Diffusion- and perfusion-weighted magnetic resonance imaging of focal cerebral ischemia and cortical spreading depression under conditions of mild hypothermia

Brain Research 885 (2000) 208–219 www.elsevier.com / locate / bres

Research report

Diffusion- and perfusion-weighted magnetic resonance imaging of focal cerebral ischemia and cortical spreading depression under conditions of mild hypothermia Midori A. Yenari a,b,d , *, David Onley a,d , Maj Hedehus c,d , Alexander deCrespigny c,d ,1 , Guo Hua Sun a,d , Michael E. Moseley c,d , Gary K. Steinberg a,d a

Department of Neurosurgery, Stanford University Medical Center, 120 Welch Road, HSLS Bldg. P304, Stanford, CA 94305 -5487, USA b Department of Neurology, Stanford University Medical Center, Stanford, CA, USA c Department of Radiology, Stanford University Medical Center, Stanford, CA, USA d Stanford Stroke Center, Stanford University Medical Center, Stanford, CA, USA Accepted 29 August 2000

Abstract In a model of experimental stroke, we characterize the effects of mild hypothermia, an effective neuroprotectant, on fluid shifts, cerebral perfusion and spreading depression (SD) using diffusion- (DWI) and perfusion-weighted MRI (PWI). Twenty-two rats underwent 2 h of middle cerebral artery (MCA) occlusion and were either kept normothermic or rendered mildly hypothermic shortly after MCA occlusion for 2 h. DWI images were obtained 0.5, 2 and 24 h after MCA occlusion, and maps of the apparent diffusion coefficient (ADC) were generated. SD-like transient ADC decreases were also detected using DWI in animals subjected to topical KCl application (n54) and ischemia (n56). Mild hypothermia significantly inhibited DWI lesion growth early after the onset of ischemia as well as 24 h later, and improved recovery of striatal ADC by 24 h. Mild hypothermia prolonged SD-like ADC transients and further decreased the ADC following KCl application and immediately after MCA occlusion. Cerebral perfusion, however, was not affected by temperature changes. We conclude that mild hypothermia is neuroprotective and suppresses infarct growth early after the onset of ischemia, with better ADC recovery. The ADC decrease during SD was greater during mild hypothermia, and suggests that the source of the ADC is more complex than previously believed.  2000 Elsevier Science B.V. All rights reserved. Theme: Disorders of the nervous system Topic: Ischemia Keywords: Mild hypothermia; Rat; Magnetic resonance imaging; Stroke; Spreading depression

1. Introduction While mild hypothermia has been shown to be neuroprotective against cerebral ischemia, the precise mechanisms are not well known. Alterations in cerebral metabolism and blood flow are known to play some role in this protective effect [33,39], although this cannot fully explain *Corresponding author. Tel.: 11-650-723-4448; fax: 11-650-7234451. E-mail address: [email protected] (M.A. Yenari). 1 Present address: NMR Center, Massachusetts General Hospital, Boston, MA, USA.

the extent of protection seen with only modest reductions in brain temperature [14]. Diffusion-weighted magnetic resonance imaging (DWI) is capable of non-invasively detecting fluids shifts following experimental stroke with quantitative estimates determined by computing the apparent diffusion coefficient (ADC) [38,40]. ADC abnormalities detected by DWI have been correlated with ATP depletion, tissue acidosis, decreases in Na 1 ,K 1 -ATPase activity and K 1 concentration [2,38]. These metabolic disturbances occur when the ADC decreases 35–50% of normal or more. If mild hypothermia’s protective effects are, in part, due to preservation of ion gradients, DWI should be capable of detecting this. Perfusion-weighted

0006-8993 / 00 / $ – see front matter  2000 Elsevier Science B.V. All rights reserved. PII: S0006-8993( 00 )02942-5

