HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Figures Only Full Text (PDF) All Versions of this Article: 2273012169v1 227/3/725    most recent Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Hunter, G. J. Articles by Gonzalez, R. G. Search for Related Content PubMed PubMed Citation Articles by Hunter, G. J. Articles by Gonzalez, R. G.

Whole-Brain CT Perfusion Measurement of Perfused Cerebral Blood Volume in Acute Ischemic Stroke: Probability Curve for Regional Infarction¹

¹ From the MGH Perfusion and Physiology Analysis Laboratory (G.J.H., H.M.S., L.M.H.) and Departments of Neurology (W.J.K., F.S.B., L.H.S., G.A.R.) and Radiology (R.G.G.), Massachusetts General Hospital, Gray Bldg, Rm 285, 55 Fruit St, Boston, MA 02114. From the 2000 RSNA scientific assembly. Received January 16, 2002; revision requested February 15; final revision received October 7; accepted October 23. Address correspondence to G.J.H. (e-mail: gjhunter@partners.org).

	ABSTRACT

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
PURPOSE: To determine the probability curve for regional cerebral infarction as a function of percentage normalized perfused cerebral blood volume (pCBV) in patients with acute ischemic stroke.

MATERIALS AND METHODS: The authors retrospectively analyzed whole-brain computed tomographic (CT) perfusion scans from 28 patients with acute stroke (<6 hours) due to major arterial occlusion, without intracranial hemorrhage. Each patient had a positive follow-up CT scan 1–4 days later, without interval thrombolysis. Normalized pCBV, expressed as a percentage of contralateral normal brain pCBV, was determined in the core infarction and in regions just inside and outside the boundary between infarcted and noninfarcted brain. These regions were dichotomized into infarcted (core and inner band) and noninfarcted (outer band) categories. Logistic regression analysis was then used to create a reference curve of probability of infarction as a function of percentage normalized pCBV.

RESULTS: Normalized pCBV values in the core, inner band, and outer band were 24.5% ± 2.3, 36.3% ± 2.4, and 72.1% ± 2.4, with corresponding probabilities of infarction of .99, .96, and .11. The normalized pCBV at which the probability of survival reached .5 was 58.0% ± 0.5. Sensitivity, specificity, and accuracy of the reference probability curve were 90.5% (209 of 231), 89.5% (212 of 237), and 90.0% (421 of 468), respectively. Negative and positive predictive values were 90.6% (212 of 234) and 89.3% (209 of 234), respectively. R² was 0.73, and differences in perfusion between core and inner and outer bands were highly significant (P < .0001).

CONCLUSION: A probability of infarction curve can help predict the likelihood of infarction as a function of percentage normalized pCBV.

© RSNA, 2003

Index terms: Brain, blood flow, 10.76, 10.12119, 17.7214 • Brain, infarction, 10.781 • Brain, perfusion, 10.12115, 10.12119 • Computed tomography (CT), perfusion study, 10.12119 • Thrombolysis, 10.1265

	INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
In a patient with signs and symptoms of stroke who presents to the emergency room within 6 hours of the onset of symptoms, there is now the possibility that some ischemic brain may be salvaged if blood flow is restored in a timely manner (1–3). However, restoring blood flow is itself a procedure that carries some risk. The decision regarding whether or not to attempt revascularization is determined by weighing the potential benefit against the risk of the procedure. One of the important factors considered in determining this decision is the assessment of amount and location of potentially salvageable ischemic brain.

A method for the measurement of perfused cerebral blood volume (pCBV) with use of conventional radiographic contrast material and helical computed tomographic (CT) scanning through the whole brain has been developed in an animal model of stroke (4).This whole-brain CT perfusion method has also been implemented in patients with acute stroke syndrome (5). The whole-brain CT perfusion study consists of helical, nonenhanced scanning followed by repeat scanning during contrast material infusion. From these data, an evaluation of brain anatomy, presence of hemorrhage, and the location and extent of hypoperfused brain regions can be performed within 10–15 minutes of a patient’s arrival in the CT suite. Furthermore, the presence of treatable vessel occlusion can also be determined with construction of CT angiograms (5). The purpose of our study was to determine the probability curve for regional cerebral infarction as a function of percentage of normalized pCBV in patients with acute ischemic stroke.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Patients
Under the auspices of our institutional review board, imaging data from patient studies obtained as part of the standard of clinical care for acute stroke at our institution were retrospectively reviewed; informed consent was not required by our institutional review board. Data were included in this analysis if the patient presented to the emergency room within 6 hours of onset of stroke symptoms, there was no evidence of intracerebral hemorrhage, the intracranial CT angiogram showed a unilateral intracranial major vessel occlusion without extracranial carotid occlusion, conventional treatment alone was used, and a follow-up CT scan was obtained 1–4 days after emergency room presentation that showed the presence of a territorial infarction in the distribution of a major artery, without evidence of interval hemorrhagic conversion since presentation. Major vessel occlusion was considered as involvement of any M1 or M2 segment of the middle cerebral artery, the A1 segment of the anterior cerebral artery, the distal internal carotid artery, the P1 segment of the posterior cerebral artery, or the basilar artery.

