Low Cerebral Blood Volume Is Predictive of Diffusion Restriction Only in Hyperacute Stroke
Background and Purpose—Diffusion-weighted MRI (DWI) demonstrates ischemic tissue with high sensitivity. Although low cerebral blood volume (CBV) is also used as a marker for infarction, the quantitative relationship between diffusion abnormalities and CBV is unknown. We tested the hypothesis that CBV would decrease proportionally to the apparent diffusion coefficient in patients with acute stroke and thus could be used as a surrogate parameter for diffusion restriction.
Methods—Perfusion-weighted imaging and DWI was performed in 54 patients within 28 hours of symptom onset. Mean apparent diffusion coefficient, cerebral blood flow, and CBV were measured within DWI lesions and contralateral regions.
Results—Within DWI lesions, CBV (3.3±1.9 mL/100 g) was significantly decreased relative to contralateral regions (4.1±2.1 mL/100 g, P<0.001). Relative CBV was not decreased in patients with evidence of early reperfusion (1.2±0.5) or mild stroke (National Institutes of Health Stroke Scale <4, 1.1±0.6). Linear regression indicated that relative CBV was predictive of relative apparent diffusion coefficient only in patients imaged within 9 hours of symptom onset (R=0.50, P=0.02). Ischemic tissue volumes generated using a CBV threshold of the 50th percentile of normal tissue were correlated with DWI lesion volumes (R=0.73, P<0.001). The mean difference between the CBV threshold of the 50th percentile of normal tissue and DWI lesion volumes was 6.3 mL (95% limits of agreement, 0.1 to 12.6 mL).
Conclusions—Decreases in relative CBV are predictive of diffusion abnormalities in ischemic stroke. The pattern of CBV changes varies with clinical severity and symptom duration. Ischemic tissue volumes comparable to DWI lesions can be generated using CBV thresholds, but the use of this method is limited in patients with minor stroke.
Diffusion (DWI) and perfusion-weighted (PWI) MRI are highly sensitive for acute ischemic changes. DWI identifies regions of bioenergetic compromise,1–3 whereas PWI provides cerebral blood flow and volume data. Although acute diffusion restriction does not predict infarction with perfect specificity, DWI lesions in most cases represent irreversibly injured tissue.4–6 Mismatch between a smaller volume of compromised tissue visualized with DWI and larger perfusion deficit demonstrated by PWI can be used as an operational definition of the ischemic penumbra.7 Although this hypothesis remains unproven, mismatch does appear to predict treatment response.
In many centers, access issues have limited the use of MRI in acute stroke. Increasingly, CT perfusion (CTP) is being used in lieu of MRI. Unlike MRI, CTP does not provide a direct marker of bioenergetic failure. However, cerebral blood volume (CBV) maps may provide an alternative marker of tissue irreversibly destined for infarction. Under ischemic conditions, as cerebral perfusion pressure falls, an initial homeostatic response is vasodilatation, facilitating increased oxygen extraction and also resulting in an increase in CBV. In tissue where vasodilatation is insufficient to maintain tissue perfusion, CBV will fall. Decreased CBV is associated with irreversible injury and ultimately infarction.8 Thus, low CBV may be used as a surrogate for diffusion restriction in acute stroke.
Although CBV thresholds for infarction have been identified, the precise relationship between diffusion changes and CBV is unknown.9 CBV measurements are used as a surrogate for irreversible tissue injury in many centers despite a lack of systematic studies comparing CBV with other measures of bioenergetic compromise (ie, DWI). We therefore undertook a comparative MRI study examining the relationship between CBV and diffusion characteristics in patients with acute stroke. We aimed to identify relative CBV thresholds for diffusion restriction. We also tested the hypothesis that the apparent diffusion coefficient (ADC) and CBV within ischemic regions are directly correlated.