M. A. Yenari et al. / Brain Research 885 (2000) 208 – 219

MRI (PWI) can be used to evaluate cerebral perfusion. Although PWI does not precisely reflect regional cerebral blood flow (rCBF), it is a non-invasive technique which can provide information regarding microvascular patency [41,62]. As these techniques [3,58,60] and hypothermic therapies [35,51] are beginning to gain interest at the clinical level, we utilize DWI and PWI to study the temporal and regional evolution of mild hypothermia in an experimental model of transient focal cerebral ischemia. DWI can also be used to detect transient declines in ADC which correspond electrophysiologically with the transient depolarizations or spreading depressions (SD) observed following ischemia [5,6,17,20,31,48–50,54]. SDs which occur during cerebral ischemia are thought to contribute to infarct growth by altering ion gradients or increasing extracellular glutamate [20]. The extent and distribution of SDs also appear to be correlated with expression of various genes, such as immediate early genes [19], cyclooxygenase [37] and protein kinase C (PKC) [29]. While the SDs may result in worsening of ischemic injury, pre-insult induction of SDs by KCl application may actually protect the brain from subsequent ischemic events (ischemic tolerance) [26,27,36]. An earlier report showed that mild hypothermia slows the propagation of the SDs within cortex, as detected by DC potential measurements [59]. We now show complementary results using DWI, which has the advantage of being non-invasive and offers anatomical resolution. To our knowledge, this is the first report characterizing the temperature dependence of SDlike ADC changes in ischemia and KCl models using DWI.

209

within cerebral cortex and striata. Brain temperature was reduced to 358C, 338C, 308C and 288C by placing the animal on a cooling blanket and spraying ethanol onto the body, then applying cool air from a hair dryer hose. Temperature was maintained for 15–20 min prior to cooling to the next level. To determine the effects of temperature changes during ischemia, temperature probes were stereotaxically placed into each striata and a third probe was placed in the rectum. Temperatures were monitored during conditions of ischemic normothermia (n51) and mild hypothermia (n51) as described subsequently.

2.2. Effects of temperature on ADC As micromolecular diffusion is altered with temperature changes, the relationship between brain ADC and temperature was established in our model. Using the correlation data from the above experiment, we monitored rectal temperature during scanning in this experiment and all subsequent experiments. Three anesthetized non-ischemic animals underwent DWI imaging with brain temperatures varied from 308C to 408C. Rectal temperatures were varied as described above and DWI images were performed at each temperature. Trace DWI images were generated as described subsequently in the MRI protocol section. Regions of interest (ROIs) were identified within regions of cortex, striatum and thalamus, and ADCs from fitted maps were measured within these ROIs. Correction curves were determined for each of the structures as a function of temperature.

2.3. Ischemia model 2. Materials and methods

2.1. Measurement of brain temperature We first established the relationship between rectal temperature and brain temperature in our model, as it is not yet feasible to measure simultaneous brain and body temperature during MRI scanning given the susceptibility artifacts caused by the brain temperature probe. Rats were placed in the MRI probe then into the scanning unit to maintain the same thermodynamic environment experienced during scanning. Because we found no differences in temperature gradients with animals in or out of the MRI unit, subsequent temperature correlations were made with the animal in the MRI probe. We applied whole body cooling and warming in the range of 30–408C (rectal temperature) to halothane-anesthetized rats (n53) and monitored for heart rate, respirations and rectal temperature. Temperature was controlled by placing a bonnet from a commercially available hair dryer directly onto the animal, and the air temperature could be adjusted by the unit’s controls. A small burr hole was drilled over the fronto-parietal cortex and temperature probes were placed

Twenty-two male Sprague–Dawley rats weighing between 290 and 320 g. were anesthetized with 3% halothane by face mask. Oxygen and air were supplied in a ratio of 0.2 l oxygen:0.8 l air. Once surgical planes of anesthesia were attained (assessed by absence of hind leg withdrawal to pinch), halothane was decreased to 1–1.5% throughout the remainder of the surgery. A femoral artery catheter was placed to monitor mean arterial blood pressure (MAP) and a rectal probe was placed to monitor body temperature. The animal was placed on a heating / cooling blanket to maintain body temperature between 378C and 388C. Arterial blood gases, serum glucose and hematocrit are measured and recorded. Ischemia was induced using an occluding intraluminal suture used previously by our group [61]. A cervical midline incision was made and the left carotid artery and branches were isolated. The common carotid artery (CCA), external carotid (ECA), and pterygopalatine (PPA) were identified. The ECA was ligated and bisected. An uncoated 30-cm long segment of 3-0 nylon monofilament suture (Ethicon, Somerville, NJ) with the tip rounded by a flame is inserted into the stump of the ECA and advanced under direct visualization into the ICA approxi-

