The Cortical Ischemic Core and Not the Consistently Present Penumbra Is a Determinant of Clinical Outcome in Acute Middle Cerebral Artery Occlusion
Background and Purpose— Patient selection for acute stroke therapy based on physiology rather than on time may lead to expansion of the therapeutic window, improved outcomes, and fewer side effects than currently achieved. This approach requires early determination of both irreversible (core) and reversible (penumbra) ischemia in acute stroke.
Methods— Using established perfusion thresholds, we characterized the relationship among core, penumbra, and brain tissue perfused above penumbral thresholds (non-core/non-penumbra [NC/NP]) in 36 patients with middle cerebral artery (MCA) stem occlusion who underwent quantitative cerebral blood flow (CBF) assessment with xenon-enhanced CT within 6 hours of symptom onset. Results were expressed as percentage of core, percentage of penumbra, or percentage of NC/NP relative to the ipsilateral cortical MCA territory and were correlated with clinical and radiological variables and with clinical outcomes.
Results— While great variability in the mean±SD percentage of core (37.6±18.7) and NC/NP (30.3±16.6) was observed, the percentage of penumbra was relatively constant from individual to individual, constituting approximately one third of the cortical MCA territory (32.1±7). In univariable and multivariable analyses, percent core and not percent penumbra was significantly associated with outcome.
Conclusions— In acute MCA occlusion, penumbra is consistently present within a relatively narrow range, despite great variability in the size of core. This may explain why the core and not the penumbra is the main determinant of outcome in our group of patients. Recanalization therapy in acute MCA occlusion should ideally be guided by diagnostic methods capable of rapidly and reliably identifying irreversible ischemia.
- brain ischemia
- cerebral blood flow
- computed tomography, x-ray computed
- middle cerebral artery occlusion
The wide spectrum of outcomes observed after recanalization therapy in patients with middle cerebral artery (MCA) occlusion1,2 suggests that the extent of reversible (penumbra) and irreversible (core) ischemic insult in acute stroke varies greatly. The ischemic core represents tissue that is irreversibly damaged. Beyond a certain time limit (probably no longer than 1 hour of continuous vessel occlusion), it corresponds to cerebral blood flow (CBF) values of <7 to 12 mL/100 g per minute.3 Thrombolytic therapy administered to patients with large volumes of infarcted tissue is associated with a substantial increase in the risk of symptomatic hemorrhage and malignant cerebral edema,4–8 underlying the importance of determining the extent of core as a guide to patient selection for thrombolytic therapy. The ischemic penumbra represents
See Editorial Comment, page 2434
tissue that is functionally impaired but structurally intact. It corresponds to a high CBF limit of 17 to 22 mL/100 g per minute and to a low CBF limit of 7 to 12 mL/100 g per minute.3 Knowledge of the extent of penumbra is also important because if the penumbra were large, aggressive therapies directed toward reperfusion might be warranted well beyond the currently accepted therapeutic window.9
Assessment of core and penumbra in human ischemic stroke by various methods of quantitative CBF measurement (positron emission tomography [PET], xenon-enhanced CT CBF [Xe-CT CBF] evaluation) has yielded contradictory results.10–12 These studies were limited by small sample sizes, prolonged time lapse from stroke onset to CBF assessment, and heterogeneous vascular occlusion sites. The objective of this study was to assess the extent of core and penumbra in acute MCA stem occlusion and to determine their relationship with clinical outcome.
Subjects and Methods
This study was conducted with institutional review board approval (institutional review board No. 020178).
Thirty-six patients with acute ischemic stroke were included according to the following criteria: onset of symptoms within 6 hours of undergoing a Xe-CT CBF study, MCA stem occlusion determined by conventional angiogram or CT angiogram at the time of the CBF study, and lack of significant motion artifact on the CBF study.
We retrospectively selected our patients from a prospective registry of 378 consecutive individuals admitted to the University of Pittsburgh Stroke Service between January 1997 and April 2001 who underwent a Xe-CT CBF study, of which 160 patients were studied within 6 hours of symptoms onset. Fifty patients had MCA occlusion. Fourteen of these 50 patients were excluded from further analysis because of excessive motion artifact. Clinical and demographic data were obtained from either a prospectively acquired clinical database or from medical records. All patients screened had a documented time of symptom onset and a documented time at which the CBF study was performed. Outcome data at 3 to 6 months were available from medical records or through telephone interview. To obtain a binary outcome, outcomes were classified as favorable, representing no or minimal disability and able to perform all prior activities (modified Rankin Scale [mRS] score 0 to 1) or unfavorable, representing death or disability (mRS score 2 to 6).
