Sensitivity and Specificity of 99mTc-HMPAO SPECT Cerebral Perfusion Measurements During the First 48 Hours for the Localization of Cerebral Infarction
Background and Purpose There is no routinely used method for imaging the location of the extent and severity of cerebral tissue perfusion changes during the first hours of ischemic stroke, the period during which therapeutic intervention is most likely to be successful. Cerebral perfusion measurements with single-photon emission CT (SPECT) may potentially provide this information rapidly and noninvasively. In this study, the sensitivity and specificity of 99mTc–hexamethylpropyleneamine oxime (HMPAO) SPECT cerebral perfusion measurements during the first 48 hours of cerebral ischemia for the localization of cerebral infarction were determined.
Methods One hundred and four patients with acute ischemic stroke underwent 99mTc-HMPAO SPECT and CT scanning during the first 48 hours. In each patient, the location of the SPECT perfusion abnormality was compared with the location of infarction on a second brain CT acquired at a mean of 8 days after stroke.
Results During the first 48 hours of ischemic stroke, the sensitivity of 99mTc-HMPAO SPECT in locating the site of infarction was 79% (110/139), and the specificity was 95% (362/381). SPECT was more sensitive in the localization of the vascular territory of cortical infarction (sensitivity, 93%) than pure subcortical infarcts (sensitivity, 47%). During the first 48 hours, SPECT was significantly more sensitive than brain CT (sensitivity of brain CT during the first 48 hours, 35%; P<.001, Mann-Whitney U test).
Conclusions HMPAO SPECT measurement provides a widely available and practical technique of locating cerebral ischemia acutely and demonstrates high sensitivity and specificity within the first 48 hours for the localization of the vascular territory of cerebral infarction. It is most sensitive for cortical ischemia but is limited by its resolution in the subcortex, particularly of white matter perfusion changes.
- cerebral blood flow
- cerebrovascular disorders
- diagnostic imaging
- tomography, emission computed
In human ischemic stroke, a potential window of uncertain duration (probably hours) exists, during which therapeutic intervention may limit infarct size and improve outcome.1 2 3 Locating the site and the severity of cerebral perfusion changes during the acute phase is likely to be important in guiding therapeutic intervention, since it is possible that some therapies will need to be targeted to specific stroke subtypes and patients in whom still-viable tissue is present. However, in current practice, acute cerebral ischemia is not routinely visualized because the structural images obtained from CT and MRI are usually normal or show minimal changes within the first few hours.
With functional imaging methods such as SPECT, it is now possible to visualize cerebral perfusion changes in the acute phase. Compared with other functional imaging techniques, this modality has the advantages of being widely available, practical, and relatively inexpensive. Several studies have demonstrated the ability of SPECT perfusion measurements to demonstrate the ischemic focus, but no large-scale study against a universally accepted gold standard of the sensitivity and specificity of 99mTc-HMPAO SPECT in localizing cerebral ischemia within the first 48 hours (the time window during which it is believed that therapeutic intervention may potentially be effective) has been undertaken previously. Earlier studies had the following limitations: (1) small patient numbers,4 5 (2) heterogeneous examination times after the onset of ischemia,4 5 (3) the use of different radiopharmaceuticals6 7 such as 99mTc-ECD or 123I-IMP, (4) the use of clinical criteria for localization rather than a universally accepted or neuroimaging-based gold standard,6 8 (5) the absence of measurements of specificity except for one study,6 and (6) retrospective analysis in some studies. Three 99mTc-HMPAO SPECT studies8 9 10 had fewer than 40 patients within the first 24 to 48 hours, and two other studies recruited patients within the first 4 to 6 days after stroke.11 12 Most other studies have involved the use of SPECT up to 5 days after stroke or beyond.
The aim of this study was to determine the sensitivity and specificity of 99mTc-HMPAO SPECT cerebral perfusion measurements during the first 48 hours of cerebral ischemia for the localization of infarction on CT. A secondary aim was to compare the sensitivities of SPECT and CT during the first 48 hours.
