no evidence that the origin of dominant collateralization influenced outcome, © 2014 ISCBFM
Predictive value of dynCTA time delay in stroke SE Beyer et al
7 i.e., the extent of collateralization did not differ among groups with different origins. However, these results have to be interpreted cautiously as there were only 15 patients with an anterior dominant pattern versus 74 patients with a posterior dominant pattern. These data must be interpreted in the context of the study design. As a retrospective study, there is a potential for bias in patient selection. The use of a large cohort of consecutive patients and standardized protocols for patients with suspected stroke, however, helped to minimize this problem. Furthermore, although mechanical thrombectomy and treatment with intravenous recombinant tissue plasminogen activator were included, the state of recanalization was not assessed. However, our data apropos the follow-up infarct volume indicate that the predictive value in the initial dynamic CTA is independent of spontaneous or therapy-induced recanalization. In addition, the use of both CT and MRI as follow-up might result in heterogeneity of volume measurements. Further, measuring regional contrast enhancement CTP parameters would be a more quantitative approach but may be more heavily influenced by factors such as severe patient motion. Finally, measurement of the delay in the M2 segment and averaging of all M2 measurements is a relatively rough assessment and we acknowledge that the association between the averaged M2 delay and the lesion size might be affected by (1) further occlusions distal to the M1 segment, (2) different prominences of M2 trunks, and (3) differences in the filling time of collaterals in different areas of the MCA territory. In conclusion, we showed that assessment of the velocity of collateral filling using dynamic CTA provides an important predictor of initial CTP and follow-up lesion size and adds functional information to assess the morphologic extent of collaterals. DISCLOSURE/CONFLICT OF INTEREST The authors declare no conflict of interest.
REFERENCES 1 Liebeskind DS. Collateral circulation. Stroke 2003; 34: 2279–2284. 2 Tariq N, Khatri R. Leptomeningeal collaterals in acute ischemic stroke. J Vasc Interv Neurol 2008; 1: 91–95. 3 Campbell BC, Christensen S, Tress BM, Churilov L, Desmond PM, Parsons MW et al. Failure of collateral blood flow is associated with infarct growth in ischemic stroke. J Cereb Blood Flow Metab 2013; 33: 1168–1172. 4 Tan JC, Dillon WP, Liu S, Adler F, Smith WS, Wintermark M. Systematic comparison of perfusion-CT and CT-angiography in acute stroke patients. Ann Neurol 2007; 61: 533–543. 5 Miteff F, Levi CR, Bateman GA, Spratt N, McElduff P, Parsons MW. The independent predictive utility of computed tomography angiographic collateral status in acute ischaemic stroke. Brain 2009; 132: 2231–2238. 6 McVerry F, Liebeskind DS, Muir KW. Systematic review of methods for assessing leptomeningeal collateral flow. AJNR Am J Neuroradiol 2012; 33: 576–582. 7 Ito K, Sasaki M, Kobayashi M, Ogasawara K, Nishihara T, Takahashi T et al. Noninvasive evaluation of collateral blood flow through circle of Willis in cervical carotid stenosis using selective magnetic resonance angiography. J Stroke Cerebrovasc Dis 2013; 23: 1019–1023. 8 Menon BK, Smith EE, Modi J, Patel SK, Bhatia R, Watson TW et al. Regional leptomeningeal score on CT angiography predicts clinical and imaging outcomes in patients with acute anterior circulation occlusions. AJNR Am J Neuroradiol 2011; 32: 1640–1645. 9 Maas MB, Lev MH, Ay H, Singhal AB, Greer DM, Smith WS et al. Collateral vessels on CT angiography predict outcome in acute ischemic stroke. Stroke 2009; 40: 3001–3005.