Xe-CT CBF Data Analysis
Details of the stable Xe-CT CBF technique have been published previously.13,14 Four CT images of 1-cm slice thickness were obtained along the orbitomeatal line (Figure 1⇓A). Although the duration of each study was not recorded, a standard head CT followed by a Xe-CT CBF study typically requires approximately 20 minutes for acquisition and 5 additional minutes for CBF calculation and display. Our analysis included the ipsilateral and contralateral outer 2 cm of cortical MCA territory of all 4 levels, which, for standardization purposes, was defined by anatomic templates that are part of the Xe-CT computer software (Diversified Diagnostic Products Inc) (Figure 1⇓B). A computer-generated division of the entire cortex into 20 regions of interest (ROIs) was also obtained (Figure 1⇓C), with 6 ROIs per slice (3 to 8 on the right and 13 to 18 on the left) corresponding to the computer-generated template representing the cortical MCA territory. Thus, the total number of ROIs corresponding to the cortical MCA territory on each side is 24.
Voxels (1×1×10 mm) corresponding to a CBF range of 0 to 8 mL/100 g per minute were separated out of each level and counted by the computer software in the ipsilateral and contralateral MCA cortex (Figure 1⇑D). Areas of lack of flow due to prominent cortical sulci were not excluded from analysis. For each side, voxels corresponding to this flow range in all 4 levels were summed, and the number obtained was divided by the sum of voxels corresponding to the entire (ie, not separated by flow thresholds) ipsilateral cortical MCA territory in the same 4 levels and expressed as percentage of ischemic core to the cortical MCA territory. In a similar fashion, the number of voxels in the cortical MCA territory corresponding to a CBF range of 8 to 20 mL/100 g per minute (Figure 1⇑E) and >20 mL/100 g per minute (Figure 1⇑F) were summed and divided by the total number of voxels for the MCA cortex to yield percentage of penumbra and non-core/non-penumbra (NC/NP) values, respectively. To include the deep structures supplied by the MCA in our analysis, mean CBF values for the entire ipsilateral and contralateral MCA territory (cortex and deep structures) were obtained with the use of previously established anatomic templates,15 according to which the MCA territory was manually tracked. In these tracked areas, regional CBF values were computed by the Xe-CT software package, yielding mean total MCA CBF values for each of the 4 levels. These values were then averaged into 1 value representing mean total MCA CBF.
ROI-based analysis of percent core, penumbra, and NC/NP for ipsilateral and contralateral cortical MCA territory was performed to compensate for potential errors induced by a voxel-based analysis.
The number of cortical ROIs (mean, 314 voxels per ROI) with a mean CBF of <8, 8 to 20, and >20 mL/100 g per minute, respectively, was divided by 24. These values were correlated with the percentages of core, penumbra, and NC/NP obtained through voxel-based analysis.
Other Imaging Data Analysis
Noncontrasted baseline CT scans were reviewed by a blinded neuroradiologist (S.G.) and assessed for early CT changes involving greater or less than one third of MCA territory. Follow-up imaging studies (CT or MRI) were reviewed by a blinded coinvestigator (H.Y.), and levels in the brain similar to the levels used by the Xe-CT CBF software were analyzed and divided into the same ROIs. Each cortical MCA ROI was deemed infarcted or noninfarcted, and percent ipsilateral cortical MCA infarction on follow-up imaging was determined by dividing the number of infarcted ipsilateral cortical ROIs by 24.
In the 23 patients who underwent intra-arterial thrombolytic therapy, MCA recanalization status at 2 hours after angiogram was categorized as absent (Thrombolysis in Myocardial Infarction [TIMI] grades 0 and 1) or present (TIMI grades 2 and 3).