Between January 1995 and May 2001, three authors (G.J.H., H.M.S., L.M.H.) found in the radiology report database 210 cases of patients who had presented to the emergency room with strokelike symptoms and who underwent whole-brain CT perfusion imaging. Of these 210 potential studies, 28 fulfilled the inclusion criteria and were analyzed in the present study. There were 12 female patients with a mean age of 66 years and a median age of 69 years and 16 male patients with a mean age of 68 years and a median age of 70 years. For each patient, the site of occlusion and the occluded arterial territory were recorded, as was the time of presentation following the ictus. The mean time of presentation was 3 hours 30 minutes ± 12 minutes (range, 1 hour 30 minutes to 6 hours; median, 3 hours). The demographic information for each patient is presented in the Table.

Follow-up Brain CT Imaging
Follow-up transverse scans were obtained by using a routine nonenhanced CT protocol with the same imaging parameters as mentioned previously. This provided images through the brain from the foramen magnum to the vertex. The follow-up scan was used to identify infarcted brain. Newly infarcted brain was identified as tissue with hypoattenuation on the follow-up scan not present on the initial nonenhanced scan obtained at the patient’s presentation to the hospital with acute symptoms.

Data Analysis
Brain regions analyzed.—Normalized pCBV values were determined on a region-of-interest (ROI) basis with consensus among three of the authors: a neuroradiologist (G.J.H.), a neuroradiology fellow (H.M.S.), and a clinical medical physicist (L.M.H.). Four regions were identified in the following manner. For each section in the follow-up study that contained acutely infarcted brain, an ROI was drawn around the hypoattenuating tissue. These ROIs were then transferred to the initial contrast-enhanced scans (Fig 1). For each of these ROIs, inner and outer contours were then constructed parallel to and 5 mm from the ROI boundary that defined the acute infarction. The inner band ROI was inside and the outer band ROI was outside the ROI with the acute infarct. The three resultant ROIs corresponded to core infarction and inner and outer boundaries of infarcted tissue. The two ROIs that corresponded to newly infarcted brain (core and inner band) were reflected about the midline to the contralateral nonischemic hemisphere and fused to produce a single ROI that represented the nonischemic control region. From all ROIs, identifiable contrast material–filled vessels, skull, and cerebrospinal fluid–filled spaces were excluded by means of inspection (Fig 2). All the ROIs were then transferred from the initial contrast-enhanced scans to the initial nonenhanced scans for the calculation of the normalized pCBV values for each region. This process was facilitated by means of manual anatomic coregistration of the individual sections.

View larger version (87K): [in this window] [in a new window] [Download PPT slide]   Figure 1. Patient 4. Two series of transverse sections show the initial contrast-enhanced scans. Top: Scans obtained at time of stroke. Bottom: Scans obtained at 3-day follow-up. ROIs surrounding the infarction as defined on the follow-up scans are shown in yellow and have been transferred to the corresponding sections from the initial whole-brain CT perfusion imaging set; those defined on the initial scans are shown in red after large vessels and cerebrospinal fluid have been excluded from consideration. Note that there are some areas on the initial scans that are of normal attenuation and progress to infarction, whereas other areas of low attenuation on the initial scans do not progress to infarction. This is likely a reflection of local heterogeneity in perfusion at the margins of a territorial infarction.

View larger version (111K): [in this window] [in a new window] [Download PPT slide]   Figure 2. First two transverse sections obtained in patient 4 in Figure 1. The eventual infarction ROI is outlined in red, and the core infarction ROI is shown in yellow. A 5-mm band inside the eventual infarction margin is shown in green, and a 5-mm band outside the eventual infarction margin is shown in blue. The control region is shown in orange in the contralateral hemisphere. Such ROIs were obtained for all infarcted areas in each patient. Large vessels and cerebrospinal fluid spaces were excluded from consideration in every case.

