Cerebral Hemodynamics in Human Acute Ischemic Stroke: A Study with Diffusion- and Perfusion-Weighted Magnetic Resonance Imaging and SPECT

Abstract

Nineteen patients with acute ischemic stroke (<24 hours) underwent diffusion-weighted and perfusion-weighted (PWI) magnetic resonance imaging at the acute stage and 1 week later. Eleven patients also underwent technetium-99m ethyl cysteinate dimer single-photon emission computed tomography (SPECT) at the acute stage. Relative (ischemic vs. contralateral control) cerebral blood flow (relCBF), relative cerebral blood volume, and relative mean transit time were measured in the ischemic core, in the area of infarct growth, and in the eventually viable ischemic tissue on PWI maps. The relCBF was also measured from SPECT. There was a curvilinear relationship between the relCBF measured from PWI and SPECT (r = 0.854; P < 0.001). The tissue proceeding to infarction during the follow-up had significantly lower initial CBF and cerebral blood volume values on PWI maps (P < 0.001) than the eventually viable ischemic tissue had. The best value for discriminating the area of infarct growth from the eventually viable ischemic tissue was 48% for PWI relCBF and 87% for PWI relative cerebral blood volume. Combined diffusion and perfusion-weighted imaging enables one to detect hemodynamically different subregions inside the initial perfusion abnormality. Tissue survival may be different in these subregions and may be predicted.

Keywords

Diffusion Human Magnetic resonance imaging Perfusion SPECT Stroke

Positron emission tomography (PET) has largely improved our knowledge of the pathophysiology of ischemic stroke in the living human brain (Furlan et al., 1996; Hakim et al., 1989; Heiss et al., 1992; Lassen, 1985; Marchal et al., 1996, 1999). Because of its high cost and limited availability, PET is of limited value in clinical practice. Because diffusion-weighted (DWI) and perfusion-weighted (PWI) magnetic resonance (MR) imaging scans have higher spatial resolution and are available in most institutions equipped with high-field MR units, they have been expanding our knowledge of ischemic stroke in vivo. A better understanding of the hemodynamic changes in the ischemic brain tissue may be helpful in predicting the outcome of stroke, thereby improving patient management and evaluation of the efficiency of novel therapies.

PWI can provide information about the capillary “perfusion” in the brain. DWI is sensitive to the molecular diffusion of water and detects decreased diffusion in ischemic brain tissue (Baird and Warach, 1998). Although the lesions seen in DWI have been reversible in some animal models (Minematsu et al., 1992b) and in patients with transient ischemic attacks (Kidwell et al., 1999), there are only a few observations of recovery of ischemic brain tissue with decreased diffusion in humans (Marks et al., 1996; Sorensen et al., 1996). In a recent study by Marks et al. (1999), apparent diffusion coefficient maps showed lesion shrinkage in patients treated with thrombolysis. However, fluid-attenuated inversion recovery images showed that this was not a true beneficial effect but merely pseudonormalization of the apparent diffusion coefficient (Marks et al., 1999). Thus, because the lesion on DWI is rarely reversible, DWI is at present the most accurate clinically available method to visualize infarcted brain tissue at the acute stage. By using combined DWI and PWI, it may be possible to distinguish the infarcted tissue from the hypoperfused but still viable ischemic tissue (Baird et al., 1997; Sorensen et al., 1996). The mismatch between PWI and DWI lesions has been shown to predict infarct growth (Baird et al., 1996; Karonen et al., 1999; Sorensen et al., 1999). The mismatch between single-photon emission computed tomography (SPECT) and DWI lesions has also been shown to predict infarct growth (Karonen et al., 2000).

Various types of perfusion maps can be generated from PWI data (Sorensen et al., 1999). Their significance in clinical stroke has not yet been fully clarified. Moreover, defining lesion volumes visually on PWI maps may impair the predictive capability (Sorensen et al., 1999), because minor hemodynamic changes in the ischemic territory on PWI maps are visually indistinguishable. It is crucial to establish empirical hemodynamic threshold parameters that could distinguish between brain tissue that is potentially salvageable with early reperfusion and irreversibly damaged tissue that will inevitably suffer infarction.

SPECT is a well-characterized method used to measure cerebral blood flow (CBF), and it has a well-established record in terms of predicting outcome relative to a fixed ischemic threshold (Shimosegawa et al., 1994; Ueda et al., 1999; Watanabe et al., 1999). Recently, a significant correlation in CBF in normal subjects was found between PWI and SPECT (Ernst et al., 1999), but the correlation between the two methods has not been previously studied in ischemic stroke.

The purposes of this study were (1) to clarify the relationship between the CBF values measured from SPECT and PWI in human ischemic stroke; (2) to determine the characteristics of cortical cerebral perfusion for ischemic tissue; and finally, (3) to determine the cutoff percentage of the hemodynamic parameters between ischemic tissue that ultimately remains viable and ischemic tissue that eventually suffers infarction.

MATERIALS AND METHODS

Patients

Nineteen patients (11 women, 8 men; mean age ± SD, 71 ± 8 years) with first-ever acute infarction in the territory of the anterior cerebral circulation were enrolled into this study. Because various subgroups of cerebral infarction have different clinicopathologic characteristics (Bamford et al., 1991; De Reuck et al., 1981), patients with lacunar infarcts or infarcts in the territory of posterior circulation were not included in this study. The inclusion criteria were as follows: (1) the patient had symptoms and signs of acute hemispheric ischemic stroke; (2) cerebral hemorrhage and other nonischemic causes for symptoms were excluded by computed tomography; (3) MR imaging could be performed within 24 hours of the onset of symptoms; (4) the follow-up MR imaging could be performed at 1 week after the symptoms onset; and (5) the patient did not receive thrombolytic or experimental neuroprotective agents. Use of anticoagulation or antiplatelet drugs was not an exclusion criterion.

On admission to the hospital, the mean volume of infarct (the lesion volume on DWI) was 36.1 ±33.5 cm³ (mean ± SD), and the National Institutes of Health Stroke Scale was 14 ± 6. The time delay from the onset of symptoms to the initial MR imaging was 10.0 ± 4.6 hours (Table 1). The onset of symptoms was defined as the time the patient was last known to be without symptoms.

TABLE 1.

