| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
(Stroke. 1995;26:667-675.)
© 1995 American Heart Association, Inc.
Articles |
Sensitivity of Magnetic Resonance Diffusion-Weighted Imaging and Regional Relationship Between the Apparent Diffusion Coefficient and Cerebral Blood Flow in Rat Focal Cerebral Ischemia
From the Departments of Biomedical Engineering (A.D.P.-T., S.C.J.), Neurology (A.J.F.), Neurosurgery (I.A.A.), and Radiology (M.X., T.C.N., A.W.M.), Cleveland Clinic Foundation, Cleveland, Ohio.
Correspondence to Stephen C. Jones, PhD, Cerebrovascular Research Laboratory, NC30, Department of Biomedical Engineering, Cleveland Clinic Foundation, 9500 Euclid Ave, Cleveland, OH 44195-5286.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Background and Purpose Magnetic resonance (MR) diffusion-weighted imaging (DWI), a noninvasive procedure, may play an important role in detecting and accurately localizing the extent of evolving infarction within the period immediately following stroke. We evaluated the sensitivity and specificity of DWI in detecting ischemia and compared a quantitative measure derived from the DWI, the apparent diffusion coefficient (ADC), with autoradiographic cerebral blood flow (CBF) in an experimental model of focal cerebral ischemia in rats.
Methods MR imaging data were obtained with a General Electric 4.7-T horizontal bore magnet CSI II system with self-shielded gradients. DWI was acquired within 41±6 minutes (mean±SD) after onset of ischemia and repeated at 169±14 minutes, followed by CBF determination at 237±21 minutes. DWI, ADC, and CBF images from each animal were then compared.
Results The sensitivities for detecting an abnormality at 1 and 3 hours for DWI were significantly different, and the sensitivity of 3-hour DWI did not differ from the CBF sensitivity of 99%. A mean±SD ADC threshold of 460±95 µm2/s was defined as 45% higher than the low ADC in the ischemic core compared with the contralateral ADC. Subthreshold ADC area and ischemic area were significantly correlated (r2=.69, P<.05). In 19 of 48 regions of interest classified as ischemic (<35 mL · 100 g-1 · min-1) from both the 3-hour ADC and CBF images, 3-hour ADC correlated significantly with CBF (r2=.27, n=19, P<.05), whereas in the nonischemic regions ADC was inversely correlated with CBF. Several ischemic regions showed a sharp drop in ADC to 37% (P<.001, n=5) compared with all other regions (n=43) from 1 to 3 hours.
Conclusions Because of the change in the sensitivity of detecting ischemia with DWI, the difference in correlation of CBF with ADC between ischemic and nonischemic cortex, and the presence of several regions in which ADC dropped to 37% from 1 to 3 hours, our data suggest that ADC values potentially can be used to monitor evolving infarction.
Key Words: cerebral blood flow • cerebral ischemia, focal • diagnosis • magnetic resonance imaging, diffusion-weighted • rats
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The time window for effective treatment of cerebral ischemia is 4 hours or less in monkeys1 and rats.2 A reliable diagnostic study that can confirm and determine the extent of ischemia has not been available during these critical early hours for clinical stroke trials that use this short a therapeutic window.3 4 Conventional T2-weighted magnetic resonance (MR) imaging does not fully reflect the extent of ischemic changes in the first 12 hours from stroke onset.5
Diffusion-weighted imaging (DWI) is a relatively new MR modality that is sensitive to the microscopic motion of water molecules and permits quantitation of this phenomenon through the calculation of apparent diffusion coefficient (ADC) images.6 7 ADC may be calculated, on a pixel-by-pixel basis, from at least two images with different diffusion weighting. The potential role of DWI in stroke is based on the observation that the apparent diffusion rate of water protons in ischemic brain is much lower than in normal brain. DWI has been used to demarcate very early (within 148 and 30 minutes9 ) changes in the evolution of cerebral ischemia in animals10 11 12 13 and in humans.14
In addition, it has been shown by Busza et al15 in the gerbil that at a cerebral blood flow (CBF) below 15 to 20 mL · 100 g-1 · min-1, the signal-intensity ratio of DWI suddenly increases. Similarly, Roberts et al16 have provided evidence of a threshold by comparing contralateral with ipsilateral ADC ratios and an MR index of blood volume in a cat model of partial middle cerebral artery (MCA) stenosis at 1 and 6 hours. At 7 hours after MCA occlusion, Back et al17 compared ADC with pH, lactate, and ATP distributions, reaching the conclusion that all of the parameters are highly correlated at this period.
Our strategy in this study was to focus on areas that these studies have not explored. First, with objective image observation by blinded observers and statistical analysis, we determined the sensitivity and specificity of DWI for detecting ischemia produced with an embolus model of focal ischemia in rats, using a clinical radiology paradigm. Then, we used image analysis of ADC and CBF in regions of interest (ROIs) that included ischemic, normal cortex, and intermediate areas to quantitatively explore the relationship between regional ADC and autoradiographic CBF, using the ischemic threshold18 to define those regions that will eventually infarct.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Fourteen Sprague-Dawley rats weighing 355±27 g (mean±SD) underwent tracheotomy and were mechanically ventilated (model 683, Harvard Apparatus). Anesthesia was induced with 2% to 3% halothane in a mixture of 70% N2O and 30% O2. For the CBF study, a maintenance dose of 0.5% to 1.0% halothane was adjusted to minimize abrupt changes in heart rate, mean arterial pressure, respiratory rate, or pupillary diameter in response to tail pinch. Both femoral arteries and veins were cannulated with PE-50 polyethylene catheters. During surgery and the CBF study, body temperature was maintained at 37°C with a servo-controlled heating lamp and a rectal thermistor (YSI 74). Mean arterial pressure was continuously monitored from a femoral artery with a strain-gauge transducer (MP-15D, Micron). Heart rate and end-tidal CO2, as determined with an infrared CO2 analyzer (model 223, Puritan-Bennett), were monitored. PaO2, PaCO2, and pH levels were determined with a blood gas analyzer (ABL-3, Radiometer). All procedures were approved by the institutional committee dealing with animal research.