210

M. A. Yenari et al. / Brain Research 885 (2000) 208 – 219

mately 19–20 mm from the bifurcation in order to occlude the ostium of the MCA. A temporary aneurysm clip was also placed on the CCA. Animals were either kept normothermic (normo, rectal5378C, n511) or rendered mildly hypothermic (hypo, rectal5308C, n511) within 15–20 min after MCA occlusion, and maintained for 2 h. After 2 h of ischemia, the suture was withdrawn with the animal remaining in the MRI scanner. The animal was rewarmed within 15–20 min using a warming blanket and by blowing warm air into the scanner bore. Following the final scan, the animal was returned to the operating table. The aneurysm clip was removed and the surgical incisions closed. The animal was allowed to recover, then transported to the intensive care unit at the animal facility for post-operative monitoring. At the completion of the experiment, the animal was euthanized with a barbiturate overdose.

2.4. KCl-induced SD experiments In separate, non-ischemic, halothane anesthetized animals (n54), a KCl chamber was placed 2 mm anterior to the lambdoidal suture and 2 mm lateral to the sagittal suture over the right hemisphere. A DC electrode was placed 2 mm anterior to the coronal suture and 2 mm lateral to the sagittal suture. KCl (3 M) was applied during imaging acquisition and a DC shift was confirmed by simultaneous recordings from the electrode. Rectal temperature was varied and images were obtained at 308C, 338C and 378C, corresponding to brain temperatures of 338C, 358C and 388C, respectively. DWI imaging (described below) commenced immediately before and up to 20 min after KCl application.

2.5. Ischemia-induced SD experiments In six separate animals, MCA occlusion was induced remotely, with the animal inside of the MRI scanner. This method has been published previously [49], but some modifications have been made. PE-50 tubing 10 mm in length was tapered on one end. The opposite end of the tubing was inserted approximately 5 mm into PE-90 tubing which extended the length of the MR cradle (|30 cm). The tapered tip of the PE-50 tubing was inserted 5 mm into the ECA stump and secured. A 30-mm monofilament suture was attached to another segment of PE-50 tubing (45 cm in length) and fed through the P-90 tubing and into the ECA approximately 15 mm from the CCA bifurcation. This design permitted the suture to be advanced no more than 20 mm from the CCA bifurcation. Two minutes after the beginning of image acquisition, the suture was advanced another 2–5 mm from the CCA bifurcation. Two slice DWI imaging to detect SDs continued for a total of 30 min, followed by multislice protocols for DWI and PWI to confirm ischemia. Animals were maintained normother-

mic (n53), or were cooled to 308C (rectal temperature, n53) immediately prior to suture insertion.

2.6. MRI scanning DWI and PWI images were obtained intra-ischemically (30 min following MCAO), post-reperfusion (within 5–10 min of reperfusion) and 24 h later. Imaging was performed on a Bruker CSI 2.0 Tesla system. The complete protocol (for each DWI and PWI scan set) took approximately 30 min to complete. Because of the prolonged time under halothane anesthesia, many animals were incapable of surviving the full 24 h. Only four normothermic and three mildly hypothermic animals were able to complete imaging at all three time points. Therefore, subsequent animals were randomized into two groups. In the first group (three normothermics, four mildly hypothermics), animals were imaged intra-ischemically and early into reperfusion. In the second group (four normothermics and four mildly hypothermics), the animal was removed from the scanner after obtaining the intra-ischemic images and the mild hypothermia was completed. The suture was removed and the animal was allowed to recover. The following day, a second set of scans at 24 h was performed. For all groups, the total duration under anesthesia was similar (4.5 h). A total of 22 animals, 11 normothermic and 11 mildly hypothermic, were imaged for this part of the study.