Basic characteristics of the study population

Patient/age/gender	Time from ictus to MRI (h)	NIHSS	Total lesion volume on DWI at acute stage (cm³)	Lesion size on DWI	Lesion size on the initial MTT (PWI) and DWI at one week	Level of arterial occlusion on MRA at acute stage
1/73/F*	6.25	10	11.6	final > initial	MTT_init > DWI_final	Carotid siphon (R)
2/56/M*	9.5	16	32.1	final > initial	MTT_init > DWI_final	Carotid siphon (R)
3/72/M*	8.5	19	130.2	final > initial	MTT_init > DWI_final	Carotid siphon (R)
4/81/F*	13.0	7	3.4	final = initial	MTT_init > DWI_final	MCA, M1-segment (R)
5/53/M*	21.75	8	54.6	final = initial	MTT_init = DWI_final	MCA, anterior branch (L)
6/74/M*	3.5	12	32.9	final = initial	MTT_init = DWI_final	MCA, anterior and posterior branch (R)
7/84/F*	10.25	10	34.8	final > initial	MTT_init > DWI_final	MCA, anterior branch (R)
8/71/F*	8.0	6	20.4	final > initial	MTT_init > DWI_final	MCA, posterior branch (L)
9/75/F*	7.0	14	6.1	final > initial	MTT_init > DWI_final	MCA, posterior branch (L)
10/71/M*	14.75	20	63.9	final = initial	MTT_init = DWI_final	MCA, anterior branch (R)
11/68/M*	11.5	5	6.4	final > initial	MTT_init > DWI_final	Negative
12/59/M	6.5	19	50.6	final > initial	MTT_init > DWI_final	MCA, anterior branch (R)
13/78/M	7.5	19	33.6	final = initial	MTT_init > DWI_final	Carotid siphon (R)
14/71/F	6.75	22	11.9	final > initial	MTT_init = DWI_final	Carotid siphon (L)
15/68/F	10.0	23	40.5	final > initial	MTT_init > DWI_final	MCA, M1-segment (L)
16/71/F	6.25	13	21.7	final > initial	MTT_init > DWI_final	MCA, anterior branch (R)
17/81/F	16.0	18	103.5	final > initial	MTT_init = DWI_final	Carotid siphon (L)
18/72/F	6.5	20	15.7	final > initial	MTT_init > DWI_final	Carotid siphon (L)
19/72/F	17.0	9	12.9	final = initial	MTT_init > DWI_final	MCA, anterior branch (R)

Indicates the patient also underwent single photon emission computed tomography at acute stage. MRI, magnetic resonance imaging; NIHSS, National Institute of Health Stroke Scale on the admission to the hospital; DWI, diffusion weighted imaging; PWI, perfusion weighted imaging; MTT, mean transit time; MRA, magnetic resonance angiography; MCA, middle cerebral artery; R, right; L, left.

Eleven patients (5 women, 6 men; mean age, 71 ± 9 years) also underwent SPECT immediately after MR imaging at the acute stage. The tracer for SPECT was injected within 1 hour of PWI, which makes the findings in SPECT and PWI comparable. Informed consent was obtained from the patient or the patient's relative. The study design was approved by the Ethics Committee of Kuopio University Hospital.

MR imaging

Each patient was imaged at the acute stage and 1 week after the onset of symptoms. All MR imaging studies were performed with a 1.5-T whole-body scanner capable of echo planar imaging (Magnetom Vision, Siemens Medical Systems, Erlangen, Germany) by using a head coil. The patient's head was fixed with standard restraints used in routine clinical MR imaging. Each MR imaging examination consisted of DWI, PWI, 20-phase-contrast MR angiography of the circle of Willis, T2-weighted and proton density-weighted axial fast spin echo imaging, and precontrast and postcontrast T1-weighted axial spin echo imaging. The total imaging time was approximately 20 minutes.

DWI was performed with a single-shot echo planar spin echo sequence (repetition time, 4,000 milliseconds; echo time, 103 milliseconds). Other imaging parameters were as follows: slice thickness, 5 mm; interslice gap, 1.5 mm; field of view, 260 mm; matrix size, 96 × 128 interpolated to 256 × 256; and total acquisition time, 20 seconds. Nineteen axial slices parallel to the orbitomeatal line were imaged. Four images per slice were obtained, i.e., one T2-weighted image without diffusion weighting (b-value, 0 s/mm²) and three diffusion-weighted images with orthogonally applied diffusion gradients (b-value, 1,000 s/mm²). To avoid the effects of diffusion anisotropy, trace images (trace of the diffusion tensor) were calculated as an average of the signal intensity in each diffusion-weighted image on a voxel-by-voxel basis.

PWI was also performed with a single-shot echo planar spin echo sequence (repetition time, 1,500 milliseconds; echo time, 78 milliseconds, field of view, 260 mm; matrix size 116 × 256). Seven 5-mm-thick axial slices with 1.5-mm interslice gaps were imaged at the slice positions containing the largest diffusion defect. Forty images per slice were acquired with 1.5-second intervals. A 0.2 mmol/kg dose of gadopentetate dimeglumine (Magnevist, Schering AG, Berlin, Germany) was injected into an antecubital vein (rate 5 mL/s) followed by a 15-mL bolus of saline. The injection was given with an MR compatible power injector (Spectris, Medrad, Pittsburgh, PA, U.S.A.) after four baseline measurements.

Perfusion raw images were postprocessed to generate maps of cerebral blood volume (CBV), CBF, and mean transit time (MTT) (Østergaard et al., 1996a, b ). CBV was determined on a voxel-by-voxel basis by numerical integration of the first-pass concentration–time curve. The shape of the arterial input function was determined from the voxels located at a branch of the middle cerebral artery, showing large signal losses during the bolus passage. The tissue impulse response function was determined by deconvolving the tissue concentration–time curve with the arterial input function. CBF was subsequently determined as the height of the deconvolved tissue impulse response. MTT was then calculated according to the central volume theorem as the CBV/CBF ratio.

SPECT imaging

Technetium-99m ethyl cysteinate dimer (^99mTc-ECD) was used as the tracer. A dose of 550 to 720 MBq, which varied according to the body weight of the patient, was injected into an antecubital vein within 1 hour of the PWI. Within 60 minutes after the injection, high resolution SPECT acquisition was performed with a gamma camera equipped with fan-beam collimators (MultiSPECT 3; Siemens Medical Systems, Inc., Hoffman Estates, IL, U.S.A.). A full 360° rotation was acquired (40 views/detector, each for 35 seconds). The matrix size was 128 × 128 and the imaging resolution was 8 to 9 mm. Contiguous 6-mm-thick axial slices aligned parallel with the orbitomeatal line were reconstructed by filtered backprojection and a uniform attenuation correction was performed.

