Relative Cerebral Blood Volume as a Marker of Durable Tissue-at-Risk Viability in Hyperacute Ischemic Stroke
Background and Purpose—Selection of best responders to reperfusion therapies could be aided by predicting the duration of tissue-at-risk viability, which may be dependant on collateral circulation status. We aimed to identify the best predictor of good collateral circulation among perfusion computed tomography (PCT) parameters in middle cerebral artery (MCA) ischemic stroke and to analyze how early MCA response to intravenous thrombolysis and PCT-derived markers of good collaterals interact to determine stroke outcome.
Methods—We prospectively studied patients with acute MCA ischemic stroke treated with intravenous thrombolysis who underwent PCT before treatment showing a target mismatch profile. Collateral status was assessed using a PCT source image–based score. PCT maps were quantitatively analyzed. Cerebral blood volume (CBV), cerebral blood flow, and Tmax were calculated within the hypoperfused volume and in the equivalent region of unaffected hemisphere. Occluded MCAs were monitored by transcranial Duplex to assess early recanalization. Main outcome variables were brain hypodensity volume and modified Rankin scale score at day 90.
Results—One hundred patients with MCA ischemic stroke imaged by PCT received intravenous thrombolysis, and 68 met all inclusion criteria. A relative CBV (rCBV) >0.93 emerged as the only predictor of good collaterals (odds ratio, 12.6; 95% confidence interval, 2.9–55.9; P=0.001). Early MCA recanalization was associated with better long-term outcome and lower infarct volume in patients with rCBV<0.93, but not in patients with high rCBV. None of the patients with rCBV<0.93 achieved good outcome in absence of early recanalization.
Conclusions—High rCBV was the strongest marker of good collaterals and may characterize durable tissue-at-risk viability in hyperacute MCA ischemic stroke.
- arterial recanalization
- collateral circulation
- ischemic stroke
- perfusion computed tomography
- thrombolytic therapy
The development of more effective reperfusion strategies for ischemic stroke patients with acute intracranial large artery occlusions is a main research priority.1 Although endovascular therapy may result in higher early arterial recanalization rates compared with intravenous thrombolysis,2 recently published randomized clinical trials have failed to prove superiority of first-wave endovascular devices compared with best medical treatment available.3–5 Almost 40% of patients in the medical arm of Interventional Management of Stroke-3 (IMS-3) study achieved good functional outcome, which raises concern about the methodology used for patient selection.3 Furthermore, the results of Mechanical Retrieval and Recanalization of Stroke Clots Using Embolectomy (MR-RESCUE) study suggest that the penumbral pattern alone lacks specificity as a marker of durable tissue viability.5 Thus, variability in the endurance of tissue at risk against hypoperfusion may be critical to determine whether ischemic tissue can resist and be saved even after delayed reperfusion, which is often the case for patients treated with intravenous thrombolysis. Therefore, beyond improving the demarcation of salvageable tissue, selection of best responders to endovascular therapy could be aided by neuroimaging markers of expected resistance of tissue at risk in the setting of a persistent arterial occlusion.
Collateral circulation has been postulated as a critical determinant of the duration of ischemic tissue viability in acute stroke.6 Different scores to assess leptomeningeal collaterals (LMC) have been described using digital subtraction cerebral angiography,7 computed tomography angiography,8 and, more recently, perfusion computed tomography source images (PCT-SI).9 Whereas noninvasive analysis of collateral circulation is time-consuming and dependant on reader’s interpretation, new perfusion imaging processing software automatically and rapidly yield quantitative parameters to assess brain perfusion status.10 However, the relationship between quantitative brain perfusion maps and collateral circulation scores has not been sufficiently explored. Searching for imaging markers of durable tissue-at-risk viability in hyperacute ischemic stroke, we aimed to identify the best predictor of good collateral circulation among quantitative PCT parameters and to analyze how early response to intravenous thrombolysis and PCT-derived markers of good collaterals interact to determine stroke outcome.