Statistical analyses were performed with the Intercooled Stata 7.0 (Stata Inc) statistical software package. Pearson and Spearman correlation coefficients were used for correlations involving normally distributed and nonnormally distributed variables, respectively. The t test was used to compare means of normally distributed variables, and the Wilcoxon matched-pairs signed rank test was performed to assess the equality of rank distributions between nonnormally distributed variables. ANOVA was used to compare means of normally distributed variables across several categorical variables of interest. Multivariate analysis of the relation between several variables of interest was performed in a stepwise logistic regression model in which entry was set at a univariate association with a probability value of ≤0.1.
Median age was 69 years (range, 24 to 89 years), with a female-to-male ratio of 2:1. The median National Institutes of Health Stroke Scale (NIHSS) score was 18 (range, 2 to 25). Twenty-six patients received thrombolytic therapy; 3 patients received intravenous thrombolysis, 12 patients received intra-arterial thrombolysis, and 11 patients received combined intravenous/intraarterial thrombolysis. The median time from onset of symptoms to the CBF study was 240 minutes (range, 120 to 360 minutes). For thrombolysis-treated patients, the mean time interval between symptom onset and thrombolysis was 218 minutes (range, 120 to 360). Eleven of 36 patients (30.56%) were intubated and mechanically ventilated. Three of 36 patients (8.33%) received deep sedation with propofol, and 24 patients (66.67%) received sedation with lorazepam, midazolam, or fentanyl.
Great variability in the mean±SD percentage of core (37.6±18.7%; range, 7.6% to 70.5%) and NC/NP (30.3±16.6%; range, 4.7% to 66%) was observed. However, the percentage of penumbra was relatively constant (32.1±7%; range, 16.2% to 46.9%) (Figure 2).
Ipsilateral percentage of core and mean ipsilateral MCA CBF did not correlate with the time after symptom onset at which the CBF study was performed, with the admission serum glucose level, or with mean arterial pressure. These values did not significantly differ between sedated and nonsedated patients. Ipsilateral percentage of core was strongly inversely correlated with ipsilateral percent NC/NP and with ipsilateral mean MCA CBF and was moderately inversely correlated with percent penumbra and with NIHSS score (Table 1). There was a high correlation between ipsilateral and contralateral percentages of core, penumbra, and NC/NP obtained through voxel-based analysis and those obtained through ROI-based analysis (Spearman ρ=0.93, P<0.0001 for percent core; ρ=0.75, P<0.0001 for percent penumbra; and ρ=0.95, P<0.0001 for percent NC/NP) (Table 2).
We observed significant differences in mean CBF values between ipsilateral and contralateral total MCA CBF (17.6 versus 37.5 mL/100 g per minute, respectively; P<0.00001) and a high correlation between mean ipsilateral total MCA CBF and mean ipsilateral cortical MCA CBF (Spearman ρ=0.94, P<0.0001). Patients with favorable outcomes had significantly lower percent core than patients with unfavorable outcomes (mean±SD, 26±14.1%; range, 9.5% to 50.4% versus mean±SD, 45±17.6%; range, 7.6% to 70.5; P=0.0073). This difference remained significant (P=0.04) after we controlled for other variables (age, NIHSS score, mean arterial pressure) that were found to be significantly associated with favorable outcome. There was no significant difference in percent penumbra in the favorable outcome group compared with the unfavorable outcome group (32.6% versus 31%; P=0.5). In a stepwise logistic regression model that included core, penumbra, age, NIHSS score, and mean arterial pressure, variables that were found to be significantly associated with favorable outcomes in univariate analyses (Table 1), only percent core (odds ratio [OR], 0.85; 95% CI, 0.74 to 0 0.98; P=0.025) and age (OR, 0.85; 95% CI, 0.73 to 0.97; P=0.020) remained significantly associated with favorable outcome. When percent core and percent penumbra were substituted with ipsilateral mean MCA CBF, age (OR, 0.84; 95% CI, 0.73 to 0.98; P=0.025) and mean MCA CBF (OR, 1.39; 95% CI, 1.05 to 1.84; P=0.02) were significantly associated with favorable outcome. Substituting total ipsilateral MCA CBF for cortical MCA CBF in the logistic regression model yielded similar results; age (OR, 0.86; 95% CI, 0.76 to 0.98; P=0.0026) and ipsilateral total MCA CBF (OR, 1.33; 95% CI, 1.05 to 1.69; P=0.016) were the only factors found to be significantly associated with favorable outcome.
