|Home | About | Journals | Submit | Contact Us | Français|
The most important service that imaging provides to patients with ischemic stroke is to rapidly identify those patients who are most likely to benefit from immediate treatment. This group includes patients who have severe neurological symptoms due to an occlusion of a major artery, and who are candidates for recanalization using intravenous thrombolysis or intra-arterial intervention to remove the occlusion. Outcomes for these patients are determined by symptom severity, the artery that is occluded, the size of the infarct at the time of presentation, and the effect of treatment. MRI provides key physiological information through MR angiography and diffusion MRI that has been proven to be of high clinical value in identify patients who are in need of immediate treatment. Perfusion MRI provides information about the ischemic penumbra, but its clinical value is unproven. In current clinical practice the time since stroke onset is dominant over physiologic information provided by MRI in treatment decisions. This will change only when clinical trials prove that stroke physiology as revealed by MRI is superior to time from stroke onset in promoting good clinical outcomes.
Ischemic stroke is the most common neurological cause of severe disability and death. MRI reveals stroke biology in real-time, and may benefit the clinical management of patients, especially in the most severe strokes. Yet when the time of stroke onset is known, it is time that is usually decisive in decisions to treat. This is because the clinical trials that led to current stroke treatment used time and noncontrast CT scans for patient inclusion,(1) and will change only after clinical trials demonstrate that MRI is useful to identify patients who will benefit from treatment.
To illustrate the present circumstances in the use of MRI in acute ischemic stroke, consider the imaging data from 2 patients (Figure 1). In the first case, a man presented with severe neurological symptoms but with an unknown time of stroke onset. Imaging revealed an occlusion of a major cerebral artery, but that only a small part of the brain had undergone irreversible damage. Based on the physiology revealed by imaging, intra-arterial removal of the occlusion was executed, and the patient made a full recovery. In the second case, a man presented with moderate symptoms, but unlike the first case, the time of stroke onset was known. Once again, a major artery was occluded, but only a small potion of the territory at risk was irreversibly injured at the time of initial imaging. The decision not to intervene was made because the patient fell outside the established time windows for treatment. The infarction grew accompanied by substantial worsening of symptoms. These cases illustrate the current dominance of time over information provided by imaging in stroke management. This review presents evidence on how MRI-revealed stroke biology may lead to better outcomes in spite of time of stroke onset. But, routine use of MRI to guide treatment requires verification in clinical trials.
The role of any clinical tool is to improve patient care. To understand how MRI may be most useful in improving patient outcomes in ischemic stroke, it is important to appreciate that this disease is not a single entity, but a collection of maladies with distinct natural histories. It is useful to categorize ischemic stroke patients into the 3 groups shown in Figure 2. The most important group includes patients with severe strokes due to occlusions of large cerebral arteries including the distal internal carotid artery (ICA), proximal middle cerebral artery (MCA) and basilar artery (BA). Such occlusions nearly always produce severe neurological deficits, and poor outcomes are likely if not treated. While this group comprise between 25% and 35% of all stroke patients, they account for the large majority of deaths due to stroke, and the majority of patients with poor outcomes. (2,3) The mortality rates in patients with distal ICA occlusions are reported at 41.7-53% (4-6); in those with proximal MCA occlusion are reported at 27-78% (7,8); and those with basilar occlusions are 92% (9).
Patients who have major artery occlusions may be most successfully treated (10) using catheter-based interventions (Table 1). Because the typical outcome for most of these patients is poor, it makes sense to target these patients: identify them rapidly, and intervene to convert as many as possible from major stroke patients into minor stroke patients. Thus, the single most important role for imaging in acute ischemic stroke is to identify these patients. The next most important role is to definitively determine that ischemia is the source of the neurological symptoms.
In patients who have occlusions of branch arteries, small regions of brain are at risk, and symptoms are typically mild or moderate. They generally do well regardless of treatment. The final group includes patients that have ‘lacunar’ infarcts. These are due to occlusions of small penetrating vessels, usually in the brain stem, deep gray matter, internal capsule or corona radiata. Presenting symptoms are variable, but most will also have good outcomes.
