Ischemic Core and Penumbra in Human Stroke
To the Editor:
We would like to dispute the conclusion by Kaufmann et al1 that very little penumbra is to be found as early as 1 hour after onset of middle cerebral artery (MCA) territory ischemic stroke, a conclusion which, if true, would have obvious negative implications for acute stroke management. These authors mapped cerebral blood flow by means of the stable xenon-enhanced CT (XeCT) method in 20 patients studied 60 to 360 minutes after clinical onset of stroke due to MCA occlusion. In support of their conclusion, they offer the following pieces of evidence: (1) a one-to-one correlation between measured volumes of final infarction (estimated by the lucency in late CT scans) and of tissue with flow of <6 (or 10) mL/100 g per minute, taken therefore to represent the core of irreversible damage; and (2) a lack of rim of tissue with flow between 11 and 20 mL/100 g per minute (ie, the classical penumbra range) surrounding the core. Their conclusion conflicts with considerable evidence from earlier positron emission tomography (PET) studies by several groups for a prolonged persistence of substantial amounts of penumbral tissue in a subgroup of patients.2 3 4 5 6 7 For instance, in a series of patients studied 7 to 16 hours after onset, brain tissue with characteristics suggestive of penumbra (ie, with both flow in the ≈7 to 22 mL/100 g per minute range and very high oxygen extraction fraction) occupied up to 52% of the final infarct volume2 and represented a volume of up to 25 mL in the border of the ultimate infarct.3 Recently, tissue with perfusional characteristics also compatible with penumbra was shown to be salvaged from necrosis by intravenous thrombolysis performed within 3 hours of clinical onset7 ; furthermore, the amount of tissue so salvaged correlated with measured neurological recovery,7 consistent with earlier findings by Furlan et al3 that spontaneous neurological outcome is partly explained by the volume of penumbra that escapes infarction. Also, it would be difficult to explain the results of the National Institute of Neurological Disorders and Stroke study8 showing beneficial effects of thrombolysis up to 3 hours after onset (ie, in the 90- to 180-minute patient subgroup) if, as stated by Kaufmann et al based on their findings, “strategies to improve the outcome of many patients with acute MCA occlusion must … include interventions to reverse the ischemic process within a few minutes of onset.”
We believe several limitations due to both their retrospective study and the methodology used, some of which rightly acknowledged, may account for their surprising findings. First, regarding the correlation observed with infarct volume, the latter was not optimally assessed, because it was estimated from CT scans obtained 2 to 6 days after stroke in 6 of 12 patients, when edema may considerably blur the borders of the infarct, and between 7 and 14 days in 3 others, ie, a time period where the so-called fogging effect9 10 might obfuscate the infarct and result in gross volume underestimation. These problems may be appreciated from inspection of their Figure 2 (bottom row). Also, the image analysis was performed on 1 axial brain slice only, that passing through the basal ganglia, rather than across the whole brain volume as advocated,2 3 7 so that large portions of the penumbra, which tends to reside in the periphery rather than near the center of the MCA territory, may have been overlooked. Several patients were excluded post hoc from the cerebral blood flow maps analysis because of suspicion of large infarct on admission CT and excluded from the correlation with CT data because of lack of radiological follow-up or complications from intra-arterial thrombolysis. The CT cut onto which the flow maps were superimposed was not coregistered with the early XeCT cuts; instead, the “1 cm slice image most closely corresponding to the XeCT middle level was selected.” In 1 patient, the volume of tissue with flow of <6 mL/100 g per minute underestimated the final volume of infarction by a factor of 2, consistent with a large penumbra being present at least in this particular case, and indicating that the prediction of infarct volume with this flow threshold can be erroneous at the single-case level. In a recent voxel-based analysis of 19 patients with MCA territory stroke studied 5 to 18 hours after onset (G. Marchal, K. Benali, S. Iglesias, F. Viader, J.-M. Derlon, and J.-C. Baron, unpublished data, 1999), we found that the volume of tissue (across the whole brain) that was below a flow value of 8.4 mL/100 g per minute (determined as a probable threshold for irreversible damage at P<0.05) highly positively (P<0.001) correlated with final infarct volume (determined on coregistered CT obtained later than 14 days after onset), similar to the findings of Kaufmann et al, but—at variance with these authors—underestimated that volume by a factor of 2 on average, consistent with earlier evidence that the penumbra partly evolves to necrosis after the PET study.2 Second, Kaufmann et al analyzed their flow maps on a voxel-by-voxel basis, which is well advised in studies of the penumbra,2 3 7 but the accuracy of the XeCT method for measuring cerebral blood flow at the voxel level is questionable because of poor signal-to-noise ratio11 12 ; thus, the error in flow estimate in a single 1×1×10 mm3 voxel may be as high as 100%.12 In voxel-based PET studies of the penumbra, a procedure of voxel averaging was advocated to improve the signal-to-noise ratio.2 3 To our knowledge, the XeCT method has not been validated against a gold standard at the voxel level and was found to deviate from 133Xe single-photon emission tomography–determined cerebral blood flow already with large regions-of-interest.13 In addition, the duration of stable xenon inhalation in the study of Kaufmann et al1 was only 4.3 minutes, which may result in erroneous estimates of the blood-tissue partition coefficient, especially in small voxels, and, as a result, potentially also of flow.11
Thus, it is possible that the method is insensitive or unreliable to detect voxels with flow in the penumbral range or that it underestimates flow values such that voxels with values of <10 mL/100 g per minute might effectively be penumbral (see patient 6 of their Figure 2). All the above methodological issues need to be acknowledged before the conclusions of Kaufmann et al1 may be considered valid.