Subjects and Methods
The study was conducted prospectively from March 1991 through October 1993. Patients presenting to the Austin Hospital Stroke Unit13 within 48 hours of the onset of a focal neurological event consistent with ischemic stroke were eligible for inclusion in the study. Some patients were participants in other stroke trials in progress at the Austin and Repatriation Medical Centre, including the pilot and randomized phases of the Australian Streptokinase Trial.14 15 The study was approved by the ethics committee of the Austin and Repatriation Medical Centre and the Radiation Safety Section of the Health Department of Victoria.
Patients were assessed in the emergency department. The following data were recorded: age, sex, time of symptom onset, neurological deficit, time of CT scanning, and 99mTc-HMPAO administration. The following inclusion and exclusion criteria were applied. Inclusion criteria were CT and SPECT examination within 48 hours of the onset of symptoms, supratentorial ischemic stroke, CT exclusion of cerebral hemorrhage and other causes of pseudostroke, and informed consent (verbal or signed) obtained from the patient or closest relative if the patient was unable to give consent. Exclusion criteria were cerebral hemorrhage or pseudostroke, infratentorial ischemic stroke, or previous infarction in the clinically relevant vascular territory interfering with scan interpretation.
Each patient underwent noncontrast CT scanning on arrival in the emergency department with a Siemens DR3 scanner to exclude cerebral hemorrhage and other causes of pseudostroke. The slice thickness of the reconstructed images was 8 mm. The CT was repeated after 7 to 10 days, or earlier if clinically indicated. The second CT was performed to determine the topography of infarction, which was used as the gold standard in determining the sensitivity and specificity of SPECT. In cases where multiple CT scans were performed for clinical reasons, the scan that most clearly delineated the topography of the infarct was used as the “gold standard.”
All SPECT scans were performed within 48 hours of the onset of symptoms and as soon after the patient’s presentation as was possible. In some cases, two SPECT studies were performed during the first 48 hours (in patients entered into the Australian Streptokinase Trial14 15 ); in those cases, only the first 99mTc-HMPAO SPECT study was evaluated.
99mTc-HMPAO (15 to 25 mCi, Ceretec-Amersham Australasia) was injected as soon as possible after the CT scan had been performed, either in the emergency department or in the CT scanning suite. Scanning was performed when clinically convenient using a rotating General Electric 400 AC Starcam camera. Sixty-four images were acquired over 360° on a 128×128 matrix with a pixel size of 3.1 mm and acquisition time of 15 to 30 seconds per frame. After scatter correction and attenuation correction, 40 transaxial slices were reconstructed on a 64×64 matrix.
Analysis of Neuroimaging Studies
The neuroimaging studies were read visually by two observers blinded to the clinical localization (a neurologist and a neurology trainee) who classified the structural changes on the CT scans and the perfusion changes on the SPECT images by vascular territory. The categories of vascular territories were MCA, ACA, PCA, Ex WS, subcortical, and normal. Subcortical was defined as any region without cortical involvement.16 The subcortical categories included anterior choroidal territory,17 lacune,18 internal watershed,19 and striatocapsular20 and white matter medullary infarcts.21 Subcortical infarcts and perfusion abnormalities were grouped because the resolution of our SPECT was 1.2 cm, measured in a resolution phantom. Because subcortical infarcts are often small (lacunae are defined as <1.5 cm), it was decided that it would be difficult to distinguish between the different categories of subcortical ischemia on SPECT. A total of 14 patients with clinical stroke (including 4 with reversible ischemic neurological deficits) were excluded because the late CT was normal.
Some patients had involvement of more than one vascular territory, which the readers recorded separately. Any discordant results were discussed and consensus reached. The interobserver agreement for late CT reading was overall κ=0.85 (MCA territory, κ=0.90; ACA territory, κ=0.82; PCA territory, κ=0.85; Ex WS, κ=0.49; subcortical, κ=0.65). The interobserver agreement for SPECT reading was overall κ=0.83 (MCA territory, κ=0.80; ACA territory, κ=0.79; PCA territory, κ=0.79; Ex WS, κ=−0.01; subcortical, κ=0.73).