2.7. DWI protocols Spin echo echo planar images (SE EPI) was used with the following imaging parameters: FOV540340 mm 2 and resolution564364. For the ischemia and correlative experiments, a 5-cm diameter bird-cage radio frequency coil was used. For the SD experiments, a surface coil was used. Both isotropic diffusion-weighted (IDW) and DWI trace images were obtained. Imaging parameters were: TE / TR5 66 ms / 4 s, no. of slices54, coronal plane, slice thickness5 2.0 mm. For IDW scans, 16 images with varying gradient amplitudes (16 b-values between 100 and 1200 s / mm 2 in random order) were acquired for each slice. For the trace DWI images, eight images were acquired from each slice (eight b-values between 0 and 1400 s / mm 2 ), and repeated with the gradients applied in the x, y and z directions. Trace images were derived by averaging the images obtained from each direction. T2-Weighted (T2W) images were obtained using b-values50. In order to detect SDs, continuous, two slice EPI images were acquired. Imaging parameters were TE / TR550 ms / 2 s, slice thickness51.65 mm, no. of slices52, axial plane. Three images with varying gradient amplitude (gradients applied in x, y and z simultaneously, total b-values56, 1347, and 1347) were obtained. This was repeated 600 times with a time resolution of 2 s.

M. A. Yenari et al. / Brain Research 885 (2000) 208 – 219

2.8. PWI protocols PWI scans were performed using a T2-weighted gradient echo EPI to track the contrast bolus [48,49,62]. Three coronal slices were repeatedly acquired during rapid intravenous injection of a contrast agent (0.2 mmol / kg gadopentitate dimeglumine). Imaging parameters were the same as for DWI imaging. All three slices were acquired every 2 s prior to and during contrast injection, and continued for a total of 32 s.

2.9. MRI analysis All analyses were performed by an investigator blinded to treatment groups, although information regarding the time point of the scan sets was provided. Data sets for all MRI scans from a specific time point were pooled, regardless of when the animal was euthanized. ROIs delineating the DWI hyperintensity from four adjacent slices were traced by hand and expressed as the percent area relative to the ipsilateral hemisphere. We have previously shown that these DWI hyperintensities at 24 h strongly correlate with infarction as seen by triphenyl tetrazolium chloride (TTC) staining [61]. ROIs were also delineated within the striata (Fig. 1, Si) to determine the percent change in ADC value between the ipsilateral ischemic (Si) and corresponding contralateral non-ischemic (Sc) sides from the ADC maps. For comparison between temperature groups, all ADCs from images acquired under mild hypothermic conditions were corrected to a brain temperature in a normothermic animal using the previous correlative data.

211

Maps of bolus arrival delays were fitted from the PWI scans using a fitting routine previously reported by our group (MRVision, Menlo Park, CA) [62]. We and others have found the delay maps to be the most sensitive perfusion index compared to relative cerebral blood volume (rCBV) and bolus peak effect [41,62]. Regions of significant delay were mapped out using a thresholding routine. Only areas where bolus arrival delays were greater than 2 s compared to the contralateral hemisphere were included and expressed as a percentage of the total ipsilateral hemisphere area [62]. Because the scans could only be continued for a finite period of time (20 min for each SD trial), the total number of SDs could not be reliably determined; therefore, we chose to only analyze parameters from the initial SD wave from each KCl application or ischemia experiment. From the KCl-induced SD experiments, ADC maps were generated and ROIs were defined in the ipsilateral hemisphere near the region of KCl application. ADC changes within the same ROI were computed from the sequential images and plotted as a function of time. The maximum ADC decrease, the time from initial ADC decline to maximal ADC decrease (ADC decline time), and the time from maximal ADC decrease to ADC recovery (ADC recovery time) were measured from the first transient depolarization. For the SD experiments conducted in ischemic animals, ROIs were identified within the cortex corresponding to a peri-infarct region (P, see Fig. 1). Another ROI within the striatum was also defined (Si, Fig. 1) in order to measure the time to terminal ADC decrease, defined as the time to maximum ADC decline in regions where no ADC recovery was observed during the 20-min scanning period.

Fig. 1. Diffusion-weighted MRI scan showing a large middle cerebral artery territory lesion. Regions of interest (ROI) which were used to measure changes in apparent diffusion coefficients (ADCs) following ischemia are indicated P, peri-infarct region in cortex; Si, ipsilateral ischemic striatum; Sc, contralateral non-ischemic striatum.