- Copyright © 1999 by American Heart Association
Kaufmann AM, Firlik AD, Fukui MB, Wechsler LR, Jungries CA, Yonas H. Ischemic core and penumbra in human stroke. Stroke. 1999;30:93–99.
Marchal G, Beaudouin V, Rioux P, de la Sayette V, Le Doze F, Viader F, Derlon J-M, Baron J-C. 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.
Wise RJS, Bernardi S, Frackowiak RSJ, Legg NJ, Jones T. Serial observations on the pathophysiology of acute stroke: the transition from ischaemia to infarction as reflected in regional oxygen extraction. Brain. 1983;106:197–222.
Heiss WD, Grond M, Thiel A, Von Stockhausen HM, Rudolf J, Ghaemi M, Löttgen J, Stenzel C, Pawlik G. Tissue at risk of infarction rescued by early reperfusion: a positron emission tomography study in systemic recombinant tissue plasminogen activator thrombolysis of acute stroke. J Cereb Blood Flow Metab. 1998;18:1298–1307.
National Institute of Neurological Disorders and Stroke rt-PA Stroke Study Group. Tissue plasminogen activator for acute ischemic stroke. N Engl J Med. 1995;24:1582–1587.
Matsuda M, Lee H, Kuribayashi K, Yoshimura M, Honda T, Handa J. Comparative study of regional cerebral blood flow values measured by XeCT and XeSPECT. Acta Neurol Scand. 1996;93(suppl 166):13–16.
We thank Drs Baron and Marchal for their thoughtful and appropriately skeptical letter in response to our article that examined cerebral blood flow levels within and about acute MCA territorial infarctions. We acknowledge that the response was appropriately skeptical because the conclusions have far-reaching implications and because at first examination they appear to conflict with an established body of work that has placed the penumbra as the goal of modern stroke therapy.
We agree that this was a preliminary and limited examination of a very complex question. The conclusion that many patients with acute MCA occlusion have a severe ischemic insult that for the most part is irreversible within a few hours of onset is, however, probably valid. The concept proposed by Simone of a cortical infarction surrounded by a broad band of penumbral tissue has not been consistently validated in subsequent histological or many physiological studies.R1 Although Baron and Marchal appropriately discuss some of the articles that have examined the complex metabolic changes that occur in the 5 to 18 hours subsequent to cortical infarction, other studiesR2 R3 have supported the hypothesis that for many patients the tissue volume at risk is irreversibly injured within a few hours of onset. If the volume enlarges it may be due to many mechanisms, only one of which is inclusion of the penumbra into the core. One mechanism that has not been adequately discussed is fluid accumulation within the core that causes secondary compression of surrounding tissues and is an acknowledged cause of secondary insult after massive MCA infarctions.
In response to the concerns raised by Baron and Marchal concerning the methodology used, we agree that 30-day follow-up CT studies would have been a preferable comparator. More levels of study would have been desirable as well. We are currently beginning a far larger prospective acute stroke study that should be able to address the above issues. In regard to the concern raised by a voxel-based analysis of the stable xenon cerebral blood flow database, although Baron and Marchal are correct in noting that there may exist an error of 100% in the measurement of flow within 1 voxel measuring 1×1×10 mm3, the error in measuring 120 contiguous voxels of this size is closer to 10%.R4 The data presented in our article did not present single voxel data but instead analyzed volumes of tissue that contained contiguous voxels that fell within defined flow ranges. A stable xenon cerebral blood flow study appears to provide access to good quantitative data because each cerebral blood flow image consists of over 24 000 flow values. This very large database has therefore made it possible to examine and report the flow values within only large numbers of contiguous voxels. Consequently, the measurement of flow values near zero for 120 to 24 000 contiguous voxels on multiple brain levels has proven predictive of irreversible ischemia.R5 R6 R7 In regard to the signal-to-noise question, the noise level of modern CT scanners is ≤1 Hounsfield unit, and the signal that we have sought as the minimal adequate for quantitative flow calculations has been 8. This initially required the inhalation of 32% to 33% xenon; however, with the development of newer CT scanners with 20% lower noise levels, 27% to 28% has been able to provide the same high signal-to-noise levels with even fewer side effects.