Regional analysis

The total infarct volume at acute stage was measured from all initial DWI slices containing the lesion. Volumes of diffusion were measured by drawing regions of interest (ROIs) around the lesions on the trace images and by multiplying the lesion area by the slice thickness. The interslice gap was estimated to contain a lesion of the same size as the slice above it, and the lesion inside the gap was included in the volume calculation.

The 20-phase-contrast MR angiography studies were independently interpreted by two reviewers (J.O.K. and R.L.V.). Disagreement occurred only in one case (patient 9) in whom consensus was reached.

Gray and white matter display different vulnerability to ischemia (Moody et al., 1990). Therefore, cortical and deep white matter subregions within the ischemic brain parenchyma may show different outcomes in response to an observed blood flow deficit. In addition, SPECT does not allow accurate measurement of the blood flow in deep white matter. So, to achieve more accurate comparisons, only the cortical areas (on SPECT) were analyzed in this study. Because of the limitation of spatial resolution, “cortical areas” contained not only gray matter but also some subcortical white matter.

SPECT-MR coregistration was performed on a Hermes workstation (Nuclear Diagnostics AB, Hägersten, Sweden) with multimodality image software (MultiModality, Nuclear Diagnostics AB, Hägersten, Sweden), which enables accurate alignment of MR and SPECT images (Julin et al., 1997). The thickness of the original contiguous axial SPECT images was first reconstructed to 6.5 mm (slice thickness 5 mm + interslice gap 1.5 mm on PWI), then the SPECT images were reoriented to confirm that the boundaries of brain tissue and lateral ventricles aligned with the corresponding structures in PWI slices. All analyzes were limited to the tissue volume of seven slices that could be imaged with PWI (with the exception of the total infarct volume presented in Table 1). The number of slices analyzed for each patient varied from three to seven.

The ischemic core, the area of infarct growth, the eventually viable ischemic tissue, and the normal tissue were determined by visually reviewing the MR examinations from the acute stage and at 1 week (Fig. 1). The ischemic core was defined as the lesion on the diffusion trace images at the acute stage. The area of infarct growth was determined as the difference between lesions in the diffusion trace images obtained at the acute stage and at 1 week. The eventually viable ischemic tissue was determined as the mismatch area between the infarct area on the diffusion trace images at 1 week and the area of prolonged MTT at the acute stage PWI. This zone was determined only if the size of the lesion on MTT map at the acute stage PWI was larger than the size of the lesion on trace images at 1 week. The normal tissue was determined as the tissue that had a normal appearance on both PWI and DWI.

FIG. 1.

A 71-year-old woman (patient 8) with right hemiparesis and dysphasia 8 hours before the first MRI. DWI trace images on the first day (A) and at 1 week (B) were compared with each other and with PWI MTT map on the first day (C) when the ROIs were drawn. First, ROI 1 was drawn on the cortical region of infarcted core detected by DWI on the first day (A). Second, cortical area of infarct growth (ROI 3) was determined by comparing the trace images (A) and (B). The eventually viable cortical ischemic tissue (ROI 5) was defined as tissue with prolonged MTT (C) but which remained non-infarcted at one week (B). Finally, ROI 7 was placed on cortical tissue that appeared normal on both DWI and PWI. For calculation of relative values for perfusion parameters, the corresponding ROIs 2, 4, 6 and 8 were drawn on the contralateral normal side. The ROIs were copied from DWI to corresponding slices on the MTT (C), CBF (D) and CBV (E) maps calculated from PWI raw data, as well as on the reoriented SPECT slices (F). Relative values of various perfusion parameters were then determined by calculating the ratios of values measured from ROIs on ischemic side (for example, ROI 1)and contralateral side (for example, ROI 2).

Irregular ROIs were drawn manually by one interpreter (Y.L.) on the cortical brain areas of the four zones (ischemic core, area of infarct growth, eventually viable ischemic tissue, and normal tissue) at first on DWI trace images. The ROIs were then placed on the corresponding contralateral regions. These ROIs were transferred to the corresponding anatomic regions on PWI maps and on the reoriented SPECT slices (maximum, seven slices). If one ROI could not cover the whole specific ischemic zone, two or more ROIs were drawn. When drawing the ROIs of normal tissue, all normal appearing cortical tissue in the territory of anterior circulation was included. However, tissue near the air cavities of the skull was avoided, because of the sensitivity of echo-planar imaging sequences to susceptibility artifacts. Figure 1 demonstrates how the ROIs were drawn. On each kind of MR perfusion maps, a total of 91 ROIs were drawn within the ischemic core, 57 ROIs within the area of infarct growth, 87 ROIs within the eventually viable ischemic tissue, and 79 ROIs within the normal tissue in 19 patients. On SPECT images, 46 ROIs were drawn within the ischemic core, 20 ROIs within the area of infarct growth, 52 ROIs within the eventually viable ischemic tissue, and 38 ROIs within the normal tissue in 11 patients. In three patients (patients 9, 11, and 14), lesions on the initial trace images were localized in the white matter and no ROI was drawn on the cortex of the ischemic core. In one patient (patient 3), no ROI was drawn on the normal tissue, as the entire hemisphere showed prolonged MTT values.

The percentage ratios between the hemodynamic parameters on the ischemic side and those on the contralateral region were calculated, and the relative CBF (relCBF) from PWI and SPECT and relative CBV (relCBV) and relative MTT (relMTT) from PWI were thus determined. The mean values of relCBF, relCBV and relMTT of each zone in each patient were also calculated by averaging all individual measurements in corresponding zones on different slices. These mean values were used in the further multiple zone comparisons.

To test the reproducibility of the PWI and SPECT measurements, interobserver variability and intraobserver reproducibility were analyzed. To test the interobserver variability, two observers (Y.L. and A.H.) independently measured relative hemodynamic parameters of the three ischemic zones on PWI and SPECT slices in 10 randomly selected patients. To test the intraobserver reproducibility, one observer (Y.L.) performed the measurements twice on PWI and SPECT in five randomly selected patients with an interval of at least 1 month between the measurements.

