Relation between cerebral blood flow and neurologic deficit resolution in acute ischemic stroke

Critical to the development of cerebral infarction is the depth and duration of the ischemic insult. The goal of emergency stroke interventions, therefore, is to restore the cerebral circulation and protect the ischemic brain as rapidly as possible, before permanent ischemic injury occurs. Although this strategy of early intervention by thrombolysis using intravenous tissue plasminogen activator (tPA), for example, has shown some benefit for populations of patients, hemorrhagic complications remain problematic, and specific benefits to individuals or subgroups of patients have been difficult to demonstrate.^1,2 The ability to predict which patients will have a good outcome without treatment would allow these patients to be spared interventions that have little potential for benefit but still carry risks and unnecessary expense. Unfortunately, in the early phases of a cerebral ischemic event, clinical or radiographic distinction between patients whose deficits will later resolve spontaneously and those who will progress to infarction remains difficult.^3,4 This study explored the role of cerebral blood flow (CBF) measurements in the detection of patients who will have a rapid recovery to a normal neurologic examination before any clinical distinction could be made. The hypothesis was that patients who made a rapid clinical recovery without treatment had an early, spontaneous lysis of their thromboembolic occlusion and therefore would have normal CBF.

Methods. Patient population. Fifty-three patients who were admitted to the University of Pittsburgh Medical Center with the clinical diagnosis of acute ischemic stroke were retrospectively studied. Patients were included if they had acute hemispheric stroke symptoms consisting of hemiparesis or hemiplegia in combination with ipsilateral sensory loss or neglect, aphasia, or visual field deficts; had a CT scan that excluded a hemorrhagic etiology; and underwent a xenon-enhanced CT(XeCT) scan within 8 hours of symptom onset and before initiation of any form of stroke treatment. Excluded were patients with lacunar syndrome symptoms, events referable to the vertebrobasilar system, monocular visual complaints, or resolving deficits. Patients varied in age from 30 to 89 years (mean 66 years); 27 were men (51%) and 26 were women (49%). The mean NIH stroke scale on admission was 11.6 (range 3 to 21).

All patients were admitted to the hospital and were managed by the faculty of the University of Pittsburgh Medical Center Stroke Institute. Treatment of these patients included heparin alone in 24 patients (45%), intravenous tPA with or without heparin in 6 patients (11%), intra-arterial urokinase in 14 patients (26%), and lubeluzole in 4 patients (8%). Five patients (10%) were managed without treatment with anticoagulation, thrombolytic agents, or cerebral protective agents.

Xenon-enhanced CT cerebral blood flow testing. All patients underwent XeCT scans within 8 hours from the onset of symptoms, varying between 20 minutes after symptom onset to 7 hours, 30 minutes. The mean time to XeCT scan was 3 hours, 51 minutes. In all cases, XeCT scans were obtained immediately after initial CT scans obtained on patient arrival, before any form of stroke treatment. Quantitative CBF studies were performed on standard CT scanners to which an independent system for xenon delivery and CBF calculation were added (Xe/CT System, Diversified Diagnostic Products, Houston, TX). While patients inhaled a 33% Xe/67% O₂ mixture(XeScan, Praxair Pharmaceutical Gases, Danbury, CT) for 4.5 minutes, CT images were obtained at three levels through the brain. CBF measurements were determined by integrating the arterial and tissue accumulation curve of xenon to solve the Kety-Schmidt equation for each 1 × 1 × 10-mm³ CT voxel. The total time required to obtain CBF images (including data acquisition, CBF calculation, and display) was ∼15 minutes. Patients were closely supervised by a physician assistant during the study, which obviated the need for sedation in all cases.

Interpretation of xenon-enhanced CT scans. Xenon-enhanced CT CBF measurements were analyzed by means of a computerized data analysis program that calculates the mean CBF within a series of ∼2 × 2-cm regions of interest (ROIs) distributed throughout the cortical and subcortical regions. Each ROI consists of between 300 and 400 voxel measurements of CBF, the mean of which is determined for each ROI. Three axial CT scan slices were studied, yielding between 55 and 65 ROIs per patient. A total of ∼3,180 ROIs were analyzed in this study, representing the mean values of ∼21,000 CBF measurements for the study population. ROIs within areas that corresponded to artifacts on CT scans were excluded from the analysis. The mean CBF in each vascular territory was calculated by obtaining the mean values of all of the ROIs within the region most commonly supplied by the anterior cerebral artery (ACA), the middle cerebral artery(MCA), and the posterior cerebral artery (PCA) on each of the three CBF images obtained for each patient.