212

M. A. Yenari et al. / Brain Research 885 (2000) 208 – 219

Similar parameters from an ROI within the cortex (adjacent to the area of KCl application) were measured from the KCl experiments.

2.10. Statistical analysis Standard statistical methods were used to analyze data. Differences between groups and the various parameters was determined using one-way analysis of variance (ANOVA) and ANOVA with repeated measures to detect overall significance, followed by a post hoc multiple comparison procedure (Scheffe’s). Correlation was determined by Pearson’s correlation coefficient, followed by a linear regression. Statistical significance was determined at the P,0.05 level. All data are presented as mean6S.E.M.

3. Results

3.1. Correlative experiments We found strong correlations between rectal and brain temperature (r50.92, P,0.0001), with brain temperature somewhat higher than rectal. The relationship was not entirely linear, as we found that when rectal temperature was 378C, brain temperature was 388C, whereas when rectal temperature was 308C, brain temperature was 338C (Table 1). Furthermore, cortical and striatal brain temperatures were similar. Under ischemic conditions, striatal temperature within the non-ischemic side followed patterns similar to those observed in non-ischemic animals. However, the ischemic striatal temperature was lower than the non-ischemic striatum, and was no different from the rectal temperature (Fig. 2). Therefore, we found it possible to reliably predict brain temperature from rectal temperature measurements in this model. ADC was found to vary linearly and directly with temperature. A 18C change in brain temperature corresponded to a 1.6% change in ADC. Our results are in close agreement with others’ who measured the temperature dependence of the ADC in water and brain [18,23] as well as in a mathematical model [32]. Although these changes are quite small in comparison to the ADC reduction observed with ischemia, we used this relationship to correct all measured ADCs to the ADC corresponding to a

Fig. 2. Striatal temperature in ischemic and non-ischemic brain under normothermic and mildly hypothermic conditions. Under normothermic conditions, rectal (d) and ischemic striatal (.) temperatures were similar, whereas non-ischemic striatal temperature (j) was approximately 18C higher. Under mildly hypothermic conditions, rectal (s) and ischemic striatal (,) temperatures were similar, whereas non-ischemic striatal temperature (h) was approximately 28C higher. Cooling began immediately after occlusion of the middle cerebral artery (MCAO), and rewarming began immediately after reperfusion (arrow).

brain temperature under normothermic conditions (i.e. non-ischemic brain was corrected to 388C, and ischemic brain was corrected to 378C).

3.2. Focal cerebral ischemia There were no differences among any of the physiological parameters (except temperature) between groups. For each temperature group, 22 DWI image sets (11 mildly hypothermic and 11 normothermic sets) were available for analysis at 0.5 h post-ischemia (intraischemic), 14 image sets (seven mildly hypothermic and seven normothermic sets) were available for analysis 2.5 h post-ischemia (postreperfusion) and 15 image sets (seven mildly hypothermic and eight normothermic sets) were available for analysis at 24 h. For the PWI scans, only six image sets were available for analysis at the 24-h time point among normothermic animals, otherwise the numbers are the same as for DWI. Mild hypothermia inhibited DWI lesion growth early

Table 1 Correlation of brain and rectal temperature (n53)a Target rectal temperature (8C)

Rectal temperature (8C)

Brain temperature (8C)

39 37 35 33 30

39.260.33 37.060.2 35.260.1 32.860.05 30.460.16

40.260.34 38.060.18 37.160.08 34.960.12 32.960.08

a

r50.92, P,0.0001.

M. A. Yenari et al. / Brain Research 885 (2000) 208 – 219

213

Fig. 3. MRI of transient focal cerebral ischemia under conditions of mild hypothermia (coronal plane) (a) and normothermia (b). DWI and ADC images show that the ischemic lesion (left side of image) is reduced in size at all time points compared to normothermia. PWI delay maps show a large delay area intra-ischemically (left side of image) which reverses upon reperfusion and remains normal by 24 h. T2W image does not become positive until 24 h.