Ten animals underwent cerebral embolization with a single silicone cylinder (1 mm long, 300 µm in diameter). Under an operating microscope, the left internal carotid artery was exposed, and the pterygopalatine branch was electrocauterized and sectioned. The left common carotid artery was permanently occluded with a ligature, and a single silicone cylinder in saline was infused through the external carotid into the internal carotid.19 Because of the variable location of the infarct, this model permitted testing of sensitivity and specificity.
Four additional animals were used to determine the change of T2 from 1 to 3 hours after ischemia. The MCA was coagulated from 2 mm proximal to the olfactory tract to the inferior cerebral vein by the subtemporal route.20 21 In addition, both common carotid arteries were isolated, coagulated, and transected.22 This model permitted the estimation of the T2 changes from 1 to 3 hours.
During DWI acquisitions, the rats remained anesthetized (isoflurane 1% with 70% N2O, balance O2) and paralyzed (gallamine triethiodide, 10 mg · h-1 · kg-1), and mean arterial pressure was continuously monitored. Body temperature was maintained constant with isothermal pads, and a modified rodent ventilator was used to deliver the anesthesia during the MR study. The paralyzing agent was stopped 15 minutes before the end of the last MR data acquisition.
DWI measurements were performed with a General Electric 4.7-T CSI II unit equipped with Acustar self-shielded gradient coils. A 1H slotted tube resonator probe (diameter, 35 mm; length, 55 mm) of high B1 homogeneity was constructed for this purpose.23 DWIs were collected with a repetition time of 3 seconds and an echo time of 80 milliseconds, and 10-millisecond diffusion gradient pulses applied in the "y" direction (vertical or dorsoventral anatomic axis) with a 30-millisecond pulse interval. The gradient strength for DWI scans was set at 10.44 G/cm (b=2393 s/mm2). The scan time for each DWI data set (four coronal sections) was approximately 15 minutes. The slice thickness (z axis) was 3 mm, and the field of view was 65x65 mm. Phase encodings (x axis) of 128 steps were used in the acquisition, and the data were zero-filled to 256x256 before Fourier transformation. Immediately after the first series of DWI, T2-weighted images were acquired with the same spin-echo sequence but with the gradient pulse set to zero. A second DWI series was obtained approximately 3 hours after stroke onset. In an additional 4 rats with MCA occlusion, T2 images using the same parameters were collected approximately 1 and 3 hours after ischemia.
ADC images were calculated from DWIs and T2-weighted images according to the method described by Le Bihan et al.6 7 Using this method, the ADC is the negative of the slope of the natural logarithm of the DWI intensity plotted as a function of the gradient factor, b. The 1-hour ADC image (ADC1) was calculated for each pixel as follows: ADC1=-ln(S1b/S10)/b=ln(S10 /S1b)/b, where b is the gradient value (2393 s/mm2), S0 is the signal intensity at b=0, and Sb is the signal intensity at b=2393 s/mm2.
The 3-hour ADC image (ADC3) was calculated as follows: ADC3=ln(S10/S3b)/b, except for the ADC in ischemic cortex, ADC3(ICO), which was corrected using ADC3(ICO)=ADC3(ICO, uncorrected)+ln(SIR30/SIR10)/b, where SIR10 and SIR30 are the T2 signal-intensity ratios (ischemic cortex signal intensity/contralateral cortex signal intensity) 1 and 3 hours after ischemia, obtained from a separate group of animals. In the ischemic cortex the T2 signal-intensity ratios 1 and 3 hours after ischemia in 4 rats were 1.06±0.07 and 1.17±0.06, respectively, reflecting a 10% increase.
After the 3-hour DWI, animals were transported from the MR facility to the laboratory where CBF was measured. Before the brief transport period, a single intravenous dose of 13 mg/kg pentobarbital was administered. On arrival at the CBF laboratory, halothane and N2O anesthesia was reinstituted. The mean interval between pentobarbital administration and the CBF study was 55±10 minutes. The low pentobarbital dose and the length of time between administration and CBF study indicate that the CBF measurement was not influenced by pentobarbital.24 When combined with N2O, the cerebrovascular effects of isoflurane, used during MRI, and halothane, used during CBF determination, do not differ.25
CBF was measured with [14C]iodoantipyrine (100 µCi/kg).26 Briefly, a background arterial blood sample was taken, and [14C]iodoantipyrine was infused continuously for 45 seconds into the femoral vein. Multiple arterial blood samples were collected. These blood samples were analyzed for 14C using liquid scintillation counting. After 45 seconds the animal was decapitated. The brain was promptly removed, frozen immediately in chlorodifluoromethane (-44°C), and stored in a freezer (-80°C). Coronal sections, 20 µm thick, were obtained in a cryostat at -20°C. Every 10th section was dried on a 60°C hot plate and exposed for 4 days to x-ray film (Kodak SB5) together with eight precalibrated [14C]methyl methacrylate standards.
CBF autoradiograms were digitized using a quantitative image analysis system (MCID, Imaging Research). CBF images were produced by first converting optical density to 14C concentration using the [14C]methyl methacrylate standards and then to CBF using the 14C arterial concentration versus time and the method of Sakurada et al.26
Sensitivity and specificity were tabulated by blinded visual assessment of 1- and 3-hour DWI, T2-weighted, and CBF images from autoradiograms. Four blinded observers graded the presence of hyperintense areas from the DWIs or T2-weighted images in each set of four MR images. Similarly, the four anatomically matching CBF autoradiograms were graded for the presence of areas with hypoperfusion. Grades were 0 or 1 for the absence or presence of an abnormality, respectively.