Data Analysis and Statistical Analyses
The vascular territory of infarction on the second CT scan was used as the gold standard in determining the sensitivity and specificity of SPECT in localizing acute cerebral ischemia. In the overall data analysis, the perfusion abnormalities on the SPECT images in the first 48 hours were compared with the location of infarction on the late CT scan. For each patient and volunteer, a 6×6 contingency table was constructed with the CT location on one axis and the SPECT location of perfusion change on the other. The six categories were MCA cortical, ACA, PCA, Ex WS, subcortical, and normal.
In each patient, each of the five vascular territories was classified into true-positive, true-negative, false-positive, and false-negative categories (see below), giving a total of 520 vascular territories that were analyzed in 104 subjects. The cells of the 104 6×6 contingency tables were then summed to a 6×6 contingency table and then to an overall 2×2 contingency table that compared the HMPAO SPECT results with those of CT.
The classifications were defined as follows.6
True-positive (TP). The location of ischemia on the SPECT study corresponded to the location of infarction on the late CT, eg, when the images were concordant for MCA territory infarction and hypoperfusion.
False-positive (FP). A perfusion abnormality was detected on the SPECT image in a vascular territory in a patient who had a normal CT (a control subject) or an extra perfusion abnormality was present on the SPECT study in an area that was not infarcted on the late CT scan. For example, if a perfusion defect was seen in the PCA territory and the MCA territory on the SPECT study in a patient with isolated cortical middle cerebral infarction on the CT, the PCA perfusion defect was defined as a false-positive.
False-negative (FN). The SPECT scan was reported as normal when there was an area of infarction on the late CT or the SPECT showed a perfusion abnormality in a different location to the site of infarction on CT, eg, if a patient with an ACA territory infarct on CT had a normal SPECT study.
True-negative (TN). This classification was used when the SPECT revealed no abnormality in a patient without infarction on CT.
The sensitivity, specificity, and positive and negative predictive values were determined using the following formulas: Sensitivity=TP/(TP+FN); Specificity=TN/(FP+TN); Positive Predictive Value=TP/(TP+FP); and Negative Predictive Value=TN/(FN+TN).
For statistical power in the final analysis, the cortical infarcts were grouped. The sensitivities of CT and HMPAO SPECT were compared by χ2 analysis. Comparisons between groups were analyzed using unpaired Student’s t tests. A value of P<.05 was considered statistically significant.
A total of 104 patients were studied (Table 1⇓). There were 53 men and 51 women (Table 1⇓). The mean age of the patients was 69.3 years. The mean±SD time of 99mTc-HMPAO administration was 16.4±2.3 hours after the onset of symptoms. The second CT scan was performed at a mean time of 8.0±4.3 days. There were 139 vascular territories of infarction (80 cortical, 59 subcortical) (Table 2⇓) in the 104 patients. Eight patients had more than one cortical vascular territory involved. In 25 patients with cortical infarcts there was extension of the infarct into the lenticulostriate MCA territory.
Extra regions of hypoperfusion on SPECT in addition to the territory of infarction were identified in 19 patients: 1 patient with MCA infarction had additional PCA territory hypoperfusion; 7 patients with subcortical infarction had additional cortical hypoperfusion in the MCA territory; in another with an ACA territory infarct, there was cortical hypoperfusion in the MCA territory; and 10 patients with isolated cortical MCA infarcts also had subcortical hypoperfusion on SPECT. These extra regions may have represented areas of diaschisis or low flow or areas that were subsequently salvaged.
Sensitivity and Specificity of SPECT in Locating Ischemia During the First 48 Hours
SPECT was sensitive and specific in locating the site of ischemia during the first 48 hours (Table 3⇓), with an overall sensitivity of 79% and specificity of 95%. SPECT was more sensitive in cortical ischemia, correctly localizing the site of the ischemia in 93% of infarcted cortical vascular territories, although it was less sensitive in pure subcortical infarcts, where the sensitivity was 47%. SPECT also demonstrated high positive and negative predictive values for the acute localization of the vascular territory of infarction (Table 3⇓).