All patients were part of an ongoing study of the use of DWI and PWI in ischemic stroke. The current study was a retrospective evaluation of this database. The protocol was approved by the local Human Research Ethics Board and informed consent was obtained in all cases. Eligible patients had ischemic stroke diagnosed by a stroke neurologist, were >18 years old, and had no contraindications to MRI. Patients all had visible DWI lesions on the acute scan.
All patients were imaged with DWI and PWI using an 8-channel phased array radiofrequency head coil (MRI Devices, Waukesha, Wisc) on a 1.5-T whole-body Siemens Sonata MRI scanner (Siemens Medical Systems, Erlangen, Germany). DWI was acquired with single-shot spin-echo diffusion echoplanar imaging, 220-mm field of view, 19 5-mm axial slices with a 1.5-mm gap, b value of 1000 s/mm2 along 3 orthogonal directions, repetition time/echo time 1630/50 msec, GRAPPA R=2, and matrix size of 128×128 zero-filled to 256×256. PWI images were obtained at 60 time points per slice using single-shot gradient-echo echoplanar imaging with intravenous injection of Magnevist (Bayer HealthCare Pharmaceuticals), Gd-DTPA, at the rate of 5 mL/s, 13 to 19 5-mm axial slices with a 1.5-mm gap, repetition time/echo time 1320/50msec, and matrix size and slice orientation identical to DWI. Axial turbo spin-echo T1- and T2-weighted images were acquired; 19 slices, repetition time/echo time 680/17 msec (T1-weighted), 5800/99 msec (T2-weighted), matrix size of 128×256 (T1-weighted), 204×256 (T2-weighted), and field of view and slice orientation identical to DWI. Fluid-attenuated inversion recovery, gradient-echo, and time-of-flight angiography images were also acquired as part of the 20-minute stroke protocol.
Postprocessing of raw images was performed by a single investigator (M.E.K.). Perfusion DICOM files were imported into custom Matlab 7.4 (The Mathworks) software (PGUI Perfusion Analysis Software, CFIN Aarhus University Hospital, 2007). Maps of time to peak of the tissue response were generated. An arterial input function was manually selected from the middle cerebral artery contralateral to the DWI lesion and used to calculate deconvolved maps of cerebral blood flow (CBF) and CBV.10–12 PWI and DWI images were coregistered using statistical parametric mapping (SPM8b; Wellcome Trust Centre for Neuroimaging, London, UK).
Region of Interest Analysis
Coregistered images were imported into Analyze 8.1 for region of interest (ROI) analysis.13 The DWI lesion was defined using a semiautomated threshold intensity technique. Mirrored ROIs were manually drawn over contralateral homologous regions. Mean ADC values were measured within each ROI. Voxels with ADC >120×10−5 mm2/s were considered to contain cerebrospinal fluid and were excluded.6,14 Relative (r) ADC was calculated as the ratio of ipsilateral to contralateral values.
Normal white matter within the contralateral centrum semiovale on CBV and CBF maps was normalized to 2 mL/100 g and 22 mL/100 g/min, respectively.15,16 Voxels with CBV >8 mL/100 g or CBF >100 mL/100 g/min were assumed to contain blood vessels and removed from the ROIs.17 Relative CBF and CBV were calculated as the ratio of CBF or CBV within the DWI lesion to that of contralateral homologous regions. Perfusion status at the time of MRI was assessed using rCBF. An rCBF >0.8 (ie, blood flow within the DWI lesion at least 80% of normal) was deemed to represent reperfusion.
T1-weighted images acquired in all patients were segmented using a tissue probabilistic map of gray (GM) and white matter (WM) included in SPM8b. The DWI ROIs, and contralateral homologous regions, were then subdivided into GM and WM regions. All ROIs were transferred to coregistered PWI maps to calculate mean absolute and rCBF and CBV values within the ischemic lesions.18
Ischemic regions were also defined using CBV thresholds based on percentiles of normal contralateral tissue. Manually drawn ROIs of all visually apparent abnormal regions seen on time-to-peak maps were superimposed on CBV maps. Within each patient, normal CBV was calculated as that of the mean of the contralateral hemisphere. CBV values corresponding to cut points of the 5th, 10th, 25th, 50th, 75th, 90th, and 95th percentiles of these normal values were then calculated. These percentile values were then used to measure CBV-defined tissue infarct volumes within the oligemic region identified on time-to-peak maps.