Image analysis of both the ADC and CBF images was performed using the MCID image analysis system. ADC images were converted into the image analyzer format and used to explore the ADC threshold, ROIs, and areas of low ADC.
Transects through the region of lowest ADC in the left ischemic
hemisphere and through a corresponding area in the opposite normal
hemisphere were extracted from both the 1-hour and 3-hour ADC images
(Fig 1
). These transects consisted of a plot of the ADC
versus length along a 1-mm-wide curvilinear segment drawn through the
right and left hemispheres. As presented in Fig 1, for the transect
on the ischemic side, contiguous ADC values within 1.1 times the
minimum ADC were averaged (ADCisch). The ADCs from the
corresponding contralateral profile were averaged
(ADCcontr). The ADC threshold (ADCthres) for
each animal was calculated from the following formula:
ADCthres=0.45 · (ADCcontr-ADCisch)+ADCisch.
This ADC threshold corresponds to 45% of the difference between the
average ADCs of the ischemic and contralateral profiles. The use of
45% as the difference is justified in "Results" and
"Discussion." The ADC thresholds from each section were averaged
to produce a mean ADC threshold. All values below this threshold were
considered as subthreshold ADC. This ADC threshold was used to
establish the cross-sectional area of abnormality in the ADC
images.
|
For each animal, ROIs and cross-sectional areas were determined from the 3-hour ADC image with the largest area of low ADC, the 1-hour ADC image from the same level, and the most anatomically matched CBF autoradiogram. The areas of ischemia (CBF <35 mL · 100 g-1 · min-1)27 and of subthreshold ADC were calculated from these images. Ischemic area was determined as previously described21 and was corrected for the area of edema.28 Ten ROIs were chosen: ischemic core with CBF <35 mL · 100 g-1 · min-1 or subthreshold ADC (ICO); contralateral to ischemia (ICT); left and right white matter; left and right motor cortex; left and right somatosensory cortex; and left and right caudate putamen. The CBF and ADC values were averaged by ROI. For the correlation between ADC and CBF, each individual ROI was classified as ischemic (CBF <35 mL · 100 g-1 · min-1) or nonischemic because this level of ischemia correlates with eventual neuropathology.1 18 Changes in ADC from 1 to 3 hours were assessed using the ratio of 3-hour ADC to 1-hour ADC for each ROI.
For the analysis of sensitivity and specificity, the agreement
between observers was evaluated using 
.29 is a
chance-corrected measure of agreement among observers, which equals 0
for chance agreement only and 1 for perfect agreement among observers.
Good agreement is represented if >0.5.30
Sensitivity and specificity for each method of evaluating ischemia were
estimated using a latent class model.31 This methodology
avoids the need for a "gold standard" in comparison to the tested
technique. The basic model defines the profile of stroke detection for
each technique (1-hour DWI, T2-weighted, 3-hour DWI, and
CBF) and observer as a likelihood function with two parameters, one for
sensitivity and false-positive rate (1 minus specificity) and one for
prevalence of an abnormality. These parameters are estimated using the
maximum likelihood method via the EM algorithm. The average grade among
the four readers was used to estimate sensitivity and specificity by
technique and image.
Repeated-measures ANOVA was performed to determine differences between regions in 1- and 3-hour ADC. Linear regression and the coefficient of determination, r2, were used to describe the relationship between ischemic and subthreshold ADC areas and between CBF and ADC in individual ROIs. The slope of the regression lines and r2 were tested for differences from 0 using the t test. Paired or unpaired t tests were used for all comparisons. Values are expressed as mean±SD. Statistical significance was assumed when probability values were less than .05. The statistical analyses were performed using the SAS system.32
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Arterial PaO2 and PaCO2 remained stable in these respirated animals from surgery to the CBF study. However, pH showed a minimal but significant decrease from baseline levels. Mean arterial pressure was significantly increased during the MR study, as was heart rate before the CBF determination (Table 1
).
|
Table 2 presents the grouping and characteristics of
the 10 animals used in this study. The sensitivity and specificity of
DWI for detecting ischemic regions were estimated in all 10 rats. It
was not possible to obtain 1-hour or 3-hour ADC values or images from
rat 8 and 3-hour ADCs from rat 1 because of technical errors during the
transfer of the digitized images; therefore, it was not possible to
determine these subthreshold ADC areas. Ischemic and subthreshold ADC
areas were measured in all other stroke animals. The 1-hour DWI was
started at 41±6 minutes and the 3-hour DWI at 169±14 minutes. Even
though the mean times for these images were 41 and 169 minutes,
respectively, they will be referred to as the 1-hour images and the
3-hour images. T2-weighted images were started at 73±6
minutes. Finally, CBF was measured at 237±21 minutes.
|
Fig 2 top shows the 1-hour ADC at 50 minutes with very
slight changes in signal intensity in the left cortex. Fig 2 middle
shows the 3-hour ADC at 175 minutes; a well-demarcated area with low
ADC is present in a similar location where CBF is below the
ischemic threshold, as shown in Fig 2 bottom (images from rat 5; see
Table 2
).
|
Sensitivity and specificity were assessed using four blinded observers.
Each observer graded four sets (1-hour DWI, T2-weighted,
3-hour DWI, and CBF) of four images for each of the 10 animals, except
for eight images that were not available for grading (second DWI from
rat 1 and T2 from rat 8; see Table 2). The values
ranged from a low of 0.48 to a high of 0.89 and were statistically
significant, suggesting good agreement among the four observers
(P<.001).