Patients and Methods
We prospectively evaluated patients with an acute nonlacunar middle cerebral artery (MCA) ischemic stroke admitted to our stroke unit from October 2009 to August 2012. All patients fulfilled criteria to receive intravenous thrombolysis with tissue plasminogen activator (tPA) and underwent PCT before tPA bolus injection. From September 2008, after approval by our ethics committee, our protocol allowed PCT-guided extension of intravenous thrombolysis therapeutic window in patients presenting >4.5 hours from onset or in cases with unclear onset time, whenever inclusion in an ongoing clinical trial was impossible. Indication of intravenous thrombolysis within the first 4.5 hours was based on plain CT images. Patients admitted >4.5 hours received intravenous tPA according to previously published PCT criteria.11 Initial candidates had to fulfill the following additional criteria to enter this study: (1) no contraindication for iodinated contrast agent administration, (2) single acute ischemic lesion affecting MCA territory, (3) PCT mean transit time maps showing acute ischemic involvement of ≥2 contiguous MCA cortical ASPECTS (Alberta Stroke Program Early CT Score) areas (M1–M6), (4) good PCT-SI quality allowing LMC interpretation and automatic postprocessing of perfusion maps, and (5) target mismatch profile as defined by quantitative perfusion maps.
The study protocol was approved by the local ethics committee, and informed consent was obtained from all patients or their legal representatives.
All included patients were managed according to our institutional protocol, which is based on updated international guidelines. Intravenous thrombolysis was administered in a 0.9 mg/kg tPA dose.12 Stroke severity was assessed with the National Institutes of Health Stroke Scale (NIHSS) on admission and periodically during the next 24 hours.
Stroke subtypes were classified using modified Trial of Org 10172 in Acute Stroke Treatment (TOAST) criteria,13 in agreement with the results of the additional diagnostic procedures performed. Early neurological improvement was defined as a decrease ≥4 points in NIHSS score during the first 24 hours, or as a NIHSS score 0 or 1 at 24 hours. Symptomatic hemorrhagic transformation was defined as any hemorrhage causing an increase of ≥4 points in the NIHSS. Good long-term functional outcome was defined as a modified Rankin scale score ≤2 at day 90.
Imaging Protocol: CT Acquisition
Cerebral CT scans were performed before tPA bolus and repeated after 24 hours and whenever neurological deterioration occurred. Early ischemic changes on CT at admission were evaluated using ASPECTS.14 On 24-hour CT, brain hypodensity volume was calculated using the formula for irregular volumes (A×B×C/2), where A is the hypodensity’s largest diameter; B, the perpendicular diameter; and C, the coronal diameter. Both noncontrast CT and PCT were performed in either a GE Lightspeed scanner with 64 rows of detectors or a Toshiba Aquilion equipped with 32 rows of detectors depending on their availability (see online-only Data Supplement).
Imaging Protocol: PCT-SI–Based Collateral Circulation Assessment
Dynamic PCT-SI was used to assess LMC. Our PCT-SI–based LMC scoring protocol has been described elsewhere.9 Basically, the degree of LMC arterial filling within ischemic tissue boundaries was categorized as follows: 0, absence of vessels; 1, collateral supply filling <50%; 2, collateral supply filling between >50% and <100%; and 3, equal or more prominent when compared with the unaffected hemisphere. The scale was divided into 2 groups: good (scores 2–3) and poor (scores 0–1) collaterals. Image review was performed on a picture archiving and communications system workstation (Impax 6.0:AGFA Technical Imaging System, Mortsel, Belgium) by 2 neurologists experienced in stroke imaging interpretation (A.I.C., J.F.A.). Disagreements in LMC scoring were resolved by independent evaluation of a neuroradiologist (S.P.).
Imaging Protocol: Quantitative PCT Processing
All PCT-SIs were postprocessed using commercially available automatic software (Perfscape 2.0, Neuroscape 2.0; Olea Medical). The contralateral anterior cerebral artery was selected to provide arterial input function,15 whereas venous outflow function was derived from the superior sagital sinus. Singular value decomposition (SVD) deconvolution was performed with standard (sSVD; delay-sensitive) and oscillatory index–regularized block circular (oSVD; delay-insensitive) algorithms to create maps of cerebral blood flow (CBF), mean transit time, and time to peak of the deconvolved tissue residue function (Tmax). Cerebral blood volume (CBV) and time to peak were calculated from the concentration time curve. In line with previous studies, infarct core threshold was defined by a relative CBF map <31% to the mean of contralateral hemisphere, whereas brain tissue-at-risk outer boundary was defined by Tmax >6 seconds.16,17 To avoid falsely low CBF in periventricular and leukoaraiosis areas, we manually excluded regions with low CBF outside the hypoperfused volume. Mismatch volume was defined as hypoperfused volume minus infarct core. Volumes of hypoperfused tissue, infarct core, and mismatch region were automatically generated by the software. Target mismatch profile was defined by a hypoperfused volume ≥15 mL, a mismatch ratio ≥1.8, and an infarct core <70 mL.18
Once perfusion maps were available, regions of interest were selected in both cerebral hemispheres. The hypoperfused areas, as defined in Tmax maps, were manually delineated on all axial sections in the ischemic hemisphere, and the corresponding areas in the unaffected side appeared automatically demarcated. Manual delineation of hypoperfused areas on Tmax maps was performed by consensus of 2 observers (E.C., J.F.A.) to avoid discrepancies in region of interest definition. The software provided CBV and CBF values from every region of interest, and relative CBF (rCBF) and relative CBV (rCBV) in every section were calculated as the quotient between the values from ischemic and contralateral regions of interest. Finally, mean rCBV and rCBF were calculated by adding the values obtained in every section divided by the number of sections. All PCT image processing was performed by E.C., who was blinded to LMC assessment.