Statistical analysis was performed using SPSS 17.0. Linear regression was used to assess the relationship between ADC and CBV. Paired t tests were used to compare DWI and PWI measures between the ischemic and unaffected hemispheres. Given the heterogeneous composition of the patient population, a number of a priori-defined groups was analyzed. Patients were analyzed with respect to time between symptom onset and MRI (<9 hours versus >9 hours), stroke severity (National Institutes of Health Stroke Scale <4 versus ≥4), and perfusion status (rCBF >0.8 versus <0.8). One-way analysis of variance and post hoc Tukey tests were used to compare DWI and PWI measures between patients in these groups. The limits of agreement between ADC- and CBV-defined ischemic volumes were calculated and illustrated using Bland-Altman plots.
Baseline Clinical Data
There were 103 patients with acute ischemic stroke enrolled. Fourteen patients were excluded due to an absence of DWI changes. Another 35 patients were excluded due to absent or uninterpretable PWI data. A total of 54 patients were included in the final analysis (36 males and 18 females; age 71.1±12.1; age range, 41 to 93 years). The median National Institutes of Health Stroke Scale was 6 (range, 0 to 27). Sixteen patients (30%) had minor stroke (National Institutes of Health Stroke Scale <4). The median time to MRI was 13 hours (range, 0.5 to 28 hours) and 21 patients (39%) were scanned within 9 hours of symptom onset. Twenty-two patients (41%) showed evidence of reperfusion at the time of MRI, whereas 32 had persisting hypoperfusion.
CBV Within DWI Lesions
Within DWI lesions, mean CBV in all patients was 3.3±1.9 mL/100 g and 4.1±2.1 mL/100 g (P<0.001) in contralateral regions. A number of CBV patterns were evident (Figure 1). In patients with major stroke (National Institutes of Health Stroke Scale ≥4), mean CBV was significantly lower in areas of diffusion restriction (2.88±1.43 mL/100 g, n=38) relative to contralateral homologous regions (3.2±1.14 mL/100 g, P<0.001). This pattern was not seen in patients with clinically minor stroke (National Institutes of Health Stroke Scale <4), in which mean CBV was not reduced in ischemic regions (3.48±1.61 versus 3.43±1.72 mL/100 g, P=0.84, n=16). Mean DWI lesion volume in patients with minor stroke (6.7±7.3 mL) was significantly smaller than in patients with major stroke (17.7±2.9 mL, P=0.018).
Mean rCBV within the DWI lesions was significantly lower in patients with persisting hypoperfusion (0.79±0.26, n=32) than in those with reperfusion (1.16±0.52, n=22, P=0.029). An elevation of CBV (rCBV >1.0) was seen in 15 of 22 (68%) of reperfused patients and only 4 of 32 (13%) of those with persisting hypoperfusion. In contrast, rADC was decreased in all patients, although it was lower in those with persisting hypoperfusion (0.69±0.09) than those with reperfusion (0.79±0.10, P=0.001).
Effect of Time Between Symptom Onset and MRI
The relationship between ADC and CBV was assessed with linear regression. There was not a significant relationship between rCBV and rADC (R=0.20, P=0.30) when all patients were assessed together. Similarly when GM and WM ADC and CBV were assessed separately, no significant relationship was found (GM: R=0.28; P=0.087; WM: R=0.04; P=0.77). Only in patients imaged <9 hours after symptom onset (n=21) was there a significant relationship between rADC and rCBV (R=0.50, P=0.02; Figure 2). Within the <9-hour group, this relationship remained significant in GM (R=0.51, P=0.025; 95% CI, 0.022 to 0.295; n=19, 2 patients had only WM lesions). The relationship was not significant in WM (R=0.42, P=0.056; 95% CI, −0.002 to 0.168; Figure 3). In patients imaged >9 hours after onset, there was no longer a relationship between rADC and rCBV (R=0.05, P=0.81).