Statistical analysis

With the exception of relMTT of the ischemic core, the relative hemodynamic parameters proved to be normally distributed by the Kolmogorov-Smirnov test. Intraclass correlation coefficient (r_i) was used to analyze the reproducibility of the measurements. Pearson's correlation coefficient (r) was used to calculate the correlation between the relCBF in all ROIs consisting of the ischemic zones and normal brain area measured from SPECT and PWI. The difference between the relCBF in ROIs measured from PWI and SPECT in each corresponding ischemic zone was analyzed with a paired Student's t-test. Discriminant analysis was performed to estimate cutoff values for relCBF and relCBV in ROIs to discriminate between the area of infarct growth and the eventually viable ischemic tissue. The sensitivity, specificity, and efficiency to predict tissue outcome (infarction or noninfarction) were calculated by using these cutoff values. P < 0.05 was considered statistically significant.

The differences between the mean values of relCBF, relCBV, and relMTT of the three ischemic zones and normal brain were studied by the Kruskal-Wallis test. If the differences were significant, paired comparisons were performed with the Mann-Whitney U test.

RESULTS

All patients had findings consistent with infarct (decreased molecular diffusion) at the acute stage. In 13 patients, the volume of the infarcted tissue increased on DWI images. In six patients, the volume of the infarct did not increase during the first week when evaluation of the DWI images occurred. In 14 cases, the initial MTT perfusion defect was larger than the final size of the infarct on the DWI images 1 week after the onset of the symptoms. In 5 patients, the initial perfusion defect on the MTT maps was equal to the final size of the infarct on the DWI images at 1 week (Table 1). In all 19 patients, there was an ischemic core; in 13 patients, there was a zone consistent with infarct growth; and in 14 cases, there was a zone corresponding to eventually viable ischemic tissue. The MR angiography findings are summarized in Table 1.

Reproducibility of the measurements

The reproducibility analysis was performed between the individual measurements on various slices. The intraclass correlation coefficients of the relCBF (PWI), relCBV, relMTT, and relCBF (SPECT) in interobserver variability analysis were 0.848, 0.864, 0.796 (P < 0.01; n = 100) and 0.892 (P < 0.01; n = 45), respectively, and in intraobserver reproducibility analysis 0.931, 0.938, 0.835 (P < 0.01; n = 40) and 0.934 (P < 0.01; n = 26), respectively.

Comparison of SPECT and PWI

In the 11 patients who underwent both MR imaging and SPECT during the acute stage, the relCBF measured from SPECT showed a curvilinear relationship with those measured from PWI. A statistically significant correlation existed between the relCBF measured from SPECT and PWI when a logarithmic transformation was performed for the relCBF measured from PWI (linear correlation, r = 0.777, n = 156; and linear-logarithmic correlation, r = 0.854, P < 0.001, n = 156) (Fig. 2). The relCBF values measured from SPECT were significantly higher than those measured from PWI in each of the three ischemic zones (ischemic core, area of infarct growth, and eventually viable ischemic tissue) (P < 0.001; n = 46, 20, and 52, respectively).

FIG. 2.

The relative CBF measured from SPECT shows significant correlation with the logarithmic transformed relative CBF measured from PWI (r = 0.854; P < 0.001). Values of the infarcted core (○, n = 46), the area of infarct growth (●, n = 20), the eventually viable ischemic tissue (Δ, n = 52), and the normal tissue (▲, n = 38) are included.

Discriminant analysis

According to the discriminant analysis, the best cutoff value to discriminate between the area of infarct growth and the eventually viable ischemic tissue was 48% for PWI relCBF and 87% for PWI relCBV (Fig. 3). The sensitivity, specificity, and efficiency of discrimination to predict the tissue outcome are summarized in Table 2.

FIG. 3.

According to the discriminant analysis, the best cutoff value to discriminate between the area of infarct growth (○) and the eventually viable ischemic tissue (●) was 48% for relative cerebral blood flow (A) and 87% for relative cerebral blood volume (B). A total of 57 samples in the area of infarct growth in 13 patients, and 87 samples in the eventually viable ischemic tissue in 14 patients, were obtained in each perfusion maps.

TABLE 2.

Percentage values of discriminant analysis for determining abilities of relCBF and relCBV to predict the outcome of ischemic tissue

	Cut-off value	Sensitivity	Specificity	Efficiency
RelCBF	48	88	66	74
RelCBV	87	83	82	82

RelCBF, relative cerebral blood flow; RelCBV, relative cerebral blood volume.

Relative hemodynamic parameters

The mean values of the relative hemodynamic parameters from affected slices of each patient in the four zones are listed in Table 3. In PWI, relCBF values in the ischemic core, the area of infarct growth, and the eventually viable ischemic tissue were 13% ± 6%, 35% ± 12%, and 66% ± 16%, respectively. In SPECT, relCBF values in the corresponding zones were 27% ± 14%, 69% ± 15%, and 87% ± 7%, respectively. The relCBV values in the corresponding zones were 25% ± 10%, 69% ± 19%, and 113% ± 13%, respectively. The relMTT values in the corresponding zones were 199% ± 75%, 209% ± 46%, and 184% ± 42%, respectively. The tissue proceeding to infarction during the follow-up had significantly lower initial relCBF and relCBV values on the PWI maps (P < 0.001) than the eventually viable ischemic tissue had. Also the relCBF measured from SPECT was lower in the zone proceeding to infarction than in the eventually viable ischemic tissue in seven patients (P = 0.011). For both relCBF and relCBV measured from PWI, the values of the four different zones differed statistically significantly (P ≤ 0.003). The relCBF values were significantly lower, but the relCBV values higher, in the eventually viable ischemic tissue than in the normal brain. For relMTT, the values differed statistically significantly only when normal tissue was compared with ischemic zones (Table 4).

TABLE 3.