Statistical Analysis
Means of normalized pCBV values for each ROI category in each patient were calculated. The data were analyzed by using analysis of variance with randomized block terms for patient versus region of infarction, with independent variables being the patient and the location of the infarction and the dependent variable being the percentage of normalized pCBV. This tested the hypothesis that regions with an infarct have low pCBV and those without an infarct have higher pCBV values. Pairwise post hoc evaluation of the null hypothesis that there are no differences among the core, inner band, and outer band regions was performed by using Fisher protected least-significant-difference, Scheffé, and Bonferroni-Dunn methods. Software (SAS; SAS Institute, Cary, NC) was used for this purpose.

A logistic regression model with categoric response variables of "infarction" or "no infarction" was used to determine the probability of brain infarction as a function of the percentage of normalized pCBV (6,7). A pCBV value was classified in the infarction category if it was from the core or the inner band and in the no infarction category if it was from the contralateral nonischemic brain or the outer band. On the basis of this dichotomous separation into infarction and no-infarction groups, a log-normal regression line was calculated by using the percentage of normalized pCBV values from the three ROIs in each section that contributed to a patient’s infarction. This provided the probability of infarction as a function of the percentage of normalized pCBV. The coefficient of determination (R²), 95% confidence limits, accuracy, sensitivity, specificity, and negative and positive predictive values were obtained as part of the regression process (6,7). The same software as mentioned previously was used for this purpose.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
There were a total of 117 sections that contributed to the regions of infarction and were identified on the follow-up CT scans obtained within 4 days after stroke onset. Demographic data and the average percentage of normalized pCBV value for each region type in each patient are presented in the Table. In addition, in all patients, an overall mean and standard error of the mean value for each region type (core, inner band, outer band) is provided in the Table and shown in Figure 3.

View larger version (23K): [in this window] [in a new window] [Download PPT slide]   Figure 3. Bar graph shows mean pCBV values, normalized as a percentage of the contralateral control region, for the infarct core, inner band, and outer band in the 28 patients. The asterisks represent statistically significant differences in pCBV between each of the affected regions (P < .001). The horizontal line at a normalized pCBV of 58% represents the threshold at which brain tissue is as likely to infarct as to survive.

The results of logistic regression are summarized in Figure 4. The y axis on the left is the probability of infarction; zero corresponds to the certainty of no infarction, and 1.0 corresponds to the certainty of infarction. The x axis is the pCBV normalized to the contralateral nonischemic control region expressed as a percentage of the contralateral nonischemic brain tissue pCBV. The lower y axis on the right belongs to the frequency distribution of percentage normalized pCBV values that belong to the infarcted category. The upper y axis on the right has been inverted for display purposes and belongs to the frequency distribution of percentage normalized pCBV values in the noninfarcted category.

View larger version (58K): [in this window] [in a new window] [Download PPT slide]   Figure 4. Graph of results of logistic regression analysis. The solid line represents the logistic regression curve of the probability of infarction (y axis) as a function of percentage normalized pCBV (x axis). The median survival (.5 probability of infarction) occurs at 58% normalized pCBV. Also shown are the frequency histograms for infarcted (lower left) and noninfarcted (upper right) regions of brain. The histogram in the lower left corner has its y axis on the lower right, and the histogram in the upper right corner is inverted for display purposes, with its y axis on the upper right. Both histograms use the same x axis as the logistic regression curve.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
With increasing awareness among both the public and physicians that potentially useful therapies exist for the treatment of acute stroke, focus has turned to rapid evaluation of the condition of an individual patient and subsequent choice of therapy. In particular, attention has been directed to the effective triaging of patients to intraarterial thrombolysis (1–3,8–13). Key considerations in deciding whether or not to attempt intraarterial thrombolysis include the time of triaging with respect to symptom onset, the presence of an accessible arterial occlusion, the location and amount of potentially salvageable ischemic brain, the expected functional outcome should the tissue infarct, and the outcome of risk-benefit discussions with the patient and/or family. In general, the earlier the patient receives appropriate treatment, the better the likely outcome; similarly, the larger the ratio of salvageable-to-infarcted brain, the better the likely outcome after thrombolysis.