CBF studies were labeled as normal and without evidence of cerebral ischemia (nonischemic) if there were no mixed cortical ROIs with CBF 20 mL · 100 g^-1 · min^-1 in any vascular territory. The mean CBF of the symptomatic (affected) vascular territory or territories were also calculated and symptomatic (affected) vascular territories were those that were determined to be responsible for the patients presenting clinical symptoms and signs. This determination of which vascular territories to include as symptomatic (affected) was made before the calculation of the mean CBF for each territory so that the CBF data could not bias the interpretation of the clinical syndrome. The patients' deficits were attributed to the MCA territory in 46 patients (87%), the PCA territory in 3 patients (6%), and both the ACA and MCA territories in 4 patients (8%).

Results. Patients with complete resolution of deficits. Eight patients (15%) had a complete resolution of their symptoms within 24 hours that was not attributable to stroke therapy. None of these patients underwent thrombolytic or cerebral protective therapy. Five(63%) of these patients were treated with heparin alone and 3 (37%) managed without the use of heparin. The mean time to XeCT scan in this group of patients was 4 hours, 25 minutes (range 2 hours, 40 minutes to 6 hours). All eight patients had undergone XeCT scanning before any clinical improvement that might have suggested that their symptoms would subsequently completely resolve. The mean admission NIH scale score was 7.4 (varied from 3 to 16) for this group of patients.

One additional patient from the study population had a complete resolution of his deficits within 24 hours that might have been attributable to treatment. This patient underwent successful intra-arterial thrombolysis of an M1 segment occlusion with urokinase.

Relation between cerebral blood flow and deficit resolution. The mean CBF in the affected vascular territories was 35.4 ± 8.1 mL· 100 g^-1 · min^-1 in the group of patients whose deficits resolved (not attributable to therapy) compared with 17.3± 9.3 mL · 100 g^-1 · min^-1 in the remainder of the study population. This difference was statistically significant (p = 0.0002, t-test, two-tailed). The mean CBF of the symptomatic contralateral vascular territories was 38.6 ± 7.1 mL · 100 g^-1 · min^-1 in the group of patients whose deficits resolved (not attributable to therapy) compared with 33.7 ± 12.9 mL · 100 g^-1 · min^-1 for the remainder of the study population, not a statistically significant difference. Furthermore, there were no statistically significant differences between the mean CBF of the symptomatic vascular territories of the group of patients with deficits that resolved compared with the mean CBF of the asymptomatic control vascular territories of either the group of patients whose deficits resolved or those with evolving strokes. These relations are illustrated in figure 1.

All patients (100%, 9 of 9) with deficits that resolved had normal CBF studies that did not reveal any evidence of cerebral ischemia in any vascular territory (figure 2 and 3). Eight of these patients had deficits that resolved without specific stroke treatment other than heparin, whereas one patient received intra-arterial urokinase. Two of the 44 patients (5%) whose deficits did not resolve within 24 hours had normal CBF studies at the time of admission. Eight-eight percent of patients(7 of 8) whose deficits resolved within 24 hours (not related to treatment) had a mean CBF of the symptomatic vascular territory 25 mL/100 g/min compared with 18% of patients (8 of 45) whose deficits did not resolve. Ninety-seven percent of patients (37 of 38) with a mean CBF of <25 mL · 100 g^-1 · min^-1 had evolving strokes. Patients with a mean CBF 25 mL · 100 g^-1 · min^-1, therefore, were 32 times more likely to have a complete resolution of their symptoms within 24 hours compared with patients who did not. These associations were statistically significant (p = 0.0003, Fisher's exact test) These associations are shown in figure 4.

Differences between patients with deficit resolution and study population. There were no statistically significant differences between the group of patients with resolving deficits (not related to treatment) and the remainder of patients with respect to age, sex, percent receiving heparin, and time to XeCT scanning. The admission mean NIH Stroke Scale score of the group of patients with deficits that resolved was 7.6 compared with 12.4 for the remainder of the study population, suggesting a slightly better initial neurologic status in the patients who subsequently returned to normal.