The relative hemodynamic parameters of the four zones at the acute stage

	Ischemic core				Area of infarct growth				Eventually viable tissue				Normal tissue

Patient	SPECT	CBF	CBV	MTT	SPECT	CBF	CBV	MTT	SPECT	CBF	CBV	MTT	SPECT	CBF	CBV	MTT
1	44	17	24	185	77	35	84	219	85	50	102	209	95	98	100	99
2	26	6	10	188	85	16	53	288	95	50	109	224	101	114	119	96
3	16	8	19	201	74	33	82	255	87	72	128	190	—	—	—	—
4	29	21	47	171	—	—	—	—	77	62	116	185	97	99	100	98
5	54	25	36	197	—	—	—	—	—	—	—	—	96	130	108	90
6	20	16	20	137	—	—	—	—	—	—	—	—	98	91	88	98
7	17	16	37	140	72	31	60	193	92	100	140	124	97	110	103	103
8	26	9	22	187	51	44	86	204	94	67	107	171	96	95	96	101
9	—	—	—	—	46	43	95	205	75	57	98	184	90	100	98	95
10	9	11	22	214	—	—	—	—	—	—	—	—	132	119	107	96
11	—	—	—	—	81	28	66	265	87	57	111	215	101	103	106	100
12	np	9	20	203	np	40	76	178	np	61	121	189	np	88	95	101
13	np	5	11	188	np	—	—	—	np	71	111	162	np	100	96	103
14	np	—	—	—	np	27	35	152	np	—	—	—	np	87	89	99
15	np	22	26	129	np	23	36	143	np	66	92	137	np	92	102	99
16	np	13	30	269	np	44	75	198	np	70	124	178	np	96	97	98
17	np	7	15	184	np	60	80	155	np	—	—	—	np	104	94	102
18	np	5	29	449	np	25	67	258	np	42	99	285	np	98	97	111
19	np	18	29	149	np	—	—	—	np	95	117	126	np	96	99	97

Mean	27	13	25	199	69	35	69	209	87	66	113	184	100	101	100	99
± SD	± 14	± 6	± 10	± 75	± 15	± 12	± 19	± 46	± 7	± 16	± 13	± 42	± 12	± 11	± 7	± 4

The figures are percentages calculated by comparing the ischemic side to the contralateral side.

Dash indicates that regions of interest can not be drawn in the zone. np = not performed. SPECT, single photon emission computed tomography; CBF, cerebral blood flow; CBV, cerebral blood volume; MTT, mean transit time.

TABLE 4.

Statistical significance (P values) between the relative hemodynamic parameters of the four zones

Zone	Ischemic core (zone 1)	Area of infarct growth (zone 2)	Eventually viable ischemic tissue (zone 3)
RelCBF (PWI)
Zone 2	<0.001
Zone 3	<0.001	<0.001
Zone 4	<0.001	<0.001	<0.001
RelCBV (PWI)
Zone 2	<0.001
Zone 3	<0.001	<0.001
Zone 4	<0.001	<0.001	0.003
RelMTT (PWI)
Zone 2	0.188
Zone 3	0.708	0.167
Zone 4	<0.001	<0.001	<0.001
RelCBF (SPECT)
Zone 2	0.002
Zone 3	0.001	0.011
Zone 4	<0.001	0.001	0.001

Significance in differences between zones was studied with Mann-Whitney U test and interpreted as significant at P < 0.05. Zone 4 = normal tissue. RelCBF, relative cerebral blood flow; PWI, perfusion weighted imaging; RelCBV, relative cerebral blood volume; RelMTT, relative mean transit time; SPECT, single photon emission computed tomography.

DISCUSSION

At present, the efficient therapeutic time window for ischemic stroke is limited to 6 hours within the onset of symptoms (Albers, 1999). This rigid time window has been challenged by studies on human ischemic stroke (Baron et al., 1995). There is evidence that many patients have potentially salvageable tissue even up to 24 hours after stroke onset (Albers, 1999; Baron et al., 1995; Fisher, 1997). The potential effectiveness of therapy does not depend only on the time delay from the onset of symptoms, but obviously also on the presence of salvageable tissue at the time of administration of the treatment (Albers, 1999; Baron et al., 1995; Fisher, 1997). Although only one patient in the present study had the first MR imaging within 6 hours, our results reflect the clinical reality; i.e., only a minority of patients with ischemic strokes are hospitalized within 6 hours of the onset of symptoms. In contrast, the patient imaged within 6 hours of the onset of symptoms (patient 6) did not seem to have had salvageable ischemic tissue left.

Ischemia causes a decrease in the molecular diffusion of water. The main mechanism for this change is thought to be cytotoxic edema (Baird and Warach, 1998). DWI is able to detect ischemic cellular injury earlier than any other imaging method in vivo. Early ischemic changes have been detected with DWI even when standard T2-weighted images are normal, both in animal models and in humans (Moseley et al., 1990; Sorensen et al., 1996). In a rodent stroke model with focal middle cerebral artery occlusion, the DWI lesion volume became equivalent in size to the histologically measured final infarct size within 24 hours (Knight et al., 1994; Minematsu et al., 1992a). Schwamm et al. (1998) reported that in humans, the DWI lesion volume reached its maximum, on average, 70 hours after onset of stroke. The lesion on the initial DWI can be considered to represent the ischemic core. Subsequently, the lesion detected with DWI at 1 week can be considered to represent the final infarcted tissue. The difference between the lesions assessed by DWI on the initial examination and on the examination at 1 week can be determined as the area of infarct growth. However, because T2 effect also contributes to the signal intensity on DWI, accurate determination of the extent of the infarcted tissue at 1 week may be impaired by vasogenic edema.

Using PET, Schumann et al. (1998) suggested that CBF/CBV ratio (1/MTT) is directly proportional to the cerebral perfusion pressure. MTT may be prolonged even when there is only a modest decrease in the perfusion pressure, such as in oligemic states. Consequently, because MTT is a very sensitive marker of decreased perfusion pressure in acute stroke, it may be used for detecting tissue that is at risk of infarction but that may also remain viable if adequate perfusion is restored.

Comparison of SPECT and PWI

There was a curvilinear relationship between the relCBF measured from ^99mTc-ECD-SPECT and PWI. The reason behind this phenomenon may be that the ^99mTc-ECD uptake ratio shows a curvilinear relationship with CBF measured with PET (Tsuchida et al., 1994). In contrast, the CBF measured from PWI is in a linear relationship with absolute CBF measured with PET (Østergaard et al., 1998).