Whole-brain CT perfusion imaging combined with large-vessel CT angiography enables identification of areas of decreased cerebral perfusion, as well as the presence of vessel thrombus potentially amenable to intraarterial thrombolysis (5). Thresholds of cerebral blood flow below which brain will infarct have been determined in primate models and depend on both the degree and duration of ischemia (14,15). Marginally ischemic tissue will eventually infarct unless it is reperfused. The aim of intraarterial thrombolysis is to reperfuse this potentially salvageable tissue before it infarcts. Whole-brain CT perfusion imaging is a straightforward technique, readily available in the emergency room, that allows assessment of cerebral perfusion based on the tenet that the amount of contrast material reaching any part of the brain is proportional to the amount of blood flowing into it. The physiologic parameter provided by whole-brain CT perfusion imaging is a measurement of pCBV. This parameter is dependent on the blood flow and blood volume in a region under investigation (4). Figure 3 shows that there is an increase in pCBV from the core to the inner band to the outer band in the region of cerebral ischemia and infarction. This progression in pCBV is statistically highly significant and serves to reinforce the concept of relatively less perfusion in the center, or core, of an infarction when compared to its periphery (inner band) and the tissue immediately outside the infarct zone (outer band), where there is likely to be better preservation of perfusion from collateral flow.

To construct the curve of probability of infarction versus pCBV, we used the statistical technique of logistic regression (6,7). This fits a log-normal curve to pCBV data that have been dichotomized into infarct and noninfarct groups. Figure 4 shows that there is overlap between the frequency distributions of pCBV values seen in surviving and infarcting regions. This is not unexpected, as inherent noise in the scanning and data reconstruction process, imperfect anatomic coregistration, and regional differences in collateral flow and cerebrovascular physiology all contribute to a spreading of the distribution of observed pCBV values. As only patients with major arterial occlusion were analyzed, the regions used to create the reference curve all came from territorial infarctions; no lacunar infarctions were seen in these patients. The results of the logistic regression analysis suggest that the probability-of-infarction curve is suitable as a basis for creating probability maps that can be used to predict the likelihood of infarction, as a function of the percentage of normalized pCBV, in any given region of brain, in patients with major arterial occlusion and territorial infarction. Creation of such maps depends on adequate registration between the nonenhanced and contrast-enhanced scans, as well as sufficient contrast material filling of the intracranial vasculature to allow distinction between normally perfused and ischemic regions of the brain. Furthermore, the smallest region that could be usefully evaluated is unknown at present. The section thickness presents a current physical lower limit of 5 mm, but this may be an underestimate of the minimum clinically important lesion size, which in turn depends on its location. In any event, it is the size of the potentially salvageable tissue that is sought for triage decision making, not the individual lesion size per se. Within these limitations, availability of such maps shortly after whole-brain CT perfusion imaging could allow a rapid decision to be made concerning deployment of appropriate therapy, including intraarterial thrombolysis. To be used clinically, generation of the probability of infarction maps has to be automatic and available in the emergency room at the time of scanning. Work is ongoing to implement this.

We created a curve of probability of infarction as a function of the percentage of normalized pCBV in patients with acute ischemic stroke who underwent whole-brain CT perfusion imaging within 6 hours of stroke onset. Normal, nonischemic tissue had a probability of infarction of less than 1% versus 99% for the core ischemic region. The band of tissue just inside the infarct margin had a probability of infarction of 96%, whereas the outer band (tissue that did not ultimately infarct) had a probability of infarction of 11%. Validity of the probability curve needs to be tested in a larger population before it can be used prospectively to assist in the triaging of patients with acute ischemic stroke.

	FOOTNOTES

 
Abbreviations: pCBV = perfused cerebral blood volume, ROI = region of interest

Author contributions: Guarantors of integrity of entire study, G.J.H., L.M.H.; study concepts, G.J.H., L.M.H., R.G.G., H.M.S.; study design, G.J.H., L.M.H., H.M.S.; literature research, H.M.S., G.J.H., L.M.H., F.S.B., W.J.K.; clinical studies, G.J.H., W.J.K., F.S.B., L.H.S., G.A.R., R.G.G.; data acquisition, G.J.H., L.M.H., L.H.S., G.A.R., W.J.K., F.S.B., H.M.S.; data analysis/interpretation, H.M.S., L.M.H., G.J.H., L.H.S., W.J.K., F.S.B.; statistical analysis, G.J.H., R.G.G., L.M.H.; manuscript preparation, H.M.S., G.J.H., L.M.H.; manuscript definition of intellectual content, editing, revision/review, and final version approval, all authors.