Transcranial Ultrasound Assessment of MCA Recanalization
All extracranial and transcranial ultrasound imaging were performed with a Toshiba Aplio XG echograph (Toshiba Medical Systems Europe, Zoetermeer, the Netherlands). Transcranial Duplex examinations were performed right before tPA infusion and 2 hours after tPA bolus. MCA occlusions were defined according to the thrombolysis in brain ischemia (TIBI) grading system,19 which establishes grades 0 to 3 as indicative of arterial occlusion. In the control examination, TIBI grades 4 and 5 were indicative of early complete recanalization, whereas an increase in TIBI grade without reaching 4 or 5 grade was considered as partial recanalization.
Statistical analyses were performed with the SPSS statistical package (version 18.0; SPSS Inc, Chicago, IL). Interobserver agreement for LMC assessment was calculated with the κ test. Statistical significance for intergroup differences was assessed by the χ2 test for categorical variables, and the Student t-test and Mann–Whitney U test for continuous variables. All continuous variables, except NIHSS score, noncontrast CT–ASPECTS, leukocyte and platelet count, were normally distributed.
Long-term clinical outcome, early neurological recovery, 24-hour hypodensity volume, and symptomatic hemorrhagic transformation were considered outcome variables. To analyze the association between PCT parameters and collaterals, and PCT and outcome variables, crude and adjusted logistic or linear regression models, whenever appropriate depending on respective outcome variables, were applied. Quantitative PCT parameters were included in the models firstly as continuous variables and secondly as dichotomous variables after best cutoff points were obtained by means of receiver operating characteristic curves. For each end point, adjustment was done by entering baseline clinical or radiological variables showing P<0.1 on respective bivariate analyses. Results of the regression analyses are expressed as odds ratios (ORs) and their corresponding confidence intervals (CIs). For all tests, a probability value <0.05 was considered statistically significant.
During the study period, 100 patients with an acute MCA stroke treated with intravenous tPA underwent PCT before tPA bolus. Of them, 68 fulfilled all imaging selection criteria and were included in the study. Reasons for exclusion of the remaining 32 patients were: insufficient PCT-SI quality (n=4), mean transit time abnormality involving <2 contiguous MCA cortical ASPECTS areas (n=15), quantified hypoperfused tissue volume <15 cc (n=8), and infarct core >70 cc (n=5). All included patients showed a target mismatch pattern on baseline PCT. Table 1 summarizes the demographic characteristics and clinical variables of the study sample. Thirty (44.1%) patients were women, mean age was 79.4 years, and median baseline NIHSS was 13. Regarding thrombolytic therapy, 30 received tPA in a standard time window and 38 beyond 4.5 hours from symptom onset.
Collateral Circulation and Perfusion CT Parameters
LMC score was distributed as follows: 3, 21, 30, and 14 patients had LMCs of 0, 1, 2, and 3, respectively. Thus, 44 (64.7%) were classified as having good collaterals, whereas 24 (35.3%) showed poor collaterals. There was a good interobserver agreement in LMC grading (κ index, 0.724; P<0.01). Demographic characteristics and baseline clinical and radiological variables, distributed attending collateral circulation status, are shown on Table 1.
Collateral score was significantly associated with mean rCBV (P<0.001), mean rCBF (P=0.005), and infarct core (P<0.001) in bivariate analyses (Table 1). rCBV, rCBF, and infarct core, expressed as continuous variables, were associated with collaterals in the crude logistic regression models. Receiver operating characteristic curve–derived best cutoff values to discriminate between good and poor collaterals were: 0.93 for rCBV (sensitivity, 70%; specificity, 87%; P<0.001; area under the curve [AUC], 0.785), 0.77 for rCBF (sensitivity, 58%; specificity, 70%; P=0.02; AUC, 0.704), and 5.30 for infarct core (sensitivity, 56%; specificity, 74%; P=0.01; AUC, 0.735). Crude and adjusted logistic regression models were applied for each perfusion parameter as shown on Table 2. A rCBV >0.93 emerged as the only independent predictor of good collaterals (OR, 12.63; 95% CI, 2.85–55.93; P=0.001). ASPECTS on plain CT misclassified a significant proportion of patients attending this rCBV cutoff; interestingly, 34% of patients with an ASPECTS of 10 showed rCBV <0.93.