214

M. A. Yenari et al. / Brain Research 885 (2000) 208 – 219

after the onset of ischemia and up to 24 h later. DWI lesion areas (% of the ipsilateral hemisphere) were intraischemic: hypo 1064%, normo 1963.2%, P50.07; post-reperfusion:

hypo 561.6%, normo 2163%, P,0.001; and 24 h: hypo 862%, normo 2864%, P,0.01 (Figs. 3 and 4a). Striatal ADC recovery by 24 h was significantly improved among mildly hypothermic animals. ADC recovery, compared to the contralateral non-ischemic striata was 10466% among mild hypothermics versus 7766% ADC recovery among normothermics (P,0.05) (Fig. 4b). Areas of significant bolus peak delays were no different between groups at any time point. Cerebral perfusion was equally delayed during MCA occlusion, but returned to near normal in both groups shortly after reperfusion (Figs. 3 and 4c).

3.3. KCl- and ischemia-induced SDs

Fig. 4. Temperature-dependent, MRI-detected ischemic lesion changes. (a) Mild hypothermia attenuates DWI lesion area. Significant reduction in DWI lesion area (% of the ipsilateral hemisphere) among the hypothermic group is observed upon reperfusion (2 h) and at 24 h. Trends showing reduced lesion size compared to the normothermic group are noted intra-ischemically (30 min post-occlusion). (b) Mild hypothermia improves recovery of the apparent diffusion coefficient (ADC) in focal cerebral ischemia. Striatal ADC (% decrease relative to the contralateral, non-ischemic side) recovers to normal values in the hypothermia group by 24 h. All ADCs are corrected to normothermic conditions. (c) Mild hypothermia does not alter cerebral perfusion. Areas of bolus arrival delay from the PWI scans are no different between the hypothermic group compared to the normothermic group. Upon reperfusion, these areas of perfusion delay reverse and no delay is observed in either group by 24 h. *P#0.05, **P#0.01 ***P#0.001 vs. normothermic group. Numbers in parentheses indicate the number of images included in the analysis for each condition and time point.

KCl application produced SD-like waves on DWI in non-ischemic animals which corresponded to the DC potential changes (data not shown). This correlation has been previously been described in detail by our group [12,48,49] and others [17,31]. Therefore, DC potential shifts were not measured for subsequent studies. Maximum ADC decreases were significantly lower with decreased temperature. Mild hypothermia prolonged ADC decline and ADC recovery times from the initial depolarization compared to normothermia (Table 2 and Fig. 5). Among normothermic animals, ischemia resulted in SDlike ADC changes within peri-infarct zones. Mild hypothermia suppressed SD generation within cortex among two of three animals, and these animals also did not have any MRI-detected cortical injury. The third mildly hypothermic animal had a cortical lesion seen on DWI, and SD-like waves were observed in peri-infarct regions. ADC decline and recovery times within the cortical peri-infarct zone were slowed in this mildly hypothermic animal compared to the normothermics. SDs were detected within peri-infarct regions in the mildly hypothermic group with prolonged ADC decline and recovery (Table 3). ADC decline and recovery times were approximately three times longer among mildly hypothermic animals compared normothermic ones (P,0.05). Time to terminal ADC decrease was twice as long in mildly hypothermic animals compared to controls. Maximum ADC decreases were lower among the mildly hypothermic group even after correction for temperature.

4. Discussion Mild hypothermia has been shown to be an effective treatment against cerebral ischemic injury by numerous groups [14,34]. Lowered temperatures are thought to preserve metabolic stores, but this has not been consistently demonstrated in the setting of mild to moderate hypothermia [8,15,25,28,53,57]. Several groups have also shown that mild hypothermia is associated with decreased accumulation of excitatory amino acids [4,7,21,33,42], and subsequent reduction in calcium-mediated toxicity and

M. A. Yenari et al. / Brain Research 885 (2000) 208 – 219

215

Table 2 Temperature dependence of SD-like ADC changes following KCl application (n54)a Brain temperature (8C)

ADC cor Max (310 24 s / cm)

ADC decline (s)

ADC recovery (s)

38 35 33 P value b

5.860.5 5.060.2 4.660.25 ,0.0001* ,‡

5167.5 5766.5 68610 ,0.01*

5465.5 7566.5 8464.5 ,0.0001* ,†,‡

a b

ADC, apparent diffusion coefficient; KCl, potassium chloride; ADC cor , ADC corrected to 388C; ADC cor Max, maximum ADC cor decrease. *338C vs. 388C; † 338C vs. 358C; ‡ 358C vs. 388C.