In 35 patients, follow-up brain imaging scans (CT, n=21; MRI, n=14) at a median number of 2.5 (range, 1 to 350) days for CT and 2 (range, 1 to 238) days for MRI were available for review. One patient, who had large areas of hypodensity on admission CT, died shortly after admission, such that no follow-up scans were performed. In this patient, areas of hypodensity on baseline CT were considered infarcted.
Median percent final infarct relative to ipsilateral cortical MCA territory was 50% (range, 0% to 100%). A strong and statistically significant correlation was found between percent core and percent final infarct (ρ=0.63, P<0.0001) in contrast to a weak and nonsignificant negative correlation found between percent penumbra and percent final infarct (ρ=−0.29, P=0.077).
A subgroup analysis was conducted on the 21 patients in whom both recanalization data at 2 hours after angiography and outcome data were available. In this group of patients, favorable outcomes were observed in 1 of 10 patients (10%) who did not recanalize (TIMI grade 0 to 1) versus 5 of 11 patients (45%) who recanalized (TIMI grade 2 to 3) (P=0.14, Fisher exact test). In both patients who did recanalize and those who did not recanalize, mean percent core was higher and mean MCA CBF was lower in patients with unfavorable outcome than in patients with favorable outcome (Table 3). These differences were significant across the 4 categories of patients both for percent core and for mean MCA CBF. By contrast, no significant differences could be detected between these 4 categories when percent penumbra was substituted for percent core (Table 3).
In patients with acute MCA occlusion, the ischemic penumbra in the cortical MCA territory is consistently present and relatively constant within 6 hours of symptoms onset. We observed high individual variability in the amount of core, which is not explained by the different time intervals between symptom onset and CBF study. Rather, it likely reflects interindividual differences in the collateral blood supply, a major determinant of regional CBF in large-vessel occlusion.
The observation that, contrary to the highly variable core, penumbra is relatively constant may explain why the volume of core and not that of penumbra is a determinant of clinical outcome in acute MCA occlusion. Our subanalysis performed on patients with known vessel recanalization status suggests that this may apply to both patients who do and patients who do not recanalize. When a large core volume is present, any benefit of penumbral salvage through vessel recanalization may be obscured by the extensive deficit and propensity to complications attributable to the core. Therefore, recanalization therapy may preferentially be guided by diagnostic methods capable of rapidly and reliably identifying irreversible ischemia.
The importance of the core as a predictor of outcome in patients undergoing intravenous thrombolysis is underscored by studies using pretreatment MRI in patients with large-vessel occlusion, showing that clinical outcome correlated significantly with lesion volume on diffusion-weighted imaging (thought to approximate core) but not with the volume on perfusion/diffusion imaging mismatch (thought to approximate penumbra).16 Other authors17 have found that in patients studied acutely with CT perfusion in whom vessel recanalization was determined at 3 days, the ratio of penumbra to penumbra plus core (but not core or penumbra alone) was predictive of clinical outcome in patients who recanalized. In general, data regarding the importance of core versus penumbra in patients undergoing thrombolysis are limited because of different methodologies used and small sample sizes involved, such that further prospective studies are needed to clarify this issue.
Penumbral tissue of an extent similar to that reported in this study, present even beyond 6 hours, was reported by investigators using PET3,10,18,19 or MRI20–22 in acute stroke. When penumbra was estimated on the basis of perfusion/diffusion imaging mismatch on brain MRI, its presence was found to be highly correlated with the presence of large-vessel occlusion.21 We analyzed only patients with same vascular occlusion site to eliminate an important confounding factor in the assessment of penumbral extent. As such, our results may apply only to a selected group of patients. The high degree of correlation between the presence of penumbra and the presence of large-vessel occlusion may explain why other studies,11 which do not include data regarding vessel patency, failed to confirm the existence of significant penumbral areas in acute stroke.