The sensitivities and specificities for the four techniques are
presented in Table 3. The sensitivities for 3-hour
DWI and CBF are 94% and 99%, respectively, with overlap of their 95%
confidence intervals, whereas the 56% sensitivity of 1-hour DWI
differs from the sensitivities for both 3-hour DWI and CBF. The
sensitivity of T2-weighted imaging is not statistically
different from zero. All techniques had high specificity. In the 1-hour
DWI, 5 animals showed hyperintense areas in the same location where CBF
showed hypoperfusion and 3 animals did not. In the 3-hour DWI, all 8
stroke animals manifested hyperintense areas in the same location where
CBF indicated hypoperfusion.
|
Animals with larger ischemic areas showed DWI changes at 1 hour, while
those with smaller ischemic areas displayed DWI abnormalities at 3
hours. The rats that showed a hyperintense area in the 1-hour DWI
showed a strong trend for larger ischemic areas (35±20
mm2, n=5) compared with those in whom hyperintensity
did not appear in the 1-hour DWI (7.9±2.9 mm2, n=3;
P=.06) (Fig 3).
|
The transects through the ischemic and contralateral hemispheres are
presented in Fig 1A, and the plot of these transects (ADC versus
distance along the cortical mantle) is presented in Fig 1B for rat
5 (see Table 2). The threshold for this animal, defined as 45% of the
difference between the ischemic and contralateral ADC, was 309
µm2/s. The mean ADC threshold obtained in 7
animals from seven transects from the 1-hour ADC and from six from the
3-hour ADC was 460±95 µm2/s.
In Fig 4, the correlation between ischemic area and the
areas of deficit in the 1-hour and 3-hour ADCs is plotted. Only the
3-hour subthreshold ADC area showed a significant correlation with
ischemic area (r2=.69, n=6,
P<.05), and the linear regression slope of 0.85 was not
significantly different from a slope of 1. In contrast, the 1-hour
subthreshold ADC area and the ischemic area were not correlated.
|
)
was correlated (r2=.69,
P<.05) with ischemic area (CBFa), whereas the 1-hour ADC
subthreshold area (ADC1a, 
) was not. The correlation of the area
of the 3-hour ADC and CBF lend support to the choice of the 45% level
between ischemic cortex ADC and contralateral cortex ADC as an
appropriate threshold. Since the ischemic area represents
tissue that will eventually infarct, the significant correlation and
similarity between slopes of the linear regression between the ischemic
area and the subthreshold ADC area at 3 hours suggest that the ADC
threshold can be used in the same way as the ischemic threshold to
predict infarction.
The 1-hour and 3-hour ADCs from various ROIs are presented in Fig 5. The ADCs for normal gray and white matter were 593
and 544 µm2/s, respectively. Both ischemic ADCs
differed from contralateral ADCs (ICT, P<.01). The mean
ratios of contralateral (ICT) to ischemic cortex (ipsilateral, ICO) ADC
were 67% and 54% at 1 and 3 hours, respectively.
|
P<.01 between ICO and
ICT).
ADC and CBF values from ischemic (identified as CBF <35 mL · 100
g-1 · min-1) and nonischemic gray matter
from the 3-hour ADC images (Fig 6A) were linearly
correlated, but the slope for ischemic regions was 12 and the slope for
nonischemic regions was -0.8
(µm2/s)/(mL · 100
g-1 · min-1). Thus the relationship
between ADC and CBF is different for ischemic and nonischemic
regions.
|
) and ischemic (CBF
<35 mL · 100 g-1 · min-1,
),
whereas the rest are near 1 (
). In these five selected regions,
ADC fell significantly from 1 to 3 hours after ischemia (0.37±0.13 vs
1.05±0.12, n=43, P<.001).
In Fig 6A, there are five regions with ADC3 below 200
µm2/s. When the ratio of ADC3 to ADC1 in each ROI
is expressed as a histogram (Fig 6B), these five regions form a
separate group in which the ratio of 0.37±0.13 is significantly
different (P<.001) from the mean ratio of the other regions
(1.04±0.13, n=43).
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The changing pathophysiology during the early stages of ischemia has made it difficult to predict the ultimate outcome on the basis of clinical observation alone. Angiography, computed tomography,33 single-photon emission tomography, positron emission tomography, or conventional MR34 all have limited utility in the very early diagnosis of ischemic stroke. MR is noninvasive, has high spatial and temporal resolution, allows imaging in any orientation, and is relatively inexpensive. However, conventional T2-weighted MR images do not faithfully depict the temporal and spatial progression of the parenchymal changes in the early stages (1 to 6 hours) of cerebral ischemia.5 13 15 35 In contrast, DWI demonstrates large changes within 30 minutes.9 Recently, Knight et al35 showed that neither T1, T2, nor proton density demonstrated the area of infarction in the first 28 hours when compared with histopathology, whereas DWI did. Our data represent the first time that individual quantitative ADC values have been presented without normalization and in comparison to regional CBF. Other investigators have used normalization of DWI intensity8 10 15 or ADC14 16 to a preischemic or contralateral value.
It has been postulated that DWI is primarily sensitive to either the redistribution of water from the extracellular to the intracellular compartment36 or a change in membrane permeability37 as a result of cytotoxic edema. In normal brain, the Na+/K+ pump maintains a large space of extracellular water, which has a higher diffusional coefficient than intracellular water. In ischemia, the pump is disabled and the extracellular space is decreased.38 Although the increased DWI signal may not be due to the total water increase in cerebral ischemia, shown to be less than 5%,39 it is possible that the shift of water into the intracellular compartment36 or the associated decrease in membrane permeability due to the inactivation of the Na+/K+ pump could be the cause of the decreased ADC in early ischemia.37
We used the concept of the ischemic threshold1 18 in two different ways in this work. We made the presumption, based on the correlation of the critical level of CBF and neuropathologically defined infarction documented by Tyson et al,27 that any brain tissue with a CBF <35 mL · 100 g-1 · min-1 will eventually progress to infarction and that this ischemic area is predictive of infarct area. We also classified 19 of 48 gray matter regions for the 3-hour ADC as ischemic because their CBFs were <35 mL · 100 g-1 · min-1, again with the presumption that this tissue classified as ischemic will eventually infarct.