The brief inhalation period appropriately raised concern by Baron and Marchal that the measurement of blood-tissue partition coefficient may be inaccurate and lead to inaccurate flow values. Although very accurate lambda maps can be generated by longer periods of xenon inhalation, a shorter duration of inhalation was chosen to minimize the sensorial side effects of xenon inhalation as well as to minimize patient motion. This study format has led to a clinically acceptable study with good clinical data obtained in over 90% of over 10 000 routine clinical studies. In fact, the errors of lambda (portion coefficient) and K (the flow constant) are inverse, so that the calculation of flow that is the multiple of λ and K remains remarkably stable.R3 This fact, combined with the observation that a brief inhalation period combined with early data acquisition also minimizes the inaccuracies that could be caused by flow activation due to xenon inhalation,R8 R9 combine to make the flow values generated by stable XeCT uncommonly accurate.R10 R11 R12 R13 Although information contributed by λ are not optimum with our current format, we believe that inclusion of this variable does provide greater accuracy of flow values even in abnormal brain regions.
We disagree with the statement that our conclusions have negative impact on acute stroke management. From an examination of a consecutive series of 53 acute hemispheric stroke patients studied with XeCT within 6 hours of stroke onset, we have observed a significant group of patients that did have a significant volume of the MCA territory with flow levels within the penumbral range. Rather than being within the rim around a core of irreversible ischemia, they involved entire vascular territories distal to MCA occlusions.R14 Thrombolysis in the presence of penumbral levels of distal MCA flow has been a rewarding experience, and conversely, thrombolysis despite flow levels in the range of irreversible ischemia has been consistently unrewarding.R15 Therefore, cerebral blood flow studies in acute stroke may identify areas of potentially reversible ischemia representing the therapeutic target, and ultimately identify patients most likely to benefit from thrombolysis interventions.
Lassen NA, Astrup J. Ischemic penumbra. In: Wood JH, ed. Cerebral Blood Flow: Physiologic and Clinical Aspects. New York, NY: McGraw-Hill Book Co; 1987;458–466.
Heiss WD, Grond M, Thiel A, Ghaemi M, Sobesky J, Rudolf J, Bauer B, Wienhard K. Permanent cortical damage detected by flumazenil positron emission tomography in acute stroke. Stroke. 1998;29:454–461.
Heiss WD, Graf R, Fujita T, Ohta K, Bauer B, Lottgen J, Wienhard K. Early detection of irreversibly damaged ischemic tissue by flumazenil positron emission tomography in cats. Stroke. 1997;28:2045–2051. comment discussion 2051–2052, 1997 Oct.
Yonas H, Gur D, Claassen D, Wolfson SK Jr, Moossy J. Stable xenon enhanced computed tomography in the study of clinical and pathologic correlates of focal ischemia in baboons. Stroke. 1988;19:228–238.
Good W, Gur D. Xenon-enhanced CT of the brain: effect of flow activation on derived cerebral blood flow measurements. AJNR Am J Neuroradiol. 1991;12:83–85.
Wolfson SK Jr, Clark J, Greenberg JH, Gur D, Yonas H, Brenner RP, Cook EE, Lordeon PA. Xenon-enhanced computed tomography compared with [14C]-iodoantipyrine for normal and low cerebral blood flow states in baboons. Stroke. 1990;21:751–757.
Yonas H, Obrist W, Gur D. Cross correlation of CBF derived by 133Xe and Xe/CT in normal volunteers. J Cereb Blood Flow Metab. 1989;9(suppl 1):S409. Abstract.
Yonas H, Darby JM, Marks EC, Durham SR, Maxwell C. CBF measurement by XeCT: approach to analysis and normal values. J Cereb Blood Flow Metab. 1991;11:716–725.
Rubin G, Firlik AD, Yonas H, Wechsler LR, Jungreis CA. Relationship between cerebral blood flow and early clinical outcome in acute ischemic stroke. Stroke. 1998;29:311. Abstract.
Rubin G, Firlik AD, Pindzola RR, Levy EI, Yonas H. The effect of reperfusion therapy on cerebral blood flow in acute stroke. J Stroke Cerebrovasc Dis. 1999;8:1–9.