	REFERENCES

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 

1. The National Institute of Neurological Disorders and Stroke rt-PA Stroke Study Group. Tissue plasminogen activator for acute ischemic stroke. N Engl J Med 1995; 333:1581-1587.[Abstract/Free Full Text]
2. del Zoppo GJ, Higashida RT, Furlan AJ, Pessin MS, Rowley HA, Gent M. PROACT: a phase II randomized trial of recombinant pro-urokinase by direct arterial delivery in acute middle cerebral artery stroke. PROACT Investigators. Prolyse in Acute Cerebral Thromboembolism. Stroke 1998; 29:4-11.[Abstract/Free Full Text]
3. Furlan A, Higashida R, Wechsler L, 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.[Abstract/Free Full Text]
4. Hamberg LM, Hunter GJ, Kierstead D, et al. Measurement of cerebral blood volume with subtraction three-dimensional functional CT. AJNR Am J Neuroradiol 1996; 17:1861-1869.[Abstract]
5. Hunter GJ, Hamberg LM, Ponzo JA, et al. Assessment of cerebral perfusion and arterial anatomy in hyperacute stroke with three-dimensional functional CT: early clinical results. AJNR Am J Neuroradiol 1998; 19:29-37.[Abstract]
6. Andersen EB. Logistic regression analysis In: Introduction to the statistical analysis of categorical data. Berlin, Germany: Springer-Verlag, 1997; 157-203.
7. Ryan TP. Logistic regression In: Modern regression methods. New York, NY: Wiley, 1997; 255-314.
8. von Kummer R, Forsting M, Wildemann B, Sartor K. Thrombolysis in acute cerebral ischemia. Aktuelle Radiol 1993; 3:351-355. [German].[Medline]
9. del Zoppo GJ. Thrombolysis in acute stroke. Neurologia 1995; 10(suppl 2):37-47.
10. Tong DC, Yenari MA, Albers GW. Intravenous thrombolytic therapy in acute stroke. Vasc Med 1997; 2:51-60.[Medline]
11. Hacke W, Brott T, Caplan L, et al. Thrombolysis in acute ischemic stroke: controlled trials and clinical experience. Neurology 1999; 53:S3-S14.
12. Derex L, Nighoghossian N, Turjman F, et al. Intravenous tPA in acute ischemic stroke related to internal carotid artery dissection. Neurology 2000; 54:2159-2161.[Abstract/Free Full Text]
13. Lees KR. Thrombolysis. Br Med Bull 2000; 56:389-400.[Abstract/Free Full Text]
14. Jones TH, Morawetz RB, Crowell RM, et al. Thresholds of focal cerebral ischemia in awake monkeys. J Neurosurg 1981; 54:773-782.[Medline]
15. Symon L. Recovery of brain function following ischemia. Acta Neurochir Suppl 1993; 57:102-109.[Medline]

This article has been cited by other articles:

				G TAN and T GODDARD Neuroimaging applications of multislice CT perfusion Imaging, June 1, 2007; 19(2): 142 - 152. [Abstract] [Full Text] [PDF]

				K W Muir, H M Halbert, T A Baird, M McCormick, and E Teasdale Visual evaluation of perfusion computed tomography in acute stroke accurately estimates infarct volume and tissue viability J. Neurol. Neurosurg. Psychiatry, March 1, 2006; 77(3): 334 - 339. [Abstract] [Full Text] [PDF]

				D. C. Alsop, E. Makovetskaya, S. Kumar, M. Selim, and G. Schlaug Markedly Reduced Apparent Blood Volume on Bolus Contrast Magnetic Resonance Imaging as a Predictor of Hemorrhage After Thrombolytic Therapy for Acute Ischemic Stroke Stroke, April 1, 2005; 36(4): 746 - 750. [Abstract] [Full Text] [PDF]

				M. Wintermark, N. J. Fischbein, W. S. Smith, N. U. Ko, M. Quist, and W. P. Dillon Accuracy of Dynamic Perfusion CT with Deconvolution in Detecting Acute Hemispheric Stroke AJNR Am. J. Neuroradiol., January 1, 2005; 26(1): 104 - 112. [Abstract] [Full Text] [PDF]

				M. Galvez, G. E. York II, and J. D. Eastwood CT Perfusion Parameter Values in Regions of Diffusion Abnormalities AJNR Am. J. Neuroradiol., August 1, 2004; 25(7): 1205 - 1210. [Abstract] [Full Text] [PDF]

				S. Warach and J.-C. Baron Neuroimaging Stroke, February 1, 2004; 35(2): 351 - 353. [Full Text] [PDF]

This Article Abstract Figures Only Full Text (PDF) All Versions of this Article: 2273012169v1 227/3/725    most recent Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Hunter, G. J. Articles by Gonzalez, R. G. Search for Related Content PubMed PubMed Citation Articles by Hunter, G. J. Articles by Gonzalez, R. G.