Prognostic Value of PCT Quantitative Parameters
More detailed information regarding prognostic variables and the predictive capacity of PCT parameters is provided in the online-only Data Supplement. No PCT parameter alone was an independent predictor of short- and long-term clinical outcome. Good functional outcome was achieved by 34 (50%) patients. Infarct core volume and rCBV were associated with long-term outcome in bivariate analyses, but this association disappeared after adjustment by baseline NIHSS and noncontrast CT–ASPECTS. Regarding 24-hour hypodensity volume, a multiple adjusted linear regression analysis identified baseline infarct core (β=0.434; P=0.001) as an independent predictor of a larger infarct volume.
Early MCA Recanalization, rCBV, and Stroke Outcome
Baseline transcranial Duplex before tPA bolus was performed in 57 patients, of whom 52 showed a TIBI pattern indicative of MCA occlusion. All of them underwent control transcranial Duplex 2 hours after tPA start. Early complete or partial MCA recanalization was observed in 18 (35%) patients. Time from stroke onset to treatment start was not significantly associated with the probability of early recanalization (P=0.73).
The study sample was divided in 2 groups attending rCBV cutoff value of 0.93, and the prognostic impact of early MCA recanalization was assessed in both rCBV groups. In the whole sample, early recanalization emerged as an independent predictor of long-term outcome (OR, 7.5; CI, 1.1–52.6; P=0.04). In patients with rCBV <0.93, early MCA recanalization was associated with a higher probability of good long-term outcome (P<0.001), early neurological improvement (P<0.001), and lower 24-hour hypodensity volume (P=0.003). However, early recanalization was not significantly associated with good long-term outcome (P=0.286), early neurological recovery (P=0.09), or infarct volume (P=0.212) in patients with rCBV >0.93. Whereas none of the patients with a low rCBV had a good 90-day outcome in absence of early MCA recanalization, 9 (52.9%) patients with a high rCBV became functionally independent despite having a persistent MCA occlusion 2 hours after tPA bolus. Figures 1 and 2 illustrate the differential response profile of both rCBV groups to early recanalization.
Our study was performed in acute MCA ischemic stroke patients treated with intravenous thrombolysis who showed a target mismatch pattern on PCT performed before tPA bolus. All brain perfusion imaging was postprocessed using automatic software, with the general purpose of searching for quantitative markers of durable tissue-at-risk viability. The study had 2 main findings. First, relative CBV, defined as the ratio between CBV measured in hypoperfused cerebral tissue and CBV obtained in the correspondent volume of the unaffected hemisphere, appeared as the most robust marker of LMC status. And second, rCBV and early MCA response to intravenous tPA may interact to determine stroke outcome, in the way that early MCA recanalization seemed to have a greater impact on long-term outcome in the group of patients with low rCBV. After these observations, high rCBV emerges as a potential marker of durable tissue-at-risk viability, an indicator of the capacity of salvageable ischemic tissue to resist longer in absence of early arterial recanalization.
Noninvasive assessment of collateral circulation represents a promising strategy to help select best candidates for reperfusion therapies.20 Collateral perfusion has a great prognostic impact in acute ischemic stroke, and recent literature has consistently shown that a better grade of collaterals is predictive of good clinical outcome.3,5,18,20 Our group recently proposed a PCT-SI–based score to assess collateral status, which predicted long-term clinical outcome in acute stroke patients treated with intravenous thrombolysis.9 PCT-SI has a video clip format that shows a dynamic flow motion in LMCs, allowing a clear distinction between arterial filling and emptying, which may represent a potential advantage over computed tomography angiography in collateral assessment. However, collateral grading using PCT-SI or other angiographic techniques has limited applicability in the hyperacute setting, because it is laborious, requires eye-training, and depends on observer’s interpretation. In contrast, perfusion postprocessing software automatically generates perfusion maps and calculates perfusion parameters in a few minutes.10 Given that perfusion parameters may indirectly reflect collateral status, we tried to correlate PCT-SI collateral score with quantitative parameters and found that rCBV was the strongest indicator of LMC circulation. Interestingly, rCBV >0.93 predicted good collaterals, and this finding warrants validation in a larger sample.