Kaufmann et al,12 using quantitative CBF assessment with Xe-CT, reported that in 20 patients with large-vessel occlusion studied within 6 hours, ipsilateral penumbra was negligible. Several methodological differences between the analysis of Kaufmann et al and ours may account for these discrepant conclusions. (1) Kaufmann et al inferred that the “real” penumbra represents the difference between mean percentages of brain perfused at penumbral flow levels in the ipsilateral versus the contralateral hemisphere. Since this difference was small, Kaufmann et al concluded that penumbra was small. However, several PET,23 Xe-CT,24 and Xe-133 studies25 as well as the data presented in this report (Table 2) have shown that the contralateral cortical regional CBF is in fact commonly significantly reduced in large hemispheric strokes. Therefore, ipsilateral cortical CBF should be analyzed independent of contralateral CBF. (2) The analysis of Kaufmann et al included nonischemic territories (anterior and posterior cerebral arteries) as well as subcortical structures, resulting in a significant increase in the proportion of white matter, for which the perfusion thresholds that define core and penumbra are unknown. (3) Finally, the extent of penumbra may have been underestimated by restricting the analysis to only 1 axial level, that passing through the basal ganglia, and excluding areas of the cortical MCA territory that are more prone to receiving pial collaterals.
Analysis of Xe-CT data at voxel level may be prone to measurement errors due to system noise and filtering.26 We believe, however, that the results of our voxel-based analysis are valid. While the measurement error of a single voxel may be 100%, the error in measuring 100 contiguous voxels is approximately 10%.14 Because in all of our 36 patients the voxels falling within each defined flow range were clustered (Figure 1D, 1E, and 1⇑F) rather than randomly scattered, the degree of measurement error is closer to 10%. The validity of our voxel-based analysis is supported by the high degree of correlation between percent core, percent penumbra, and percent NC/NP obtained through voxel-based analysis and those values obtained through measurement of mean CBF in cortical ROIs (Table 2). The latter has been validated as a reliable quantitative CBF measurement tool.14,26
This study has several limitations, mainly derived from its retrospective design. Patients were included in our study according to criteria that were meant to select a patient population with the same vascular occlusion site who underwent reliable quantitative CBF assessment within a time frame relevant for thrombolytic therapy. The lack of patient inclusion according to a prospectively established protocol introduces the possibility of selection bias. Xe-CT CBF studies were ordered during the specified study dates by all physicians involved as the standard of care for acute stroke. It is therefore unlikely that a significant number of patients were omitted from our study because of different Xe-CT CBF ordering patterns. The outcome assessment may be confounded by several factors. Outcome was assessed in a retrospective fashion, at different times, and not in a blinded manner. Different treatment modalities were used, and some patients were lost to follow-up. However, the results of our clinical-based outcome analysis are paralleled by the imaging-based outcome analysis in which percent core and not percent penumbra was strongly correlated with percent final infarct.
Another limitation is that the flow thresholds chosen to define core and penumbra have not been validated as true indicators of tissue viability by follow-up imaging or (ideally) histological studies. Nevertheless, numerous animal and human studies, including studies performed with Xe-CT CBF assessment, PET, and H2 clearance, have confirmed the existence of perfusion thresholds in cerebral ischemia.3,27,28 The exact CBF thresholds that define core or penumbra are not known, and, given their presumed dependence on time from stroke onset and on other variables (eg, body temperature, blood glucose levels, genetic factors), it is unlikely that they are defined by a fixed CBF value. Our values chosen for these thresholds correspond to those of 8 mg/100 mL per minute for core and 20 mg/100 mL per minute for penumbra, which were recently established by human PET studies.3,10,18,29
- Received May 21, 2002.
- Revision received May 13, 2003.
- Accepted May 27, 2003.
Furlan A, Higashida R, Wechsler L, Gent M, Rowley H, Kase C, Pessin M, Ahuja A, Callahan F, Clark WM, et al. Intra-arterial prourokinase for acute ischemic stroke: the PROACT II study: a randomized controlled trial: Prolyse in Acute Cerebral Thromboembolism. JAMA. 1999; 282: 2003–2011.
Arnold M, Schroth G, Nedeltchev K, Loher T, Remonda L, Stepper F, Sturzenegger M, Mattle HP. Intra-arterial thrombolysis in 100 patients with acute stroke due to middle cerebral artery occlusion. Stroke. 2002; 33: 1828–1833.
Baron JC. Perfusion thresholds in human cerebral ischemia: historical perspective and therapeutic implications. Cerebrovasc Dis. 2001; 11: 2–8.