Sensitivity of CT in the Localization of Ischemic Change During the First 48 Hours
As a corollary, the sensitivity of the acute (admission) CT scans performed during the first 48 hours were compared with that of SPECT in the cohort of 104 patients with acute stroke. In this cohort, the CT scans on average were performed significantly earlier than the SPECTs (mean time for CT was 8.6 hours compared with 16.4 hours for SPECT, P<.01, Student’s t test). The overall sensitivity of CT in the acute phase was 34.5%, the sensitivity for detecting cortical lesions was 42%, and it was 18% for subcortical lesions. Therefore, during the first 48 hours of cerebral ischemia, SPECT was significantly more sensitive than CT in locating the site of eventual infarction (P<.01, χ2 analysis).
This study confirmed the hypothesis that 99mTc-HMPAO SPECT is sensitive and specific in detecting the location of cerebral ischemia during the first 48 hours. 99mTc-HMPAO SPECT was also significantly more sensitive than CT scanning during the same time period. The study had the advantages that a large cohort was studied, the gold standard against which sensitivity and specificity were measured was based on the topography of infarction rather than on clinical criteria, and patients were recruited extremely early compared with previous studies, on average at 16.4 hours for the SPECT study. The early recruitment of our patients reflects the changing approach to stroke in recent years, in which the importance of the first few hours in the assessment and management of this condition has been recognized.22 The finding that 99mTc-HMPAO is sensitive during the first 48 hours compares favorably with the results of previous studies using 123I-IMP SPECT,7 11 133Xe SPECT,23 and 99mTc-ECD SPECT.6 In 18F fluorodeoxyglucose positron emission tomography studies, higher detection rates in the first 48 hours of cerebral ischemia than CT have also been demonstrated.24 25 26 27 Beyond the first 48 hours, the sensitivity of CT approaches that of SPECT. In the subacute phase, SPECT may be even less sensitive because of the presence of spontaneous reperfusion. CT becomes more sensitive because of the evolution of tissue infarction.8 28 The importance of determining the sensitivity of SPECT in the acute phase, in the first 48 hours, is that this time window is most likely to be the most important for tissue rescue.
99mTc-HMPAO SPECT was particularly sensitive in locating cortical ischemia acutely (93% of cases). It was less sensitive for subcortical ischemia because subcortical infarcts are often smaller and located in the white matter. White matter perfusion is poorly resolved on SPECT because white matter has only one quarter the blood flow of the gray matter.29 30 For cortical ischemia, SPECT might have been expected to be 100% sensitive. Reasons for this may be that small cortical infarcts were not located on blinded reading or that spontaneous reperfusion may have occurred in previously ischemic tissue. Perfusion defects were found in 4 patients with reversible ischemic neurological deficits who were excluded because their late CT scans were normal.
CT was less sensitive than SPECT within the first 48 hours because the structural changes of infarction evolve over hours to days: early hypodensity on CT is due to early edema.31 In previous studies, CT was found to be positive in 20% of cases in the first 6 hours32 and 50% of cases in the first 48 hours.33 In this study, CT was 35% sensitive in the localization of infarction within the first 48 hours.
In studies of acute ischemic stroke using MRI,34 35 36 it was demonstrated that an abnormal signal could be detected within the first 2 to 4 hours in some cases. T2-weighted MRI has a higher sensitivity than CT but still detects only 30% of cases in the first 24 hours and 50% in the first 48 hours.34 35 36 Future improvements in sensitivity may be anticipated as higher resolution functional imaging technology becomes available, eg, multiple detector SPECT devices or functional MRI. However, functional MRI awaits full validation.37
SPECT showed an overall specificity of 95%, which demonstrated the ease with which normal perfusion may be identified. In a previous study in which specificity was determined in a subgroup of patients, a high specificity was also found (98% in the study of Brass et al6 using 99mTc-ECD). The 19 false-positives on SPECT reflect the fact that the extent of hypoperfusion on SPECT is often greater than the area of infarction delineated on CT, and they most likely arose from the changes of diaschisis or low flow or as a result of tissue salvage occurring after spontaneous reperfusion. For example, subcortical infarction may be associated with cortical hypoperfusion if there is subcortical diaschisis,38 and cortical hypoperfusion may occur if there are significant internal carotid artery or MCA stenoses and exhausted autoregulatory reserves.39 Early reperfusion of MCA territory ischemia may limit infarction to the structures, especially in the presence of a good collateral circulation,40 as demonstrated in patients with spectacular shrinking deficits.41 Despite these false-positives, the specificity was high because of the high number of true-negative vascular territories (362); five vascular territories were analyzed per patient because a large number of patients had more than one vascular territory involved.