The ADC threshold could be similar to the ischemic threshold and
possibly indicate tissue disruption and the area destined to infarct.
Busza et al15 demonstrated in the gerbil stroke model a
sharp increase in the ratio of ischemic to contralateral DWI signal
intensity at CBFs of <15 to 20 mL · 100
g-1 · min-1. Although the interpretation
of the proposed ADC threshold warrants caution and needs further
investigation, we have demonstrated that the area of tissue with a
3-hour ADC below 460 µm2/s correlates with the
ischemic area (Fig 4). The rationale for choosing 45% between ischemic
and normal ADC to define an ADC threshold is the correlation between
the ischemic and ADC area at 3 hours in Fig 4. This correlation
supports the choice of 45% as the appropriate threshold percentage.
This proposed ADC threshold at 3 hours could very well not be valid for
earlier or later times. The possibility of defining such a threshold,
based on the quantitative capacity of an ADC image, gives credence to
the prospective use of DWI in cerebral ischemia. Dardzinski et
al40 obtained a similar ADC threshold of 550
µm2/s 2 hours after MCA occlusion using the suture
model from the comparison with 24-hour neuropathology.
Our values of ADC are consistent with the results of others in normal7 36 41 42 43 and ischemic8 11 13 brain. Our data showed a drop in the ADC values (calculated as ipsilateral/contralateral) to 67% within 1 hour and 54% at 3 hours. Other investigators have observed decreased ADC ratios from ischemic to normal cortex from 38% at 33 minutes8 to 57% at 1 hour.16 These differences could be due to the severity and variability of ischemia in relation to volume of the region sampled.
Several aspects of our data support the concept that ADC is sensitive
to the progression of the pathological process, in particular the
almost linear progression of cytotoxic edema,38 during the
initial 3 hours of ischemic stroke. First, the increase in sensitivity
for detecting the ischemic deficit from 1 to 3 hours (Table 3)
indicates that smaller deficits that were not detected at 1 hour became
larger areas of DWI hyperintensity by 3 hours. Second, the difference
in correlation of CBF and ADC between ischemic gray matter regions and
nonischemic gray matter (Fig 6A) indicates that the severity of
ischemia is related to the depression of ADC. Finally, the ratio of 3-
to 1-hour ADC was severely depressed in five regions compared with all
the others (Fig 6B). Busza et al15 recently reported a
relationship in the gerbil stroke model between DWI signal-intensity
ratio and CBF below a threshold of 15 mL · 100
g-1 · min-1 that is similar to the one we
show in our rat model. If low ADC is synonymous with the spread of
cytotoxic edema, then the sharp decrease in ADC at a CBF of 35
mL · 100 g-1 · min-1, as shown
in Fig 6A, could be used as a diagnostic tool as well as an outcome
predictor in cerebral ischemia.
The five ROIs with 3-hour ADCs <200 µm2/s in Fig 6A showed a dramatic drop in ADC from 1 to 3 hours and also had low
ipsilateral, in relation to contralateral, CBF. Whereas the other ROIs
classified as ischemic did not show a drop in ADC from 1 to 3 hours,
these five points showed a mean drop of 37%, suggesting a more rapid
progression than the other ischemic ROIs.
The use of the latent class model for the evaluation of sensitivity without the use of a "gold standard" is a powerful statistical method. Using this methodology, we showed that at 1 hour DWI is less sensitive for detecting ischemia than at 3 hours and that DWI and CBF have similar sensitivities at 3 hours. The sensitivity of DWI has been judged by computed tomography14 or histopathology at 4 hours,8 6 to 8 hours,13 35 or 24 hours.9 However, comparison with pathology does not truly test whether DWI is detecting changes related to early ischemia. Our lack of early sensitivity may be related to the partial volume effect due to the 3-mm slice thickness of our DWI images44 and suggests that the pathological process that causes DWI hyperintensity may not have progressed sufficiently at the 1-hour image.
We were able to detect a low initial sensitivity of DWI because of the
inherent variability in location and extent of ischemia with our
embolus model. Other investigators who have reported early and
sensitive detection of ischemia with DWI have used the thread model in
rats,9 the MCA occlusion model in cats,13 or
naturally occurring strokes in humans14 ; all of these
insults produce a large ischemic deficit in relation to the slice
thickness. In our study, the larger ischemic areas were detected in the
1-hour DWI (Fig 3). With a large volume of ischemia, the tissue changes
that produce a hyperintense DWI response are maximized, whereas with a
small ischemic insult there is a greater capacity for collateral
circulation to reverse the ischemic deficit. The variability of
location and volume of ischemia from the embolus model is a desirable
feature for the assessment of sensitivity and specificity of DWI that
is not available with other models of ischemia.
Our protocol used the combination of a 1-hour T2-weighted image to calculate both 1- and 3-hour ADC images in normal cortex. The ADC in ischemic cortex was corrected for the slight increase in T2 in ischemic cortex from 1 to 3 hours after ischemia, which was determined to be 10% in a separate group of animals. Much evidence has been published that T2 increases slightly in the initial 3 hours after ischemic onset. In cats, the T2 signal-intensity ratio increased 7% from 1 to 3 hours from the onset of ischemia5 and 13% 8 hours after MCA occlusion,13 increases that are comparable to those in this study. Using the gerbil45 and rat46 models of focal ischemia, increases of under 10% were found in T2 between 1 and 3 hours.