No significant relationship between rCBV and rADC was demonstrated in minor (R=−0.01, P=0.85) or major stroke groups (R=0.04, P=0.49). In addition, no clear relationship between rCBV and rADC was found in patients with persisting hypoperfusion (R=0.09, P=0.63) or those who were reperfused at the time of scanning (R=0.06, P=0.78).
Low CBV-Defined Ischemic Lesions
Visible time-to-peak deficits were present in 46 patients. The lesion volumes defined by low CBV in these patients were significantly correlated with DWI lesion volumes (Table 1). The mean differences with 95% limits of agreement between DWI and CBV-defined ischemic lesion volumes ranged from −11.8 (−1.7 mL; 5th CBV percentile) to 40.3 (24.6, 56.0 mL; 95th percentile). Low CBV cut points tended to underestimate DWI lesion volumes (Table 2). Agreement was reasonable at the 50th percentile CBV cut point, which resulted in a mean overestimation of ADC lesion volume by 6.3 (0.1, 12.6) mL (Figure 4). At all CBV percentile cut points there was a tendency in small infarcts for CBV-defined lesion volumes to overestimate DWI volumes. Conversely, CBV-defined volumes underestimated ADC volumes in larger infarcts.
This study confirms that low CBV is correlated with diffusion restriction in patients with acute stroke. With increasing time from symptom onset, however, the relationship between ADC and CBV becomes uncoupled. Smaller lesions visible on DWI, particularly those restricted to WM, may not be evident on CBV maps. These findings are relevant to future studies in which CBV is used as a surrogate for DWI changes.
Physiological Significance of CBV Changes
The patterns of CBV changes in the 54 patients with acute stroke studied were extremely variable. The majority of patients demonstrated decreased CBV within bioenergetically compromised tissue, as demonstrated by DWI. The largest decreases were seen in patients with more severe clinical deficits. Many patients with DWI lesions had no corresponding regions of depressed CBV. This discordance between CBV and ADC changes appears to be related to 3 factors: stroke severity, tissue type, and time from onset.
Patients with milder clinical syndromes had smaller DWI lesions. In many of these patients, a corollary lesion was not evident on CBV maps (Figure 1). Thus, CBV appears to be a less sensitive marker for ischemic injury. It is possible that CBV normalized after reperfusion in these cases. Alternatively, many of these milder strokes were restricted to WM, where the relationship between ADC and CBV is not apparent.
There is only 1 other published comparison of ADC and CBV,19 which included only 13 patients, all of whom had large cortical infarcts imaged within 3 hours of symptom onset. The authors reported strong correlations between the volume of tissue with decreased CBV, calculated by both PWI and CTP techniques. In our larger, more heterogeneous population, we have demonstrated that the relationship between CBV and ADC is less predictable.
Normalization or elevation of rCBF within the DWI lesion indicated that reperfusion had occurred in a number of patients by the time of the MRI scan (Figure 1). This reperfusion is nonnutritive and is not associated with tissue salvage, because infarction has already occurred.20 During an acute cerebrovascular syndrome, perfusion is a dynamic process. Spontaneous reperfusion is more likely to be seen with longer intervals between symptom onset and PWI acquisition. In contrast to PWI changes, ADC is an “all-or-none” phenomenon. The ADC within oligemic areas decreases once an ischemic threshold is reached and with few exceptions4,21,22 does not normalize irrespective of perfusion changes. This is not the case with perfusion and the resulting dissociation between ADC and CBV becomes more evident with increasing duration of symptoms.