One possible limitation of using CT after 7 days as the gold standard should be acknowledged: infarcts may go through an isodense phase, which may mean that the true extent of infarction was not recognized in some cases.
This study appeared to represent a standard stroke population in terms of the frequency and type of supratentorial cortical infarction, although lacunae, which may account for up to 15% to 25% of infarcts, were underrepresented in this series because a large number of our cases had normal CT scans.18 The results in this study indicate that SPECT is sensitive and specific in detecting acute cerebral ischemia, particularly cortical ischemia, and that it will be useful in assessing perfusion in patients undergoing acute stroke reperfusion therapies during the time window of tissue viability.14
Selected Abbreviations and Acronyms
|ACA||=||anterior cerebral artery|
|ECD||=||N,N1-1,2-ethylene-diylbis-l-cysteine diethylester dihydrochloride|
|Ex WS||=||external watershed|
|MCA||=||middle cerebral artery|
|PCA||=||posterior cerebral artery|
|SPECT||=||single-photon emission computed tomography|
This work was supported by grants from the National Health and Medical Research Council of Australia and by Amersham Australasia. The staff of the Nuclear Medicine and Radiology Departments are gratefully acknowledged.
Presented in part at the Scientific Meeting of the Australian Association of Neurologists, Canberra, Australia, May 22-25, 1994.
- Received January 6, 1997.
- Revision received February 13, 1997.
- Accepted February 13, 1997.
- Copyright © 1997 by American Heart Association
Astrup J, Siesjö BK, Symon L. Thresholds in cerebral ischemia: the ischemic penumbra. Stroke. 1981;12:723-725.
Marchal G, Beaudouin V, Rioux P, de la Sayette V, Le Doze F, Viader F. 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.
Kobayashi H, Hayashi M, Kawano H, Handa Y, Nozaki J, Yamamoto S, Matsuda H. Cerebral blood flow studies using N-isopropyl I-123-p-iodoamphetamine. Stroke. 1985;16:293-296.
Brass LM, Walovitch RC, Joseph JL, Léveillé J, Marchand L, Hellman RS, Tikofsky RS, Masdeu JC, Hall KM, Van Heertum RL. The role of single photon emission computed tomography brain imaging with 99mTc-bicisate in the localization and definition of mechanism of ischemic stroke. J Cereb Blood Flow Metab. 1994;14(suppl 1):S91-S98.
Hill TC, Magistretti PL, Holman BL, Lee RGL, O’Leary DH, Uren RF, Royal HD, Mayman CI, Kolodny GM, Clouse ME. Assessment of regional cerebral blood flow (rCBF) using SPECT and N-isopropyl-(I-123)p-iodoamphetamine (IMP). Stroke. 1984;15:40-45.
Laloux P, Doat M, Brichant C, Cauwe F, Jamart J, De Coster P. Clinical usefulness of technetium-99m HMPAO SPECT imaging to map the ischemic lesion in acute stroke: a reevaluation. Cerebrovasc Dis. 1994;4:280-286.
Just A, Schröter J. SPECT des Gehirnes mit 99mTc-HMPAO bei Patienten mit zerebrovaskulärer Erkrankung: Verleich mit der CT. Fortschr Röntgenstr. 1989;151:611-615.
Baird AE, Donnan GA, Austin MC, Fitt GJ, Davis SM, McKay WJ. Reperfusion after thrombolytic therapy in ischemic stroke measured by single-photon emission computed tomography. Stroke. 1994;25:79-85.