After the occlusion of a major cerebral artery two zones can be demarcated within the brain (Figure 3). Brain tissue immediately beyond the occlusion receives little or no flow and quickly undergoes infarction. This is known as the core of the infarct. Other parts of the brain, while still alive have abnormally low blood flow, and are in danger of proceeding to infarction. This region is known as the penumbra; it maintains viability through blood flow provided by the collateral circulation. (11) In major artery occlusions of the anterior circulation, the core and the penumbra are dependent parameters, and the relative size of each depends on the vigor of the collateral circulation. Arterial occlusion may be detected with MRA. The best current estimate of the infarct core is provided by diffusion MRI. The extent of the penumbra can be estimated using perfusion MRI. Subtle changes in the apparent diffusion coefficient of water may also provide an estimate of the penumbra. When there is an occlusion of a major anterior circulation artery, the core and penumbra are related by the collateral circulation. Therefore, knowledge of one permits a clinically relevant deduction of other. For example in a proximal MCA occlusion, if diffusion MRI reveals a small core, one can confidently deduce that the region of brain with poor blood flow is large, involving the remainder of the MCA territory. The neurological examination is also highly sensitive to abnormal blood flow to brain areas that produce symptoms. (12) However, the neurological exam cannot distinguish between the core and penumbra.
The relationships between cerebral blood flow, clinical symptoms and infarction have been elucidated in animal models and subsequently in human subjects. Figure 4 demonstrates the basic relationship between cerebral blood flow, time and infarction. This figure is derived from a study on awake rhesus macaques that had permanent or temporary MCA occlusions. (13) The major conclusions of this study was subsequently been reproduced in baboons and in humans using positron emission tomography (PET). A review of this PET literature may be found in reference 17 (11) The figure demonstrates that below a threshold level of cerebral blood flow (CBF), the animal develops paralysis. However, the presence of paralysis is not synonymous with infarction: if the CBF remains within the upper abnormal range, the brain remains viable and may, fully recover upon reperfusion. At lower rates of CBF, tissue will undergo infarction but the time required for irreversible injury is dependent on the specific CBF. At very low CBF levels, brain tissue of infarct within several minutes. Higher levels of CBF such as 15 mL/100 g tissue/minute infarction will not occur for two or more hours.
Diffusion MRI heralds the imminence of brain infarction due to ischemia which was first demonstrated in experimental animal models. (14-16) With severe reduction of CBF, a reduction in tissue water diffusion may be detected with MRI within 30 minutes. The sensitivity of diffusion MRI at the early stages of severe ischemia is illustrated in Figure 5, which shows images of a patient who developed severe left sided paralysis in the presence of an emergency department physician. If there is reperfusion shortly after the diffusion abnormality appears, it may reverse, and infarction will not occur. Clinically, spontaneous reversal is extremely rare. (17) It is also unusual with treatment. (18) In most cases, the reduction in CBF persists and infarction ensues. Later reperfusion or normalization of CBF may result in temporary normalization of the diffusion abnormality; however neuronal injury will still occur. (19) Reduced diffusivity of brain tissue detected by MRI is the most reliable early indicator of brain infarction that is widely available. (20,21)
The management of the acute ischemic stroke patient is driven by the concept of “time is brain”. In a celebrated paper (22) Saver quantified the rate of neuronal loss in stroke. By his calculations, an average of 1.8 million neurons is lost every minute that appropriate treatment is not given. However, the important issue of variability in the rate of neuronal loss amongst patients was not addressed. The rate of infarct growth may be very low as is illustrated in Figure 6. (23) In this individual, no significant expansion of the DWI lesion was observed over a four hour period.