© 2014 ISCBFM
10 Kim JJ, Fischbein NJ, Lu Y, Pham D, Dillon WP. Regional angiographic grading system for collateral flow: correlation with cerebral infarction in patients with middle cerebral artery occlusion. Stroke 2004; 35: 1340–1344. 11 Menon BK, O'Brien B, Bivard A, Spratt NJ, Demchuk AM, Miteff F et al. Assessment of leptomeningeal collaterals using dynamic CT angiography in patients with acute ischemic stroke. J Cereb Blood Flow Metab 2013; 33: 365–371. 12 Smit EJ, Vonken EJ, van der Schaaf IC, Mendrik AM, Dankbaar JW, Horsch AD et al. Timing-invariant reconstruction for deriving high-quality CT angiographic data from cerebral CT perfusion data. Radiology 2012; 263: 216–225. 13 Smit EJ, Vonken EJ, van Seeters T, Dankbaar JW, van der Schaaf IC, Kappelle LJ et al. Timing-invariant imaging of collateral vessels in acute ischemic stroke. Stroke 2013; 44: 2194–2199. 14 Calleja AI, Cortijo E, Garcia-Bermejo P, Gomez RD, Perez-Fernandez S, Del Monte JM et al. Collateral circulation on perfusion-computed tomography-source images predicts the response to stroke intravenous thrombolysis. Eur J Neurol 2013; 20: 795–802. 15 Frolich AM, Wolff SL, Psychogios MN, Klotz E, Schramm R, Wasser K et al. Timeresolved assessment of collateral flow using 4D CT angiography in large-vessel occlusion stroke. Eur Radiol 2014; 24: 390–396. 16 Siebert E, Diekmann S, Masuhr F, Bauknecht HC, Schreiber S, Klingebiel R et al. Measurement of cerebral circulation times using dynamic whole-brain CT-angiography: feasibility and initial experience. Neurol Sci 2012; 33: 741–747. 17 Bivard A, Levi C, Spratt N, Parsons M. Perfusion CT in acute stroke: a comprehensive analysis of infarct and penumbra. Radiology 2013; 267: 543–550. 18 Royalty K, Manhart M, Pulfer K, Deuerling-Zheng Y, Strother C, Fieselmann A et al. C-arm CT measurement of cerebral blood volume and cerebral blood flow using a novel high-speed acquisition and a single intravenous contrast injection. AJNR Am J Neuroradiol 2013; 34: 2131–2138. 19 Menon BK, Bal S, Modi J, Sohn SI, Watson TW, Hill MD et al. Anterior temporal artery sign in CT angiography predicts reduced fatal brain edema and mortality in acute M1 middle cerebral artery occlusions. J Neuroimaging 2012; 22: 145–148. 20 Schaefer PW, Barak ER, Kamalian S, Gharai LR, Schwamm L, Gonzalez RG et al. Quantitative assessment of core/penumbra mismatch in acute stroke: CT and MR perfusion imaging are strongly correlated when sufficient brain volume is imaged. Stroke 2008; 39: 2986–2992. 21 Thierfelder KM, Sommer WH, Baumann AB, Klotz E, Meinel FG, Strobl FF et al. Whole-brain CT perfusion: reliability and reproducibility of volumetric perfusion deficit assessment in patients with acute ischemic stroke. Neuroradiology 2013; 55: 827–835. 22 Kamalian S, Kamalian S, Maas MB, Goldmacher GV, Payabvash S, Akbar A et al. CT cerebral blood flow maps optimally correlate with admission diffusion-weighted imaging in acute stroke but thresholds vary by postprocessing platform. Stroke 2011; 42: 1923–1928. 23 Jung S, Gilgen M, Slotboom J, El-Koussy M, Zubler C, Kiefer C et al. Factors that determine penumbral tissue loss in acute ischaemic stroke. Brain 2013; 136: 3554–3560. 24 Sanossian N, Saver JL, Alger JR, Kim D, Duckwiler GR, Jahan R et al. Angiography reveals that fluid-attenuated inversion recovery vascular hyperintensities are due to slow flow, not thrombus. AJNR Am J Neuroradiol 2009; 30: 564–568. 25 Ovbiagele B, Saver JL, Starkman S, Kim D, Ali LK, Jahan R et al. Statin enhancement of collateralization in acute stroke. Neurology 2007; 68: 2129–2131. 26 Bang OY, Saver JL, Buck BH, Alger JR, Starkman S, Ovbiagele B et al. Impact of collateral flow on tissue fate in acute ischaemic stroke. J Neurol Neurosurg Psychiatry 2008; 79: 625–629. 27 Saito I, Segawa H, Shiokawa Y, Taniguchi M, Tsutsumi K. Middle cerebral artery occlusion: correlation of computed tomography and angiography with clinical outcome. Stroke 1987; 18: 863–868. 28 Shuaib A, Butcher K, Mohammad AA, Saqqur M, Liebeskind DS. Collateral blood vessels in acute ischaemic stroke: a potential therapeutic target. Lancet Neurol 2011; 10: 909–921. 29 Nagakane Y, Christensen S, Ogata T, Churilov L, Ma H, Parsons MW et al. Moving beyond a single perfusion threshold to define penumbra: a novel probabilistic mismatch definition. Stroke 2012; 43: 1548–1555. 30 Olivot JM, Mlynash M, Inoue M, Marks MP, Wheeler HM, Kemp S et al. Hypoperfusion intensity ratio predicts infarct progression and functional outcome in the DEFUSE 2 Cohort. Stroke 2014; 45: 1018–1023.
Journal of Cerebral Blood Flow & Metabolism (2014), 1 – 7