^99mTc-ECD has moderate cerebral extraction in cortical regions with high flow rates, and high cerebral extraction in regions with low flow rates (Walovitch et al., 1994). When ^99mTc-ECD-SPECT was performed on patients with acute ischemic stroke, the underestimation of CBF was more severe on the contralateral control side (with high flow rate) than on the ischemic side (with low flow rate). Therefore, the lesion-to-normal ratio (relCBF) on ^99mTc-ECD-SPECT underestimated the severity of hypoperfusion in patients with ischemic stroke. Furthermore, unlike the feature of the another perfusion-dependent tracer, technetium-99m-hexamethylpropyleneamine oxime (^99mTc-HMPAO), the uptake of ^99mTc-ECD is affected also by brain metabolism. In hypoperfused states, the uptake of ^99mTc-ECD is better preserved than would be predicted by the CBF (Shimosegawa et al., 1997). This may also contribute to the discrepancies between relCBF measured on SPECT and PWI in the present study. In clinical practice, the potential of ^99mTc-ECD-SPECT to provide information also of brain metabolism might also be of special interest.

SPECT relies on the diffusion of a lipophilic radiotracer across the blood-brain barrier and provides a good estimate of perfusion at the capillary level. PWI is relatively sensitive to the presence of large vessels, even when spin echo echo-planar imaging sequences are used, and provides perfusion information from cortex and nearby vessels in the sulci (Ernst et al., 1999). CBF in the vessels in the sulci, close to the ischemic territory, is affected by the arterial occlusion on the ischemic side. The ratio between the PWI CBF measured from the ischemic side and that from the contralateral side therefore overestimates the severity of the ischemia. Ernst et al. (1999) reported that elimination of large vessels in PWI can improve the correlation between ^99mTc-HMPAO-SPECT and PWI in normal subjects. In our material, the relatively high signal from large vessels may partly explain the different relCBF values between SPECT and PWI in ischemic stroke. In addition, the acoustic noise during MR imaging can lead to activation in the primary auditory cortex in the normal temporal lobes, which results in increased CBF in PWI (Bandettini et al., 1998).

Relative hemodynamic parameters

A study by Shimosegawa et al. (1994) showed that the relCBF values, which were measured from the ^99mTc-HMPAO-SPECT within 6 hours of stroke onset, were 48% ± 14% and 75% ± 10%, respectively, for the infarct and peri infarct regions. The relCBF of ^99mTc-ECD-SPECT for the corresponding tissues in the present study were 45% ± 26% (mean relCBF of the ischemic core and the area of infarct growth) and 87% ± 7% (the eventually viable ischemic tissue), respectively. Our relCBF value for the periinfarct tissue was therefore somewhat higher than reported by Shimosegawa et al. (1994). This discrepancy could be explained by the fact that the lesions with prolonged MTT on MTT maps were visually larger than lesions with reduced CBF on SPECT. Consequently, a part of the eventually viable tissue in the present study would have been categorized as normal tissue in their study. The effect of metabolism on the uptake of ^99mTc-ECD may also play a role in this discrepancy.

The threshold for cellular injury at a given CBF is directly related to the duration of ischemia (Hoehn-Berlage et al., 1995; Mancuso et al., 1995). Both experimental stroke models and human studies have shown that the ischemic tissue with very low CBF (<10 to 12 mL/100 g/min) will relatively rapidly lead to irreversible ischemic damage (Hakim et al., 1989; Heiss et al., 1992; Jones et al., 1981; Marchal et al., 1999). In the surrounding region of moderately ischemic tissue (approximately in the range of 10 to 20 mL/100 g/min), the tissue will gradually suffer infarction if low flow persists for a sufficiently long time (Furlan et al., 1996; Hossmann, 1994; Marchal et al., 1996). If the CBF remains in 20 to 30 mL/100 g/min, the tissue will not suffer infarction (Furlan et al., 1996). Flow of 20 to 23 mL/100 g/min seems to be the cutoff CBF value between ischemic tissue that ultimately remains viable and ischemic tissue that eventually suffers infarction (Furlan et al., 1996; Hakim et al., 1989; Jones et al., 1981; Marchal et al., 1996). The CBF of mixed gray and white matter is about 50 mL/100 g/min in normal human brain (Lassen, 1985). Therefore, cutoff relCBF value of 48% can be converted into 24 mL/100 g/min, and the relCBF in ischemic core (13%), the area of infarct growth (35%), and eventually viable ischemic tissue (66%) can be converted into 7, 18, and 33 mL/100 g/min, respectively. These values are in good agreement with the previous studies. In further agreement with our findings, Takagi et al. (1993) demonstrated in a cat stroke model that the threshold CBF for moderate glutamate release was 48% of the preischemic CBF. Glutamate is thought to be the major cause of excitatory neurotoxicity, and this is one of the earliest events occurring in stroke (Takano et al., 1997; Tatlisumak et al., 1998). Hatazawa et al. (1999) suggested that for relCBF, the cutoff value was 52% and for relCBV the cutoff value was 85% between infarction and noninfarction in human ischemic stroke.

Schlaug et al. (1999) found that in the ischemic core and in the area of infarct growth, the CBF decreased to 12% and 37% of the contralateral control region, respectively. Compared with the present study, Schlaug et al. (1999) used different methods to generate CBV, CBF, and MTT maps. They did not identify the arterial input function and determined the end of the first pass bolus separately for each region (core, contralateral control region, and penumbra), whereas we used a relatively large ROI to identify a global end point for all voxels. Thus, the relative CBV values in our study and the study by Schlaug et al. (1999) are not directly comparable. It is interesting that despite the different methods to calculate CBF maps, the CBF values for the area of ischemic core [13% in our study vs. 12% in the study by Schlaug et al. (1999)] and for the zone of infarct growth (35% and 37%, respectively) are rather similar in these two studies.

In 23 cases studied within 12 hours of onset, Sorensen et al. (1999) reported that the lesion volumes on CBV maps correlated better with the eventual infarct volumes than the lesion volumes on CBF and MTT maps did. Also in our approach to predict tissue outcome, CBV and CBF had slightly different potentials to predict whether the tissue is going to proceed to infarction. According to the discriminant analysis, the specificity to predict the tissue outcome was 66% with the relCBF cutoff value of 48% and it was 82% when the relCBV cutoff value of 87% was used.

Similar to CBF, CBV shows a gradient from the periphery to the core. The relCBV values in the ischemic core, in the area of infarct growth, and in the eventually viable ischemic tissue measured from CBV maps were 25%, 69% and 113%, respectively. These gradients are probably caused by the establishment of collateral circulation and autoregulatory vasodilatation, which both seem to play an important role in preventing the ischemic tissue from proceeding to infarct in the periphery of the ischemic zone (Yonas and Pindzola, 1994). In animal stroke models, a variety of nonvasoactive drugs are able to dramatically reduce the volume of ischemic lesion without substantially changing CBF (Beauchamp and Bryan, 1998). Some nonvasoactive drugs can reach the target region through the collateral circulation. It is noteworthy that the CBV increased to 113% in the eventually viable ischemic tissue. In feline stroke model with middle cerebral artery occlusion, Maeda et al. (1993) reported increased CBV in the affected but noninfarcted cortices. Tsuchida et al. (1997) and Wu et al. (1998) reported similar results in patients with ischemic stroke.