Discussion. Identification of patients with deficits that will resolve. This study has shown that patients with acute stroke symptoms who will return to normal within a 24-hour period can be differentiated from patients whose symptoms will not subside on the basis of CBF measurements. Patients who had a rapid return to a normal neurologic examination had, in all cases, a normal CBF study within 6 hours of symptom onset and were 32 times more likely to have a CBF 25 mL · 100 g^-1 · min^-1 in the affected vascular territory. The rapid resolution of neurologic deficits seen in these patients has been reported previously as the "spectacular shrinking deficit phenomenon,"^5,6 presumably attributable to early lysis or migration of the offending thrombus or embolus, or possibly caused by recruitment of sufficient CBF from collateral blood vessels. In a prior study, we were able to predict the presence or absence of a major vessel thrombus at angiography based on CBF measurements in acute stroke.⁷ The fact that the patients in this study had nonimproving neurologic deficits at the time of CBF imaging suggests that neurologic improvement lags behind reversal of ischemia.⁸ The exact time at which a patient's neurologic deficit becomes fixed despite reperfusion and reversal of ischemia remains unknown, but is likely dependent on several factors, the most important of which are the duration of the ischemic insult, the degree of ischemia based on collateral blood supply, and the location of the ischemia with respect to critical brain structures. Importantly, CBF measurements provide insight into each of these factors in acute stroke. It is likely, therefore, that CBF measurements will have utility in predicting patient outcome.

Because the overall outcome in patients with normal CBF in the acute stroke setting appears to be excellent, and no ongoing ischemia is present, there is likely to be no increased benefit from thrombolytic stroke therapies that would still potentially carry attendant hemorrhagic risks both systemically and in the brain or angiographic risks if an intra-arterial route for thrombolysis were selected. Patients with normal CBF, therefore, probably should not receive thrombolytic stroke therapies designed to restore cerebral perfusion. Normative studies in healthy volunteers have shown that normal mixed gray and white matter ROIs such as those used in this study have a CBF of 51 ± 10 mL · 100 g^-1 · min^-1; mixed cortical CBF 20, therefore, is markedly abnormal (>3 SDs).⁹ In a prior study of delayed ischemia after subarachnoid hemorrhage, infarction occurred rarely in patients with regions of CBF 18 mL · 100 g^-1 · min^-1, an observation leading those investigators to conclude that aggressive management of vasospasm in that setting is probably not necessary¹⁰ and prompting us to attempt to correlate these "normal" CBF values with symptom resolution in acute stroke patients in this study. Prospective data, however, will be required to determine whether CBF measurements have a sufficient ability to predict deficit resolution before they can be recommended for the routine triage of acute stroke patients and selection of patients for thrombolytic stroke therapy.

It is not known whether these patients who rapidly improved had developed small infarcts that could have been seen on MRI in clinically silent regions of brain. Irreversible ischemic changes could have ensued before spontaneous reperfusion in areas of brain that were not manifest in the return to a normal neurologic examination. If such small, silent infarcts did occur, it is not clear whether cerebral protective therapy would have provided any benefit. In addition, it is possible that detailed neuropsychological evaluation of these patients could reveal new deficits that were not detectable in the routine clinical neurologic examination these patients underwent.

Determining efficacy of intervention. Quantitative CBF measurements will likely play a role in understanding the true efficacy of current and future stroke therapies. Some patients who receive thrombolytic agents, for example, and then proceed to have a rapid return to a normal examination may do so because of the natural history of their stroke event and not because of the treatment they received. Indeed, the only patient who received stroke therapy and subsequently returned to a normal examination within 24 hours in this study had normal CBF on his admission study. It is quite likely that this patient's favorable outcome was a function of the natural history of his ischemic event rather than a result of treatment. This ability of CBF measurements to predict patients who will have a rapid reversal of symptoms without treatment may allow one to determine whether the course of patients who receive thrombolytic therapy is related to spontaneous reperfusion or an effect of the treatment drug. Furthermore, some patients have sufficient colateralization such that they have normal CBF despite arterial occlusion and do not stand to benefit from thrombolysis. In this sense, CBF measurements provide physiologic insight into individual stroke events, providing preintervention natural history probabilities against which the effects of interventions can be compared. The ability to stratify the"risk" or severity of stroke events before treatment greatly enhances the interpretation of patient outcome.