Bogousslavsky J. Topographic patterns of cerebral infarcts: correlation with etiology. Cerebrovasc Dis. 1991;1(suppl 1):61-68.
Bladin PF, Berkovic SF. Striatocapsular infarction: large infarcts in the lenticulostriate arterial territory. Neurology (Cleveland). 1984;34:1423-1430.
Bogousslavsky J, Regli F. Centrum semiovale infarcts: subcortical infarction in the superficial territory of the middle cerebral artery. Neurology. 1992;42:1992-1998.
National Stroke Association. Stroke: the first six hours. Emergency evaluation and treatment. J Stroke Cerebrovasc Dis. 1993;3:135-144.
Hughes RL, Yonas H, Gur D, Latchlaw R. Cerebral blood flow determination in the first 8 hours of cerebral infarction using stable xenon-enhanced computed tomography. Stroke. 1989;20:754-760.
Fink GR, Herholz K, Pietrzyk U, Huber M, Heiss W-D. Peri-infarct perfusion in human cerebral ischemia: its relation to tissue metabolism, morphology, and clinical outcome. J Stroke Cerebrovasc Dis. 1993;3:123-131.
Baron JC, Delattre JY, Bories J, Chiras J, Cabanis EA, Blas C, Bousser MG, Comar D. Comparison study of CT and positron emission tomographic data in recent cerebral infarction. AJNR Am J Neuroradiol. 1983;4:536-540.
Kushner M, Reivich M, Fiesvhi C, Silver F, Chawluk J, Rosen M, Greenberg J, Burke A, Alavi A. Metabolic and clinical correlates of acute ischemic infarction. Neurology. 1987;37:1103-1110.
Hayman LA, Taber KH, Jhingran SG, Killian JM, Carroll RG. Cerebral infarction: diagnosis and assessment of prognosis by using 123IMP-SPECT and CT. AJNR Am J Neuroradiol. 1989;10:557-562.
Sokoloff L. Cerebral circulation, energy metabolism, and protein synthesis: general characteristics and principles of measurement. In: Phelps ME, Mazziotta JC, Schelbert H, eds. Positron Emission Tomography and Autoradiography: Principles and Applications. New York, NY: Raven Press Publishers; 1986:1-71.
Brass LM, Rattner Z. SPECT imaging in cerebrovascular disease: brain SPECT perfusion imaging: image acquisition, processing display, and interpretation. In: Proceedings of Brookhaven National Laboratory Workshop; October 8-9, 1991; Stonybrook, NY. Pp 77-88.
Ramadan NM, Deveshwar R, Levine SR. Magnetic resonance and clinical cerebrovascular disease: an update. Stroke. 1989;20:1279-1283.
Moseley ME, Kucharczyk J, Mintorovitch J, Cohen Y, Kurhanewicz J, Derugin N, Asgari H, Norman D. Diffusion-weighted MR imaging of acute stroke: correlation with T2 weighted and magnetic susceptibility-enhanced MR imaging in cats. AJNR Am J Neuroradiol. 1990;11:423-429.
Yuh WTC, Crain MR, Loes DJ, Greene GM, Ryals TJ, Sato Y. MR imaging of cerebral ischemia: findings in the first 24 hours. AJNR Am J Neuroradiol. 1991;12:621-629.
Weiller C, Ringelstein EB, Reiche W, Buell U. Clinical and hemodynamic aspects of low flow infarcts. Stroke. 1991;22:1117-1123.
Powers WJ, Raichle ME. Positron emission tomography and its application to the study of cerebrovascular disease in man. Stroke. 1985;16:361-376.
Ringelstein EB, Biniek R, Weiller C, Ammeling B, Nolte PN, Thron A. Type and extent of hemispheric brain infarctions and clinical outcome in early and delayed middle cerebral artery recanalization. Neurology. 1992;42:289-298.
Baird AE, Donnan GA, Austin MC, McKay WJ. Reperfusion in the ‘spectacular shrinking deficit’ demonstrated by single photon emission computed tomography. Neurology. 1995;45:1335-1339.