Following the assumptions of Saver, an empirical assessment on the variability of neuronal loss amongst patients with occlusions of major arteries of the anterior circulation was performed by Hakimelahi et al. (11) The results derived from 68 consecutive patients are shown in Figure 7. The variability in the rate of neuronal loss was found to be very high, although interestingly, the average rate of neuronal loss derived empirically was similar to the theoretical average calculated by Saver. Since the occlusions in these patients involved similar, large territories at risk, the variability observed is best explained by differences in the adequacy of the collateral circulation. (24)
If there is a large variability in the vigor of the collateral circulations in patients in patients with major anterior circulation occlusions, then significant penumbras should be observed in patients even many hours after the onset of occlusion. Such observations have been reported by several investigators (25-27) including a study by Ribo et al (28), who reported that 43 of 56 stroke patients who presented 3 to 6 hours after stroke onset had a DWI/PWI mismatch (a biomarker of infarct core and the clinical penumbra) of 50% or greater. The largest study to date involved 109 patients with anterior circulation (29). All had undergone diffusion/perfusion MRI within 24 hours of stroke onset, and more than 50 percent had DWI/MTT mismatch volumes of 160% or greater (Figure 8). This was most common among patients with occlusions involving the distal internal carotid artery and/or the proximal middle cerebral artery. There was no difference between patients who were scanned at different times after ictus. Persistence of mismatch after 9 hours was common in patients with proximal occlusions of the anterior circulation.
The observation of a large penumbra in patients scanned many hours after stroke onset suggests that for a significant number of patients the ischemic penumbra, while dynamic, changes at a very slow rate and may be stable for a long period of time. For practical reasons, few longitudinal studies have been performed to evaluate the rate of change of the penumbra over first few hours and days. But as demonstrated in a study (30) in which patients underwent a diffusion-perfusion imaging at study entry, at 4 hours and at 24 hours after the initial study, as well as a follow one week later suggests that the clinical ischemic penumbra may be stable for a subset of patients. Fourteen patients with ICA and/or proximal MCA occlusion and a diffusion/perfusion mismatch at presentation were evaluated. When baseline and 4 hour scans were compared, there was no significant interval change in the mean abnormal DWI volume. Figure 9 is a graph of the mean abnormal DWI volumes for the 3 time points and mean final infarct volumes.
Taken together, the observation of a large proportion of patients with a significant clinical ischemic penumbra many hours post stroke onset, the stability of infarcts in patients with excellent collaterals, and the large variability in neuronal loss reinforces the arguments put forth by Baron and others (31) that strict reliance on time alone in the management of patients with ischemic stroke is ill considered. The evidence also suggests that attention to physiology may allow a significant number of patients to undergo treatment of a major artery occlusion even many hours after the onset of symptoms.
The most important factors that determine the outcomes of patients with acute ischemic stroke include the severity of neurological symptoms, the artery that is occluded, the size of the infarct, and whether treatment successfully keeps the infarct small. Reliable and validated instruments to assess the severity of neurological symptoms have been developed. The most widely employed and validated is the NIH stroke scale (NIHSS). (12) Using this instrument, the trained examiner rapidly evaluates the neurological status of 11 areas that involve motor function, sensory deficits and cognition. (32)The more severe the symptoms, the higher NIHSS score. Patients with an NIHSS of between 0 and 5 have mild symptoms. These patients typically recover fully, and in most centers they are not considered thrombolytic therapy such as tPA (tissue plasminogen activator). Patients with an NIHSS of 10 or greater have severe neurological deficits. These patients have a high likelihood of poor outcomes such as death or dependency if they are not successfully treated.
The specific artery that is occluded is the cause of symptom severity and is a major determinant outcome in stroke. In the absence of treatment, the worst outcomes occur in patients who have occlusions of the distal internal carotid artery, proximal middle cerebral artery and the basilar artery. The Boston Acute Stroke Imaging Scale (BASIS) is a simple dichotomized classification instrument that is based on whether or not one of these vessels are occluded or there is infarction in the vascular distribution of these vessels at the time of presentation. (2,3) Studies using this classification instrument in which MRI and/or CT were employed for classification purposes have demonstrated that between 25-35% of patients have occlusions or evidence of occlusions involving one of these major arteries. Such patients, classified as BASIS+, account for nearly all of the in-hospital deaths. In addition over 70% of these patients are discharged to rehabilitation centers. On the other hand, over 70% of BASIS− patients are discharged directly to home. In a multicenter, prospective study in which CTA was employed as part of the routine imaging evaluation of acute stroke patients, basilar and internal carotid terminus occlusions, in addition to National Institutes of Health Stroke Scale and age, independently predicted outcome. (33)The finding that major artery occlusion is independent of the NIH stroke scale in predicting outcomes indicates that the clinical examination and arterial occlusion provide complementary information, and that both are useful in predicting the clinical course of the patient.