Initially, our study was focused on the determination of the specificity and sensitivity of DWI. Later, we realized that we could compare regional quantitative ADC and CBF, but we knew that small changes in T2 would affect the calculation; we subsequently measured the change in T2 in an additional group of animals and used this data to calculate the 3-hour ADC in ischemic cortex.
A factor that would have resulted in slightly smaller ADC values is the rapid decrease in DWI signal intensity from gradient factors of zero to about 300 s/mm2,7 although others have observed a monoexponential, not biexponential, relationship between DWI signal intensity and b.40 47 If we had used a gradient factor, b, of 300 s/mm2 instead of b=0 in our T2-weighted image, we could have averted this possibility. However, we wanted to confirm our previous finding that T2-weighted imaging is not sensitive to early ischemic changes5 in comparison to DWI. An estimate of this error, based on the data presented by Le Bihan et al,7 shows that this effect leads to a 5% overestimation of ADC in our data.
Our results support the observation that DWI has a high degree of temporal and spatial sensitivity and specificity in the detection of cerebral ischemia, with the caveat that a certain critical volume of affected tissue appears to be necessary for early detectability. Furthermore, the greater extent of low ADC as the ischemic pathological process progresses from 1 to 3 hours, the correspondence between ADC and CBF in ischemic cortex, and the presence of a severe depression in ADC in some regions from 1 to 3 hours all suggest that ADC values can potentially be used to monitor ischemic changes in early stroke. Our initial estimate of an ADC threshold could be used as a gauge of this effect but must be confirmed by other studies.
|
Acknowledgments |
|---|
This work was supported in part by American Heart Association (National) Grant-in-Aid 91-1315, a Fellowship from the American Heart Association Northeast Ohio Affiliate, and National Institutes of Health grant NS30839. The authors wish to thank Kirk Easley for statistical consultation and Joseph Joyce for technical assistance.
Received July 11, 1994; revision received November 2, 1994; accepted December 29, 1994.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
1. Jones TH, Morawetz RB, Crowell RM, Marcoux FW, Fitzgibbon SJ, DeGirolami U, Ojemann RG. Thresholds of focal cerebral ischemia in awake monkeys. J Neurosurg. 1981;54:773-782. [Medline] [Order article via Infotrieve]
2.
Kaplan B, Brint S, Tanabe J, Jacewicz M, Wang XJ, Pulsinelli
W. Temporal thresholds for neocortical infarction in rats subjected to
reversible focal cerebral ischemia. Stroke. 1991;22:1032-1039.
3.
Brott TG, Haley EC Jr, Levy DE, Barsan W, Broderick J,
Sheppard GL, Spilker J, Kongable GL, Massey S, Reed R, Marler JR.
Urgent therapy for stroke, I: pilot study of tissue plasminogen
activator administered within 90 minutes. Stroke. 1992;23:632-640.
4.
Haley EC Jr, Levy DE, Brott TG, Sheppard GL, Wong MCW,
Kongable GL, Torner JC, Marler JR. Urgent therapy for stroke, II: pilot
study of tissue plasminogen activator administered 91 to 180 minutes
from onset. Stroke. 1992;23:641-645.
5.
Bose B, Jones SC, Lorig R, Friel HT, Weinstein M, Little JR.
Evolving focal cerebral ischemia in cats: spatial correlation of
nuclear magnetic resonance imaging, cerebral blood flow, tetrazolium
staining, and histopathology. Stroke. 1988;19:28-37.
6.
Le Bihan D, Breton E, Lallemand D, Grenier P, Cabanis E,
Laval-Jeantet M. MR imaging of intravoxel incoherent motions:
application to diffusion and perfusion in neurologic disorders.
Radiology. 1986;161:401-407.
7. Le Bihan D, Moonen CT, van Zijl PC, Pekar J, DesPres D. Measuring random microscopic motion of water in tissues with MR imaging: a cat brain study. J Comput Assist Tomogr. 1991;15:19-25. [Medline] [Order article via Infotrieve]
8. Mintorovitch J, Moseley ME, Chileuitt L, Shimizu H, Cohen Y, Weinstein PR. Comparison of diffusion- and T2-weighted MRI for the early detection of cerebral ischemia and reperfusion in rats. Magn Reson Med. 1991;18:39-50. [Medline] [Order article via Infotrieve]
9.
Minematsu K, Fisher M, Sotak CH, Davis MA, Fiandaca MS.
Diffusion-weighted magnetic resonance imaging: rapid and quantitative
detection of focal brain ischemia. Neurology. 1992;42:235-240.
10.
Minematsu K, Li L, Sotak CH, Davis MA, Fisher M. Reversible
focal ischemic injury demonstrated by diffusion-weighted magnetic
resonance imaging in rats. Stroke. 1992;23:1304-1311.
11.
Knight RA, Ordidge RJ, Helpern JA, Chopp M, Rodolosi LC, Peck
D. Temporal evolution of ischemic damage in rat brain measured by
proton nuclear magnetic resonance imaging. Stroke. 1991;22:802-808.
12. Quast MJ, Huang N, Hillman GR, Kent TA. The evolution of acute stroke recorded by multimodal magnetic resonance imaging. Magn Reson Imaging. 1993;11:465-471. [Medline] [Order article via Infotrieve]
13. Moseley ME, Cohen Y, Mintorovitch J, Chileuitt L, Shimizu H, Kucharczyk J, Wendland MF, Weinstein PR. Early detection of regional cerebral ischemia in cats: comparison of diffusion- and T2-weighted MRI and spectroscopy. Magn Reson Med. 1990;14:330-346. [Medline] [Order article via Infotrieve]
14.