Limitations of the study

Our study has certain limitations. First, only one patient was studied within 6 hours of onset. Whether our results are valid, considering the acute stage in which therapy is presently advocated, must be further studied. It should also be noted that we used an ROI instead of a voxel-based image analysis. Also, we defined the ROIs (the ischemic core, the area of infarct growth, and the eventually viable ischemic tissue) partly retrospectively by using information not only from the initial diffusion trace images and the initial MTT maps, but also from the diffusion trace images at 1 week, when the outcome of the tissue was known. Further studies are needed to confirm whether the tissue outcome can be predicted in the acute phase without the knowledge of the imaging finding at 1 week. In patients with ischemic stroke, CBF and CBV values provided by MR perfusion imaging have not been validated by PET. Comparison studies, with timely adequate comparison of MR imaging CBF values with PET CBF values, in stroke patients would be most helpful.

Conclusion

Combined DWI and PWI allows one to detect hemodynamically different subregions within the initial perfusion abnormality. PWI seems to have potential to replace SPECT in investigating acute ischemic stroke. PWI provides quantitative information with higher spatial resolution, and it has shown potential to predict the outcome of ischemic brain tissue.

Footnotes

Acknowledgments

The authors thank Mrs. Tuula Bruun for secretarial assistance and Minna Husso-Saastamoinen, MSc, and Tomi Kauppinen, PhD, for technical assistance.

Abbreviations used

References

Albers

(1999) Expanding the window for thrombolytic therapy in acute stroke: the potential role of acute MRI for patient selection. Stroke 30:2230–2237

Baird

Austin

McKay

Donnan

(1996) Changes in cerebral tissue perfusion during the first 48 hours of ischemic stroke: relation to clinical outcome. J Neurol Neurosurg Psychiatry 61:26–29

Baird

Benfield

Schlaug

. (1997) Enlargement of human cerebral ischemic lesion volumes measured by diffusion-weighted magnetic resonance imaging. Ann Neurol 41:581–589

Baird

Warach

(1998) Magnetic resonance imaging of acute stroke. J Cereb Blood Flow Metab 18:583–609

Bamford

Sandercock

Dennis

Burn

Warlow

(1991) Classification and natural history of clinically identifiable subtypes of cerebral infarction. Lancet 337:1521–1526

Bandettini

Jesmanowicz

Van Kylen

Birn

Hyde

(1998) Functional MRI of brain activation induced by scanner acoustic noise. Magn Reson Med 39:410–416

Baron

von Kummer

del Zoppo

(1995) Treatment of acute ischemic stroke: challenging the concept of a rigid and universal time window. Stroke 26:2219–2221

Beauchamp

Jr Bryan

(1998) Acute cerebral ischemic infarction: a pathophysiologic review and radiologic perspective. AJR Am J Roentgenol 171:73–84

De Reuck

Sieben

De Coster

Vander Eecken

(1981) Stroke pattern and topography of cerebral infarcts: a clinicopathological study. Eur Neurol 20:411–415

10.

Ernst

Chang

Itti

Speck

(1999) Correlation of regional cerebral blood flow from perfusion MRI and SPECT in normal subjects. Magn Reson Imaging 17:349–354

11.

Fisher

(1997) Characterizing the target of acute stroke therapy. Stroke 28:866–872

12.

Furlan

Marchal

Viader

Derlon

Baron

(1996) Spontaneous neurological recovery after stroke and the fate of the ischemic penumbra. Ann Neurol 40:216–226

13.

Hakim

Evans

Berger

. (1989) The effect of nimodipine on the evolution of human cerebral infarction studied by PET. J Cereb Blood Flow Metab 9:523–534

14.

Hatazawa

Shimosegawa

Toyoshima

. (1999) Cerebral blood volume in acute brain infarction: a combined study with dynamic susceptibility contrast MRI and 99mTc-HMPAO–SPECT. Stroke 30:800–806

15.

Heiss

Huber

Fink

. (1992) Progressive derangement of periinfarct viable tissue in ischemic stroke. J Cereb Blood Flow Metab 12:193–203

16.

Hoehn-Berlage

Norris

Kohno

Mies

Leibfritz

Hossmann

(1995) Evolution of regional changes in apparent diffusion coefficient during focal ischemia of rat brain: the relationship of quantitative diffusion NMR imaging to reduction in cerebral blood flow and metabolic disturbances. J Cereb Blood Flow Metab 15:1002–1011

17.

Hossmann

(1994) Viability thresholds and the penumbra of focal ischemia. Ann Neurol 36:557–565

18.

Jones

Morawetz

Crowell

. (1981) Thresholds of focal cerebral ischemia in awake monkeys. J Neurosurg 54:773–782

19.

Julin

Lindqvist

Svensson

Slomka

Wahlund

(1997) MRI-guided SPECT measurements of medial temporal lobe blood flow in Alzheimer's disease. J Nucl Med 38:914–919

20.

Karonen

Nuutinen

Kuikka

. (2000) Combined SPECT and diffusion weighted MRI as a predictor of infarct growth in acute ischemic stroke. J Nucl Med 41:788–794

21.

Karonen

Vanninen

Liu

. (1999) Combined diffusion and perfusion MRI with correlation to single-photon emission CT in acute ischemic stroke: ischemic penumbra predicts infarct growth. Stroke 30:1583–1590

22.

Kidwell

Alger

Di Salle

. (1999) Diffusion MRI in patients with transient ischemic attacks. Stroke 30:1174–1180

23.

Knight

Dereski

Helpern

Ordidge

Chopp

(1994) Magnetic resonance imaging assessment of evolving focal cerebral ischemia: comparison with histopathology in rats. Stroke 25:1252–1261

24.

Lassen

(1985) Normal average value of cerebral blood flow in younger adults is 50 mL/100 g/min. J Cereb Blood Flow Metab 5:347–349

25.

Maeda

Itoh

Ide

. (1993) Acute stroke in cats: comparison of dynamic susceptibility-contrast MR imaging with T2- and diffusion-weighted MR imaging. Radiology 189:227–232

26.