The amount of brain tissue that has undergone a reversible injury at the time of presentation is a major determinant of clinical outcomes in acute stroke. It is now well-established that diffusion MRI the best method for identifying the core of the infarct at the early stages of ischemic stroke. (21,34-36) There has been research in identifying the volume of infarct produced by an anterior circulation stroke that predicts a poor outcome regardless of intervention. Clinical tradition holds that infarcts involving approximately 1/3 of the MCA territory (~100 mL) or greater have high likelihood of poor outcomes. More recently, data is accumulating that the volume that is specific for a poor outcome is likely lower than this. Figure 10 was derived from a study of a series of patients who underwent diffusion MRI the time of admission. (37) Many of the patients received therapy including intravenous tPA or/and catheter-based intervention to open the occluded artery. It was found that all patients who had diffusion abnormality of greater than 72 mL had a poor outcome regardless of treatment. A separate study by a different group found a similar volume threshold. (38) Of course, more research is needed because, for example, it is likely that there will be different thresholds for the right and left hemispheres.
The final major factor that determines the outcomes in patients with ischemic stroke is whether or not a patient is successfully treated. The only proven treatments in ischemic stroke are agents that reopen the occluded artery (Table 1). There is a strong correlation between recanalization and outcome, and recanalization is strongly associated with improved functional outcomes and reduced mortality. (39) The spontaneous recanalization rate is ~20%, and some individuals are fortunate if the collateral circulation maintains tissue viability until the artery reopens spontaneously. Treatments for the removal of the occlusion include intravenous and intra-arterial administration of thrombolytic agents, and mechanical recanalization.
It is logical to assume that a higher proportion of good outcomes should accompany a higher rate of recanalization. This is not borne out in published clinical trials because of an important additional factor: the amount of brain already irreversibly injured at the time of recanalization. This was demonstrated by Yoo et al. (40)who showed that patients undergoing intra-arterial therapy had poor outcomes even with successful treatment if the volume of abnormal brain on DWI exceeded 70 ml. The variable dynamics of infarct growth due to differences in collateral circulation renders time from stroke onset an unreliable guide for stroke treatment aimed at alleviation of arterial occlusion. Using DWI to identify the volume of irreversibly injured brain to guide treatment is a viable alternative, but clinical trials are necessary.
The importance of identifying a major artery occlusion, which may be done with MRA,(41) and an estimate of the infarct core with DWI in acute ischemic stroke are strongly supported by the available evidence. There is much weaker evidence to support the value of perfusion imaging with MRI (or with CT) in the management of acute stroke patients. (21) The most widely employed MRI-based measure of potential for infarct growth is the diffusion/perfusion mismatch. Perfusion measurements by MRI (also CT) have many problems, most notably the lack of standards for collecting and analyzing the perfusion data. There is also measurement error that may be very large. (42) The reasons for the inadequacy of diffusion/perfusion mismatch for patient selection in ischemic stroke treatment trials were reviewed by Schaebitz.(43) The potential benefits and limitations of perfusion MRI in specific clinical circumstances are considered next.
The diffusion/perfusion mismatch has been used to identify patients that are likely to benefit from thrombolytic therapy, but are outside the traditional time window for such treatment. (44-48) Based on positive outcomes in preliminary studies, a series of clinical trials were undertaken to use this method to select patients for treatments using intravenously administered tPA or desmoteplase, a new thrombolytic agent. A meta-analysis of the DEDAS, DIAS, DIAS II, DEFUSE and EPITHET trials included 502 patients who were selected based on a diffusion/perfusion and treated with intravenous thrombolysis or placebo beyond 3 hours. (49). The meta-analysis found no significant improvement in favorable clinical outcome by thrombolytic therapy in these trials.