Warach S, Chien D, Li W, Ronthal M, Edelman RR. Fast magnetic
resonance diffusion-weighted imaging of acute human stroke.
Neurology. 1992;42:1717-1723.
15.
Busza AL, Allen KL, King MD, Van Bruggen N, Williams SR,
Gadian DG. Diffusion-weighted imaging studies of cerebral ischemia in
gerbils. Stroke. 1992;23:1602-1612.
16. Roberts TPL, Vexler Z, Derugin N, Moseley ME, Kucharczyk J. High-speed MR imaging of ischemic brain injury following stenosis of the middle cerebral artery. J Cereb Blood Flow Metab. 1993;13:940-946. [Medline] [Order article via Infotrieve]
17. Back T, Hoehn-Berlage M, Kohno K, Hossmann K-A. Diffusion nuclear magnetic resonance imaging in experimental stroke. Stroke. 1994;25:494-500. [Abstract]
18.
Heiss W-D. Experimental evidence of ischemic thresholds and
functional recovery. Stroke. 1992;23:1668-1672.
19.
Perez-Trepichio AD, Furlan AJ, Little JR, Jones SC.
Hydroxyethyl starch 200/0.5 reduces infarct volume following embolic
stroke in rats. Stroke. 1992;23:1782-1790.
20.
Jones SC, Bose B, Furlan AJ, Friel HT, Easley KA, Meredith MP,
Little JR. CO2 reactivity and heterogeneity of cerebral
blood flow in ischemic, border zone and normal cortex. Am J
Physiol. 1989;257:H473-H482.
21. Salgado AV, Jones SC, Furlan AJ, Korfali E, Marshall SA, Little JR. Bimodal treatment with nimodipine and low molecular-weight dextran for focal cerebral ischemia in the rat. Ann Neurol. 1989;26:621-627. [Medline] [Order article via Infotrieve]
22.
Chen ST, Hsu CY, Hogan EL, Maricq H, Balentine JD. A model of
focal ischemic stroke in the rat: reproducible extensive cortical
infarction. Stroke. 1986;17:738-743.
23. Xue M, Ng TC, Jones SC, Perez-Trepichio AD, Modic M. 1H NMR spectroscopic imaging with very small voxel size for the detection of metabolic injury of stroke in rat brain. In: Program and abstracts of the 10th annual meeting of the Society of Magnetic Resonance in Medicine; August 12-16, 1991; San Francisco, Calif. 1991:1051.
24. Iadecola C, Reis DJ. Continuous monitoring of cerebrocortical blood flow during stimulation of the cerebellar fastigial nucleus: a study by laser-Doppler flowmetry. J Cereb Blood Flow Metab. 1990;10:608-617. [Medline] [Order article via Infotrieve]
25.
Hansen TD, Warner DS, Todd MM, Vust LJ. Effects of nitrous
oxide and volatile anaesthetics on cerebral blood flow. Br J
Anaesth. 1989;63:290-295.
26.
Sakurada O, Kennedy C, Jehle J, Brown JD, Carbin GL, Sokoloff
L. Measurement of local cerebral blood flow with
iodo[14C]antipyrine. Am J Physiol. 1978;234:H59-H66.
27. Tyson GW, Teasdale GM, Graham DI, McCulloch J. Focal cerebral ischemia in the rat: topography of hemodynamic and histopathological changes. Ann Neurol. 1984;15:559-567. [Medline] [Order article via Infotrieve]
28. Swanson RA, Morton MT, Tsao-Wu G, Savalos RA, Davidson C, Sharp FR. A semiautomated method for measuring brain infarct volume. J Cereb Blood Flow Metab.. 1990;10:290-293. [Medline] [Order article via Infotrieve]
29. Fleiss JL. Measuring nominal scale agreement among many raters. Psychol Bull. 1971;76:378-382.
30. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics. 1977;33:159-174. [Medline] [Order article via Infotrieve]
31. Rindskopf D, Rindskopf W. The value of latent class analysis in medical diagnosis. Stat Med. 1986;5:21-27. [Medline] [Order article via Infotrieve]
32. SAS Institute. SAS/STAT User's Guide, Version 6, Volume 2. Cary, NC: SAS Institute, Inc; 1989.
33. Bryan RN, Levy LM, Whitlow WD, Killian JM, Preziosi TJ, Rosario JA. Diagnosis of acute cerebral infarction: comparison of CT and MR imaging. AJNR. 1991;12:611-620. [Abstract]
34. Yuh WT, Crain MR, Loes DJ, Greene GM, Ryals TJ, Sato Y. MR imaging of cerebral ischemia: findings in the first 24 hours. AJNR. 1991;12:621-629. [Abstract]
35. Knight RA, Dereski MO, Helpern JA, Ordidge RJ, Chopp M. Magnetic resonance imaging assessment of evolving focal cerebral ischemia. Stroke.. 1994;25:1252-1262. [Abstract]
36.
Benveniste H, Hedlund LW, Johnson GA. Mechanism of detection
of acute cerebral ischemia in rats by diffusion-weighted magnetic
resonance microscopy. Stroke. 1992;23:746-754.
37. Helpern JA, Ordidge RJ, Knight RA. The mechanism behind changes in diffusion-weighted MRI following stroke. Stroke. 1993;24:171. Abstract.
38.
Schuier FJ, Hossmann K-A. Experimental brain infarcts in cats,
II: ischemic brain edema. Stroke. 1980;11:593-601.
39.
Gotoh O, Asano T, Koide T, Takakura K. Ischemic brain edema
following occlusion of the middle cerebral artery in the rat, I: the
time courses of the brain water, sodium and potassium contents and
blood-brain barrier permeability to 125I-Albumin.