Hacke W, Kaste M, Fieschi C, Toni D, Lesaffre E, von Kummer R, Boysen G, Bluhmki E, Hoxter G, Mahagne MH, et al. Intravenous thrombolysis with recombinant tissue plasminogen activator for acute hemispheric stroke: the European Cooperative Acute Stroke Study (ECASS). JAMA. 1995; 274: 1017–1025.
Goldstein S, Yonas H, Gebel JM. Acute cerebral blood flow as a predictive physiologic marker for symptomatic hemorrhagic conversion and clinical herniation after thrombolytic therapy. Stroke. 2000; 31: 275.
Larrue V, von Kummer RR, Muller A, Bluhmki E. Risk factors for severe hemorrhagic transformation in ischemic stroke patients treated with recombinant tissue plasminogen activator: a secondary analysis of the European-Australasian Acute Stroke Study (ECASS II). Stroke. 2001; 32: 438–441.
Jovin T, Yonas H, Gebel J, Jungreis C, Horowitz M, Goldstein S, Wechsler L. In thrombolysis treated MCA occlusion, the extent of ischemic core predicts the development of life-threatening hemorrhage or edema in conjunction with vessel recanalization. Neurology. 2002; 58: A76.
Kilpatrick MM, Yonas H, Goldstein S, Kassam AB, Gebel JM Jr, Wechsler LR, Jungreis CA, Fukui MB. CT-based assessment of acute stroke: CT, CT angiography, and xenon-enhanced CT cerebral blood flow. Stroke. 2001; 32: 2543–2549.
Marchal G, Beaudouin V, Rioux P, de la Sayette V, Le Doze F, Viader F, Derlon JM, Baron JC. Prolonged persistence of substantial volumes of potentially viable brain tissue after stroke: a correlative PET-CT study with voxel-based data analysis. Stroke. 1996; 27: 599–606.
Heiss WD, Kracht LW, Thiel A, Grond M, Pawlik G. Penumbral probability thresholds of cortical flumazenil binding and blood flow predicting tissue outcome in patients with cerebral ischaemia. Brain. 2001; 124: 20–29.
Kaufmann AM, Firlik AD, Fukui MB, Wechsler LR, Jungries CA, Yonas H. Ischemic core and penumbra in human stroke. Stroke. 1999; 30: 93–99.
Tatu L MT, Bogousslavsky J, Duvernoy H. Arterial territories of human brain. In: Bogousslavsky J, Caplan L, eds. Stroke Syndromes. Cambridge, UK: Cambridge University Press; 2001: 388–399.
Nighoghossian N, Hermier M, Adeleine P, Derex L, Dugor JF, Philippeau F, Ylmaz H, Honnorat J, Dardel P, Berthezene Y, et al. Baseline magnetic resonance imaging parameters and stroke outcome in patients treated by intravenous tissue plasminogen activator. Stroke. 2003; 34: 458–463.
Wintermark M, Reichhart M, Thiran JP, Maeder P, Chalaron M, Schnyder P, Bogousslavsky J, Meuli R. Prognostic accuracy of cerebral blood flow measurement by perfusion computed tomography, at the time of emergency room admission, in acute stroke patients. Ann Neurol. 2002; 51: 417–432.
Barber PA, Davis SM, Darby DG, Desmond PM, Gerraty RP, Yang Q, Jolley D, Donnan GA, Tress BM. Absent middle cerebral artery flow predicts the presence and evolution of the ischemic penumbra. Neurology. 1999; 52: 1125–1132.
Lagreze HL, Levine RL, Pedula KL, Nickles RJ, Sunderland JS, Rowe BR. Contralateral flow reduction in unilateral stroke: evidence for transhemispheric diaschisis. Stroke. 1987; 18: 882–886.
Slater R, Reivich M, Goldberg H, Banka R, Greenberg J. Diaschisis with cerebral infarction. Stroke. 1977; 8: 684–690.
Gur D, Good WF, Wolfson SK Jr, Yonas H, Shabason L. In vivo mapping of local cerebral blood flow by xenon-enhanced computed tomography. Science. 1982; 215: 1267–1268.
Marchal G, Benali K, Iglesias S, Viader F, Derlon JM, Baron JC. Voxel-based mapping of irreversible ischaemic damage with PET in acute stroke. Brain. 1999; 122: 2387–2400.