Important factors that were not considered for inclusion/exclusion criteria in these clinical trials were the presence of a therapeutic target (a vascular occlusion documented by MRA or CTA), and the volumes of the diffusion and perfusion abnormalities. Not screening out patients with DWI volumes of greater than 70ml would have included patients that even if treated successfully would not have a good outcome. (40) Not screening out patients with small abnormal perfusion volumes would have included patients that would have a good outcome regardless of treatment. (37) The value of identifying a target occlusion has been recognized by investigators pursuing clinical trials with desmoteplase, and is implemented in an ongoing phase III clinical trial of this thrombolytic agent (Efficacy and Safety Study of Desmoteplase to Treat Acute Ischemic Stroke (DIAS-4); ClinicalTrials.gov identifier NCT00856661). In this randomized, double-blind, parallel-group placebo-controlled study, treatment can be initiated within 3 to 9 hours after the onset of stroke symptoms in patients with NIHSS Score of 4 to 24 and vessel occlusion or high-grade stenosis on MRI or CTA in proximal cerebral arteries.
As emphasized previously (Figure 2) the most important role for imaging in acute ischemic stroke is to rapidly identify patients with severe stroke syndromes that are most likely to benefit from intervention (Table 1). However, these procedures require significant resources in people and equipment, and should be limited to those who are most likely to benefit. The evidence supports proceeding to endovascular intervention when there is a significant neurological deficit, the presence of an occlusion of the major cerebral artery and a small core infarct (less than approximately 70 to 100 ml).
There is little evidence to indicate that perfusion MRI provides significant additional information in this clinical scenario. The most important reason is that the core and remaining tissue that has low blood flow are not independent parameters. The relative sizes of the core and region of abnormal perfusion are determined by the adequacy of the collateral circulation. In the presence of robust collateral circulation, the diffusion abnormality will be small and the volume of poor perfusion will be large. Conversely, if the diffusion abnormality is large, the perfusion abnormality will be small, and it is unlikely that such patients will benefit from recanalization. (24)
The linkage between the diffusion and the perfusion abnormalities by the collateral circulation suggests a simple way to identify patients who have a large mismatch without performing a perfusion MRI study. This is demonstrated in Figure 7. All patients with a diffusion abnormality volume of less than 70 ml had perfusion deficit of at least 100% greater than the diffusion lesion volume. (11,50)Physiological understanding of cerebral blood flow suggests, and the evidence supports the principle that in occlusions of major anterior circulation arteries knowledge of the diffusion abnormality volume is sufficient to deduce the approximate size of perfusion deficit. Another approach to estimate the penumbra is to further analyze the diffusion data. Reduction of blood flow will subtly decrease the apparent diffusion coefficient of water in the region surrounding the core infarct. This may be used to estimate the penumbra. (Figure 11) (51)
There is one scenario in which perfusion MRI may be of clinical value: patients that have minor neurological deficits despite a major artery occlusion. These patients will have small DWI abnormalities nearly without exception. The mild symptoms indicate that there is excellent blood flow to areas of brain beyond the occlusion because of a robust collateral circulation. Perfusion MRI demonstrating a small volume of brain with low perfusion would confirm the presence of a vigorous collateral circulation. This concept is supported by a study in which all patients with an MTT volume of <47 mL (16 of 54) and an NIHSS score <8 (17 of 54) had good outcomes. (37)
Perfusion MRI may be also be of value in patients with anterior circulation ischemic strokes due to occlusions of minor branch arteries. While this specific question has not been directly tested directly, the logic is compelling. These patients commonly present with minor neurological symptoms with NIH stroke scale values of less than 10. The major cerebral arteries are patent, and a small branch occlusion may be not be easily identified by MR or CT angiography. The value of perfusion MRI would be to confirm ischemia as the cause of the symptoms, rather than a stroke mimic such as seizure. (17,52,53) The certain identification of ischemia as the cause of a neurological abnormality has major implications for the management of such a patient, most importantly in the institution of treatment to prevent another, possibly major stroke. Additionally, if the perfusion abnormality is significantly larger than any diffusion abnormality that is identified, IV thrombolytic therapy is potentially useful, but this will require further research.