Of note, quantitative perfusion parameters showed limited prognostic value. In line with previous studies, the extent of infarct core on PCT was an independent predictor of 24-hour hypodensity volume.21 However, no perfusion parameter alone could predict clinical outcome variables. Our results reaffirm the notion that penumbral pattern on perfusion maps itself is not sufficient to predict tissue fate and patient’s outcome.22 Real-time information about initial status of the culprit vessel and response to reperfusion therapies has a crucial prognostic importance and should ideally be available to complement perfusion maps.5,23 In this context, arterial information is always available in patients treated with endovascular therapies, but is often missing in studies with patients treated with intravenous thrombolysis, who constitute the medical arm in thrombectomy trials. Thus, our methodology combining quantitative brain perfusion assessment and transcranial Duplex MCA monitoring in intravenous tPA patients may represent a strength of this study.
We found that the prognostic impact of early MCA response to intravenous tPA seems to be modulated by the PCT parameter reflecting collateral status, that is, rCBV. Whereas none of the patients with low rCBV achieved good outcome in absence of early recanalization, a substantial proportion (52.9%) of patients in the high rCBV group performed excellent even in absence of early MCA recanalization. Moreover, absence of early recanalization was not associated with a greater 24-hour hypodensity volume in the high rCBV group, which suggests that the viability of ischemic tissue lasts longer in this group. However, the distribution of modified Rankin scale in the high rCBV group suggests that early recanalization would lead to statistically better outcomes in a larger series. Taken together, these findings suggest that in patients with high rCBV, a better collateral flow may support penumbral tissue until delayed recanalization occurs, which would explain the favorable outcomes observed in absence of early recanalization only in this group of patients. In contrast, target mismatch patients with low rCBV depend dramatically on early arterial recanalization to achieve good outcome and, therefore, may benefit the most from endovascular reperfusion.
This study has limitations. First, the final sample was small in size, although highly selected, and therefore our results need to be confirmed in a larger series. Second, ultrasound protocol could not be fully accomplished in the whole study sample. Third, the adquisition of PCT source images was performed with 2 different scanners depending on their availability. However, we have not found any significant differences either in baseline or outcome variables among the groups of patients examined with each scanner. Fourth, our PCT-SI–based collateral score has not been validated against cerebral angiography.
In conclusion, high rCBV was the strongest marker of good collateral status and may help characterize durable tissue-at-risk viability in hyperacute MCA ischemic stroke. Further studies are needed to confirm the clinical applicability of this parameter in the selection of acute ischemic stroke patients for reperfusion therapies.
Sources of Funding
Drs Cortijo and Calleja were recipients of research grants from Instituto de Salud Carlos III (Río Hortega-program), Ministery of Science, Spain (2008–2011 and 2013–2014, respectively). The study was founded by the research budget of the Stroke-Program-Hospital-Clinico-Universitario-Valladolid.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.113.003340/-/DC1.
- Received August 28, 2013.
- Revision received October 4, 2013.
- Accepted October 19, 2013.
- © 2013 American Heart Association, Inc.
- Rha JH,
- Saver JL
- Bang OY,
- Saver JL,
- Kim SJ,
- Kim GM,
- Chung CS,
- Ovbiagele B,
- et al
- Christoforidis GA,
- Mohammad Y,
- Kehagias D,
- Avutu B,
- Slivka AP
- Maas MB,
- Lev MH,
- Ay H,
- Singhal AB,
- Greer DM,
- Smith WS,
- et al
- Adams HP Jr.,
- Bendixen BH,
- Kappelle LJ,
- Biller J,
- Love BB,
- Gordon DL,
- et al
- Campbell BC,
- Christensen S,
- Levi CR,
- Desmond PM,
- Donnan GA,
- Davis SM,
- et al
- Campbell BC,
- Christensen S,
- Levi CR,
- Desmond PM,
- Donnan GA,
- Davis SM,
- et al
- Demchuk AM,
- Burgin WS,
- Christou I,
- Felberg RA,
- Barber PA,
- Hill MD,
- et al
- Gasparotti R,
- Grassi M,
- Mardighian D,
- Frigerio M,
- Pavia M,
- Liserre R,
- et al
- Zhu G,
- Michel P,
- Aghaebrahim A,
- Patrie JT,
- Xin W,
- Eskandari A,
- et al
- Fiebach JB,
- Al-Rawi Y,
- Wintermark M,
- Furlan AJ,
- Rowley HA,
- Lindstén A,
- et al