Cerebral blood flow measurements in acute stroke. Despite the fact that the principal pathophysiologic event in acute ischemic stroke is a reduction in CBF, CBF measurements have had only limited application in the clinical management of acute stroke patients. This is due, in large part, to the inaccuracy and impracticality of the CBF techniques used. ¹³³Xe techniques measure only regional CBF and lack the ability to precisely define the anatomy of local variations in CBF that is necessary to have clinical utility in the diagnosis of many ischemic strokes.¹¹ Perhaps an even more notable weakness of ¹³³Xe techniques is the"look-through effect." Areas of hyperemia or normal CBF surrounding or overlying ischemic areas are "averaged" into the final CBF value, making¹³³ Xe measurements of CBF insensitive to very low flow values (0 to 8 mL · 100 g^-1 · min^-1), which are critical to understanding many ischemic events. SPECT determinations of CBF have been used in the study of acute stroke,^12,13 but they are nonquantitative and therefore rely on references to "standard" known values to make assumptions about the existence of ischemia. When CBF is lower than normal bilaterally in stroke, as it often is in our experience,⁷ SPECT will underestimate the significance of an asymmetry with respect to the degree of ischemia present. This greatly detracts from the utility of SPECT in the diagnosis of many strokes. PET technology has also been studied in patients with acute stroke, providing useful CBF and metabolic information¹⁴; however, PET studies are expensive and time consuming and therefore not practical for routine use in stroke. MRI with diffusion and perfusion weighting in acute stroke has been of increasing interest^15-17 and can allow for early identification of signal abnormalities in stroke. The information thus obtained provides early anatomic insight into the locus of the stroke event, and some physiologic information about CBF. XeCT, in distinction to all of these aforementioned diagnostic modalities, can be performed along with routine CT scans and is fast, accurate, and provides quantitative CBF information in 15 minutes.^9,18-21

Potential limitations of the study. This study is limited by its sample size and retrospective nature. We believe, however, that the 56 patients in whom XeCT scans were performed represent a population of patients with anterior circulation hemispheric ischemic stroke events typical of those who present to a tertiary care medical center. Patients were selected for inclusion in this study if their XeCT scans were performed within 8 hours of deficit onset, and most were much earlier. This time course is important in interpreting the results of this study because patients who went on to have a rapid resolution of their neurologic deficits within 24 hours likely had early cerebral reperfusion before the onset of permanent neurologic damage in critical brain regions. Of the 11 patients in this study with normal CBF at the time of admission, 8 had a spontaneous resolution of their symptoms within 24 hours (not related to treatment), 1 returned to a normal examination within 24 hours and was treated with intra-arterial urokinase, and 2 developed permanent neurologic deficits. The latter 2 patients either represent a failure of the XeCT technology to determine the presence of ischemia or they are patients who had spontaneously reperfused after the point at which irreversible ischemic damage had already ensued. Alternatively, the normal CBF in these two patients who sustained permanent neurologic deficits may represent a mismatch between CBF and metabolic need, or "luxury perfusion," first described by Lassen²² and subsequently by Ackerman et al.²³ If luxury perfusion was the sole reason for the development of an ischemic stroke despite normal CBF in these 2 patients, then it is important to note that such a phenomenon is rare indeed, occurring in only 2 of 53 patients (3.8%) within 8 hours of stroke onset. If patients were studied with XeCT later in their course, there may have been more patients in whom reperfusion had spontaneously occurred or more patients with luxury perfusion phenomena; however, more of them may have sustained permanent neurologic deficits. This would have decreased the correlation between CBF measurements and deficit resolution. Despite being only a single measurement in time in an ongoing process and not providing insight into metabolic phenomena, the data in this report indicate that XeCT CBF information obtained early after stroke onset nevertheless provide important predictive information about neurologic deficit resolution.

Implications for the management of acute stroke. The results of this study are important because they demonstrate the potential utility of XeCT CBF measurements in defining a population of patients with acute stroke symptoms with an excellent prognosis. The application of this functional imaging information in acute stroke may allow patients to be spared stroke treatments that are not likely to help them. If borne out in future prospective studies, the ability to triage acute stroke patients in such a way will represent a first step in the management of individual stroke patients based on a physiologic understanding of their ischemic events.

In addition to the value of CBF measurements in the management of individual patients, it is expected that XeCT technology will also play a major role in stroke research. The ability to stratify the preintervention severity of stroke events will greatly assist in the interpretation of the results of stroke treatment studies. It is likely that such stratification of the severity of ischemic strokes based on CBF will provide selectivity and reduce the number of patients who will need to be entered into stroke trials.

Acknowledgment

The authors thank Mr. John May for his expertise in the management of the XeCT database.