Stroke. 1985;16:101-109.
40. Dardzinski BJ, Sotak CH, Fisher M, Hasegawa Y, Li L, Minematsu K. Apparent diffusion coefficient mapping of experimental focal cerebral ischemia using diffusion-weighted echo-planar imaging. Magn Reson Med. 1993;30:318-325. [Medline] [Order article via Infotrieve]
41.
Moseley ME, Cohen Y, Kucharczyk J, Mintorovitch J, Asgari HS,
Wendland MF, Tsuruda J, Norman D. Diffusion-weighted MR imaging of
anisotropic water diffusion in cat central nervous system.
Radiology. 1990;176:439-445.
42. Turner R, Le Bihan D, Chesnick AS. Echo-planar imaging of diffusion and perfusion. Magn Reson Med. 1991;19:247-253. [Medline] [Order article via Infotrieve]
43. MacFall JR, Maki JH, Johnson GA, Hedlund LW, Cofer GP. Pre- and postmortem diffusion coefficients in rat neural and muscle tissues. Magn Reson Med. 1991;20:89-99. [Medline] [Order article via Infotrieve]
44. Hoffman EJ, Huang SC, Phelps ME. Quantitation in positron emission computed tomography, I: effect of object size. J Comput Assist Tomogr. 1979;3:299-308. [Medline] [Order article via Infotrieve]
45. Kato H, Kogure K, Ohtomo H, Izumiyama M, Tobita M, Matsui S, Yamamoto E, Kohno H, Ikebe Y, Watanabe T. Characterization of experimental ischemic brain edema utilizing proton nuclear magnetic resonance imaging. J Cereb Blood Flow Metab. 1986;6:212-221. [Medline] [Order article via Infotrieve]
46. Kotwica Z, Thuomas KA, Persson L. Magnetic resonance studies on the development of ischemic edema in an early period after occlusion of middle cerebral artery in a rat. Radiol Diagn. 1989;30:307-311.
47.
Turner R, Le Bihan D, Maier J, Vavrek R, Hedges LK, Pekar J.
Echo-planar imaging of intravoxel incoherent motion.
Radiology. 1990;177:407-414.
This article has been cited by other articles:
 |
|
 |
|
  J. V. Guadagno, P. S. Jones, T. D. Fryer, O. Barret, F. I. Aigbirhio, T. A. Carpenter, C. J. Price, J. H. Gillard, E. A. Warburton, and J.-C. Baron Local Relationships Between Restricted Water Diffusion and Oxygen Consumption in the Ischemic Human Brain Stroke, July 1, 2006; 37(7): 1741 - 1748. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
W.-R. Schabitz, H. Schade, S. Heiland, R. Kollmar, J. Bardutzky, N. Henninger, H. Muller, U. Carl, S. Toyokuni, C. Sommer, et al. Neuroprotection by Hyperbaric Oxygenation After Experimental Focal Cerebral Ischemia Monitored by MRI Stroke, May 1, 2004; 35(5): 1175 - 1179. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 V. N. Thijs, A. Adami, T. Neumann-Haefelin, M. E. Moseley, and G. W. Albers Clinical and Radiological Correlates of Reduced Cerebral Blood Flow Measured Using Magnetic Resonance Imaging Arch Neurol, February 1, 2002; 59(2): 233 - 238. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
Y. Wang, W. Hu, A. D. Perez-Trepichio, T. C. Ng, A. J. Furlan, A. W. Majors, S. C. Jones, and E. C. Haley Jr Brain Tissue Sodium Is a Ticking Clock Telling Time After Arterial Occlusion in Rat Focal Cerebral Ischemia Editorial Comment Stroke, June 1, 2000; 31(6): 1386 - 1392. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Shimizu-Sasamata, T. Kano, J. Rogowska, G. L. Wolf, M. A. Moskowitz, E. H. Lo, and C. Iadecola YM872, a Highly Water-Soluble AMPA Receptor Antagonist, Preserves the Hemodynamic Penumbra and Reduces Brain Injury After Permanent Focal Ischemia in Rats • Editorial Comment Stroke, October 1, 1998; 29(10): 2141 - 2148. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K.J. van Everdingen, J. van der Grond, L.J. Kappelle, L.M.P. Ramos, and W.P.T.M. Mali Diffusion-Weighted Magnetic Resonance Imaging in Acute Stroke Stroke, September 1, 1998; 29(9): 1783 - 1790. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
X. Di, R. Bullock, J. Watson, P. Fatouros, B. Chenard, F. White, and F. Corwin Effect of CP101,606, a Novel NR2B Subunit Antagonist of the N-Methyl-D-Aspartate Receptor, on the Volume of Ischemic Brain Damage and Cytotoxic Brain Edema After Middle Cerebral Artery Occlusion in the Feline Brain Stroke, November 1, 1997; 28(11): 2244 - 2251. [Abstract] [Full Text]
|
|
|
|
|
|
I. Loubinoux, A. Volk, J. Borredon, S. Guirimand, B. Tiffon, J. Seylaz, P. Meric, and G. A. Rosenberg Spreading of Vasogenic Edema and Cytotoxic Edema Assessed by Quantitative Diffusion and T2 Magnetic Resonance Imaging Stroke, February 1, 1997; 28(2): 419 - 427. [Abstract] [Full Text]
|
|
|
|
|
|
E. H. Lo, A. R. Pierce, J. B. Mandeville, B. R. Rosen, C. Y. Hsu, and W. Lin Neuroprotection With NBQX in Rat Focal Cerebral Ischemia: Effects on ADC Probability Distribution Functions and Diffusion-Perfusion Relationships Stroke, February 1, 1997; 28(2): 439 - 447. [Abstract] [Full Text]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||