CBV Versus ADC in GM and WM
When GM and WM were assessed independently, rADC was significantly correlated with rCBV only in GM. This may be related to the fact that GM is associated with higher infarction thresholds than WM.23,24 In addition, CBV in normal WM is already lower than that in GM, making any relative decreases less apparent.
CBV Thresholds for Diffusion Restriction
We attempted to identify CBV thresholds, based on percentile ranks of normal tissue, predictive of DWI lesion volumes. Although a CBV 50th percentile-defined lesion volume was on average predictive of areas of ADC restriction, there was a significant amount of variability between patients. Thus, identification of 1 CBV threshold reliably indicating infarct core was problematic. It is unlikely that any 1 CBV threshold for DWI changes will be identified. It has already been demonstrated using CTP that the absolute CBV threshold for infarction does vary between patients.9
The current study has several limitations. Like in all ROI-based studies, our results are complicated by lesion heterogeneity. Both ADC25 and CBV18 may vary within a single ROI. Thus, mean values within our DWI-defined lesions may not reflect the true relationship between ADC and CBV within individual voxels. In addition, because PWI acquisition is based on nondiffusible tracer kinetics, low-flow states such as in the ischemic core may result in erroneous CBV estimates.16 To ensure we captured the entire rise and fall of contrast tissue concentration curves, we acquired data over a 98-second scan time. Any scan displaying a tissue concentration curve not returning to baseline was excluded from analysis. MRI data were only obtained at 1 time point. Multiple PWI acquisition times are required to accurately characterize the temporal pattern of changes in CBV and ADC after stroke. Finally, because CTP-derived CBV is often used clinically as a surrogate for DWI restriction when MRI is not immediately available, the current study should be repeated using contemporaneous MRI and CTP.
Decreases in rCBV are predictive of diffusion abnormalities in acute ischemic stroke. The pattern of CBV changes varies with clinical severity and symptom duration. Our findings support the continued use of thresholded CBV maps as a surrogate for ischemic injury when DWI is unavailable. In patients presenting with a delay from symptom onset, however, normalization of CBV may result in underestimation of the extent of irreversibly injured tissue.
Sources of Funding
K.B. is supported by salary and grant-in-aid awards from the Heart and Stroke Foundations of Canada, Alberta, NWT, and Nunavut, the Alberta Heritage Foundation for Medical Research and the Canadian Institutes of Health Research.
- Received May 14, 2010.
- Revision received June 29, 2010.
- Accepted July 29, 2010.
Knight RA, Dereski MO, Helpern JA, Ordidge RJ, Chopp M. Magnetic resonance imaging assessment of evolving focal cerebral ischemia: comparison with histopathology in rats. Stroke. 1994; 25: 1252–1262.
Oppenheim C, Grandin C, Samson Y, Smith A, Duprez T, Marsault C, Cosnard G. Is there an apparent diffusion coefficient threshold in predicting tissue viability in hyperacute stroke? Stroke. 2001; 32: 2486–2491.
Loh PS, Butcher KS, Parsons MW, MacGregor L, Desmond PM, Tress BM, Davis SM. Apparent diffusion coefficient thresholds do not predict the response to acute stroke thrombolysis. Stroke. 2005; 36: 2626–2631.
Schlaug G, Benfield A, Baird AE, Siewert B, Lovblad KO, Parker RA, Edelman RR, Warach S. The ischemic penumbra: operationally defined by diffusion and perfusion MRI. Neurology. 1999; 53: 1528–1537.
Koenig M, Kraus M, Theek C, Klotz E, Gehlen W, Heuser L. Quantitative assessment of the ischemic brain by means of perfusion-related parameters derived from perfusion CT. Stroke. 2001; 32: 431–437.