There is little evidence that perfusion MRI is of value in assessing patients with posterior circulation strokes. The most devastating of these involve occlusion of the basilar artery, which may produce infarctions of the brainstem, cerebellum and thalamus. Even if they are small in size, symptoms may be major with a high likelihood of poor outcomes. (9) Occlusion of the basilar artery is such a devastating process that an attempt at opening this artery is often made regardless of the time stroke onset. Such procedures can be successful because the collateral circulation may be particularly permissive, and may maintain tissue viability for long periods despite severe symptoms. Data on the value of perfusion MRI in basilar artery occlusions are rare. One likely reason is that it is difficult to acquire good quality images of the posterior fossa due to magnetic susceptibility artifacts.
MRI provides physiologic information in acute ischemic stroke that is clearly superior to computed tomography (CT). Nonetheless, CT remains the most commonly employed neuroimaging method in the rapid evaluation of the ischemic stroke patient. One reason is the convenient availability of CT scanners for patients being evaluated for a sudden neurological problem. But the most important reason is that CT is recommended for patients being considered for treatment with IV tPA, (54) a recommendation based on clinical trials including the pivotal NINDS IV tPA clinical trial. (1) The use of MRI in place of CT for decisions regarding the use of IV thrombolytics requires validation in clinical trials. Moreover, CT also has proven clinical efficacy in identifying hemorrhage, which is an important possible cause of the presenting symptoms. While MRI has been shown to be more sensitive than CT in the detection of small hemorrhages, (55) this has not been shown to be of clinical relevance in acute stroke. Additionally, MRI may not detect early subarachnoid hemorrhage because of the high oxygen tension of the CSF. (55)
There have been significant efforts in developing CT perfusion methods to substitute for DWI in the detection of early infarct core. (56) While high correlations between lesions observed on diffusion MRI and perfusion CT have been reported, there are important theoretical and methodological limitations that make perfusion CT methods unreliable for determination of the core. A detailed review of this issue is beyond the scope of the present review, but it is considered in some detail in reference 11. (11) With respect to the ischemic penumbra, CT perfusion methods are similar to MR perfusion, with the exception that coverage of the head is usually more limited. However, as discussed previously, the role of perfusion MRI in the assessment of patients with major anterior circulation stroke is limited. The same limitations also hold for CT perfusion.
The Neuroradiology Division at our hospital has adopted the recommended acute stroke imaging algorithm shown in Figure 12. This algorithm was adopted after a formal review of the data collected from our acute stroke patients in the past 15 years as well as a critical review of the literature. Much of the reasoning that led to this algorithm has been presented here. The emergency department at our hospital is fortunate to have available CT and MRI at all times, and we employ both in acute stroke patients. The algorithm is explained in the figure legend, and reflects the ready availability of both technologies. It is practical and optimized for the rapid assessment of the large number of patients with neurological symptoms due to a variety of causes. How this algorithm might be applied to other stroke care programs will depend on local conditions.
Acute ischemic stroke is a heterogeneous disease, with major stroke caused by proximal artery occlusions representing the stroke subtype with the most devastating outcomes. Because there are effective treatments for major strokes, the primary role of imaging is to identify these patients rapidly and accurately. MRI is particularly powerful in depicting the most important relevant physiology in acute ischemic stroke: the occlusion site and the size of the infarct core. Patient outcomes are related to the severity of the symptoms, as measured by the NIH stroke scale, the site of occlusion, the size of the core of the infarct at the time of presentation, and the success of the treatment. In present clinical practice, time is dominant over physiology for decisions related to implementing therapy. There is evidence that suggests that the physiology revealed by MRI may be an improvement over time in clinical decisions for treatment of patients with major strokes. However, this will require clinical trials to demonstrate the value of using MRI in improving outcomes when it is used to select patients for treatment.
Supported in part by: NIH grants NS050041, NS051343