Mancuso

Karibe

Rooney

. (1995) Correlation of early reduction in the apparent diffusion coefficient of water with blood flow reduction during middle cerebral artery occlusion in rats. Magn Reson Med 34:368–377

27.

Marchal

Beaudouin

Rioux

. (1996) Prolonged persistence of substantial volumes of potentially viable brain tissue after stroke: a correlative PET-CT study with voxel-based data analysis. Stroke 27:599–606

28.

Marchal

Benali

Iglesias

Viader

Derlon

Baron

(1999) Voxel-based mapping of irreversible ischemic damage with PET in acute stroke. Brain 122:2387–2400

29.

Marks

De Crespigny

Lentz

Enzmann

Albers

Moseley

(1996) Acute and chronic stroke: navigated spin-echo diffusion-weighted MR imaging. Radiology 199:403–408

30.

Marks

Tong

Beaulieu

Albers

de Crespigny

Moseley

(1999) Evaluation of early reperfusion and IV tPA therapy using diffusion- and perfusion-weighted MRI. Neurology 52:1792–1798

31.

Minematsu

Fisher

Sotak

Davis

Fiandaca

(1992a) Diffusion-weighted magnetic resonance imaging: rapid and quantitative detection of focal brain ischemia. Neurology 42:235–240

32.

Minematsu

Sotak

Davis

Fisher

(1992b) Reversible focal ischemic injury demonstrated by diffusion-weighted magnetic resonance imaging in rats. Stroke 23:1304–1311

33.

Moody

Bell

Challa

(1990) Features of the cerebral vascular pattern that predict vulnerability to perfusion or oxygenation deficiency: an anatomic study. AJNR Am J Neuroradiol 11:431–439

34.

Moseley

Cohen

Mintorovitch

. (1990) Early detection of regional cerebral ischemia in cats: comparison of diffusion- and T2-weighted MRI and spectroscopy. Magn Reson Med 14:330–346

35.

Østergaard

Smith

Vestergaard-Poulsen

. (1998) Absolute cerebral blood flow and blood volume measured by magnetic resonance imaging bolus tracking: comparison with positron emission tomography values. J Cereb Blood Flow Metab 18:425–432

36.

Østergaard

Sorensen

Kwong

Weisskoff

Gyldensted

Rosen

(1996a) High resolution measurement of cerebral blood flow using intravascular tracer bolus passages. Part II: Experimental comparison and preliminary results. Magn Reson Med 36:726–736

37.

Østergaard

Weisskoff

Chesler

Gyldensted

Rosen

(1996b) High resolution measurement of cerebral blood flow using intravascular tracer bolus passages. Part I: Mathematical approach and statistical analysis. Magn Reson Med 36:715–725

38.

Schlaug

Benfield

Baird

. (1999) The ischemic penumbra: operationally defined by diffusion and perfusion MRI. Neurology 53:1528–1537

39.

Schumann

Touzani

Young

Morello

Baron

MacKenzie

(1998) Evaluation of the ratio of cerebral blood flow to cerebral blood volume as an index of local cerebral perfusion pressure. Brain 121:1369–1379

40.

Schwamm

Koroshetz

Sorensen

. (1998) Time course of lesion development in patients with acute stroke: serial diffusion- and hemodynamic-weighted magnetic resonance imaging. Stroke 29:2268–2276

41.

Shimosegawa

Hatazawa

Aizawa

Shouji

Hachiya

Murakami

(1997) Technetium-99m-ECD brain SPECT in misery perfusion. J Nucl Med 38:791–796

42.

Shimosegawa

Hatazawa

Inugami

. (1994) Cerebral infarction within six hours of onset: prediction of completed infarction with technetium-99m-HMPAO SPECT. J Nucl Med 35:1097–1103

43.

Sorensen

Buonanno

Gonzalez

. (1996) Hyperacute stroke: evaluation with combined multisection diffusion-weighted and hemodynamically weighted echo-planar MR imaging. Radiology 199:391–401

44.

Sorensen

Copen

Østergaard

. (1999) Hyperacute stroke: simultaneous measurement of relative cerebral blood volume, relative cerebral blood flow, and mean tissue transit time. Radiology 210:519–527

45.

Takagi

Ginsberg

Globus

. (1993) Changes in amino acid neurotransmitters and cerebral blood flow in the ischemic penumbral region following middle cerebral artery occlusion in the rat: correlation with histopathology. J Cereb Blood Flow Metab 13:575–585

46.

Takano

Tatlisumak

Formato

. (1997) Glycine site antagonist attenuates infarct size in experimental focal ischemia: postmortem and diffusion mapping studies. Stroke 28:1255–1262

47.

Tatlisumak

Takano

Meiler

Fisher

(1998) A glycine site antagonist, ZD9379, reduces number of spreading depressions and infarct size in rats with permanent middle cerebral artery occlusion. Stroke 29:190–195

48.

Tsuchida

Yamada

Maeda

. (1997) Evaluation of periinfarcted hypoperfusion with T2*-weighted dynamic MRI. J Magn Reson Imaging 7:518–522

49.

Tsuchida

Nishizawa

Yonekura

. (1994) SPECT images of technetium-99m-ethyl cysteinate dimer in cerebrovascular diseases: comparison with other cerebral perfusion tracers and PET. J Nucl Med 35:27–31

50.

Ueda

Sakaki

Yuh

Nochide

Ohta

(1999) Outcome in acute stroke with successful intra-arterial thrombolysis and predictive value of initial single-photon emission-computed tomography. J Cereb Blood Flow Metab 19:99–108

51.

Walovitch

Cheesman

Maheu

Hall

(1994) Studies of the retention mechanism of the brain perfusion imaging agent 99mTc-bicisate. J Cereb Blood Flow Metab 14(suppl 1):S4–S11

52.

Watanabe

Takagi

Aoki

Sassa

(1999) Prediction of cerebral infarct sizes by cerebral blood flow SPECT performed in the early acute stage. Ann Nucl Med 13:205–210

53.

Bruening

Berchtenbreiter

. (1998) MRI assessment of cerebral blood volume in patients with brain infarcts. Neuroradiology 40:496–502

54.

Yonas

Pindzola

(1994) Physiological determination of cerebrovascular reserves and its use in clinical management. Cerebrovasc Brain Metab Rev 6:325–340