Murphy BD, Fox AJ, Lee DH, Sahlas DJ, Black SE, Hogan MJ, Coutts SB, Demchuk AM, Goyal M, Aviv RI, Symons S, Gulka IB, Beletsky V, Pelz D, Hachinski V, Chan R, Lee TY. Identification of penumbra and infarct in acute ischemic stroke using computed tomography perfusion-derived blood flow and blood volume measurements. Stroke. 2006; 37: 1771–1777.
Hjort N, Butcher K, Davis S, Kidwell CS, Koroshetz WJ, Rother J, Schellinger PD, Warach S, Ostergaard L. Magnetic resonance imaging criteria for thrombolysis in acute cerebral infarct. Stroke. 2005; 36: 388–397.
Shih LC, Saver JL, Alger JR, Starkman S, Leary MC, Vinuela F, Duckwiler G, Gobin YP, Jahan R, Villablanca JP, Vespa PM, Kidwell CS. Perfusion-weighted magnetic resonance imaging thresholds identifying core, irreversibly infarcted tissue. Stroke. 2003; 34: 1425–1430.
Arakawa S, Wright P, Koga M, Phan T, Reutens D, Lim I, Gunawan M, Ma H, Perera N, Ly J, Zavala J, Fitt G, Donnan G. Ischemic thresholds for gray and white matter: a diffusion and perfusion magnetic resonance study. Stroke. 2006; 37: 1211–1216.
Butcher KS, Parsons M, MacGregor L, Barber PA, Chalk J, Bladin C, Levi C, Kimber T, Schultz D, Fink J, Tress B, Donnan G, Davis S; for the EPITHET Investigators. Refining the perfusion–diffusion mismatch hypothesis. Stroke. 2005: 1153–1159.
Parsons M, Pepper E, Bateman G, Wang Y, Levi C. Identification of the penumbra and infarct core on hyperacute noncontrast and perfusion ct. Neurology. 2007; 68: 730–736.
Schaefer P, Ozsunar Y, He J, Hamberg L, Hunter G, Sorensen A, Koroshetz W, Gonzalez R. Assessing tissue viability with MR diffusion and perfusion imaging. AJNR Am J Neuroradiol. 2003; 24: 436–443.
Wintermark M, Reichhart M, Cuisenaire O, Maeder P, Thiran JP, Schnyder P, Bogousslavsky J, Meuli R. Comparison of admission perfusion computed tomography and qualitative diffusion- and perfusion-weighted magnetic resonance imaging in acute stroke patients. Stroke. 2002; 33: 2025–2031.
Kidwell CS, Saver JL, Mattiello J, Starkman S, Vinuela F, Duckwiler G, Gobin YP, Jahan R, Vespa P, Villablanca JP, Liebeskind DS, Woods RP, Alger JR. Diffusion–perfusion MRI characterization of post-recanalization hyperperfusion in humans. Neurology. 2001; 57: 2015–2021.
Fiehler J, Knudsen K, Kucinski T, Kidwell CS, Alger JR, Thomalla G, Eckert B, Wittkugel O, Weiller C, Zeumer H, Rother J. Predictors of apparent diffusion coefficient normalization in stroke patients. Stroke. 2004; 35: 514–519.
Leenders KL, Perani D, Lammertsma AA, Heather JD, Buckingham P, Healy MJ, Gibbs JM, Wise RJ, Hatazawa J, Herold S, Buckingham P, Healy MJR, Gibbs JM, Wise RJS, Hatawaza J, Herold S, Beaney RP, Brooks DJ, Spinks T, Rhodes C, Frackowiak RSJ, Jones T. Cerebral blood flow, blood volume and oxygen utilization. Normal values and effect of age. Brain. 1990; 113: 27–47.
Wu O, Koroshetz WJ, Ostergaard L, Buonanno FS, Copen WA, Gonzalez RG, Rordorf G, Rosen BR, Schwamm LH, Weisskoff RM, Sorensen AG. Predicting tissue outcome in acute human cerebral ischemia using combined diffusion- and perfusion-weighted MR imaging. Stroke. 2001; 32: 933–942.