nitric oxide generation [24]. SDs have been thought to contribute to infarct growth due to increased calcium influx through glutamate activated calcium ion channels [5,13,20,43] and propagation through gap junctions [44]. DWI has gained considerable interest over the past few years as a rapid, non-invasive way of detecting ischemic lesions by taking advantage of the early fluid shifts that occur following stroke. DWI is also capable of detecting transient diffusion changes which have been correlated to SDs [17,31,49]. Given that these ADC changes reflect alterations in ion and fluid homeostasis following ischemic insults [40], we show here that protection by mild hypothermia appears to be related to enhanced recovery from pathological fluid shifts and attenuated the DWI lesion early after ischemia onset. Mild hypothermia also suppressed SD-like ADC changes. However, the ADC drop during the SD-like transients was lower among animals subjected to ischemia and KCl application at lower temperatures, and suggests that the source of the ADC is more complex than previously believed. Using PWI, we show that the protective effects of mild hypothermia do not appear to alter the area of cerebral perfusion deficit. Jiang et al. [23] previously utilized a similar rodent stroke model to examine the effects of mild hypothermia within striatum using DWI and PWI. Consistent with our results, they found that striatal ADC recovered more rapidly in animals subjected to mild hypothermia, with return of the ADC as early as 6 h post-ischemia in the mildly hypothermic group. We show that MRI-detected lesion growth is inhibited soon after application of mild hypothermia with a trend towards reduced DWI lesion size as early as 0.5 h after ischemia onset, and significant reduction in lesion size by 2 h. We also observed improved recovery of the ADC by 24 h with mild hypothermia. Since the decreased ADC is thought to reflect cytotoxic edema, mild hypothermia could protect ischemic brain by restoring these fluid shifts. On the other hand, it is possible that the early ADC recovery represents a ‘pseudonormalization’, reflecting cell lysis and the development of vasogenic edema. However, overall DWI lesion size at 24 h was attenuated among the mild hypothermia group, making it more likely that the ADC recovery represented mild hypothermic rescue of ischemic tissue. Using electrophysiological methods, others have shown that mild hypothermia delays the time to terminal depolarization following anoxia [25,42]. By applying rapid DWI during MCA

occlusion, we similarly show that mild hypothermia delays the time to maximum ADC decrease (the DWI ‘equivalent’ of terminal depolarization [48,49]). Our observations are also in line with those by Katsura et al. [25] who showed that mild hypothermia slowed the rate of extracellular potassium ion accumulation, and delayed the fall in ATP in a model of cardiac arrest. This is consistent with the notion that hypothermia preserves ATP hydrolases needed to maintain ion gradients, and subsequently delays or prevents cytotoxic edema formation. There have been a few reports in the literature examining SDs and the effects of hypothermia. To our knowledge, this is the first report of temperature dependent SD/ SDlike changes using DWI. However, other reports in the literature have used electrophysiological methods to study the temperature dependence of SDs. After topical application of KCl, Takaoka et al. [55] found that decreases in temperature did not affect the brain’s ability to depolarize, but mild hypothermia did reduce the propagation speed of SDs. Following needle stab injury [59] and focal cerebral ischemia [9], the number of SDs were decreased in number and their onset was delayed with lower temperature. Glutamate antagonists [22,45,56], calcium channel blockers [52] and some anesthetics [50] have also been shown to inhibit SDs. This led some to link SDs to glutamate accumulation and subsequent infarct growth [5,20]. In fact, administration of various glutamate antagonists have been shown to decrease the number of SDs as well as infarct size [37,56], and mild hypothermia attenuates ischemiainduced glutamate release [4,7,21,33,42]. Our data using DWI are in agreement with this. We found that SD-like transient ADC changes are prolonged with mild hypothermia compared to normothermia, and following ischemia, mild hypothermia completely inhibited these transient ADC changes in two of three cases. KCl-induced SDs have been shown to induce certain stress-associated genes [19,46,52], and application of KCl during the ischemic period worsens ischemic injury [54]. Similarly, mild hypothermia is associated with reduced expression of heat shock protein 70 (HSP70), a stress-induced gene [10]. Our finding that mild hypothermia inhibits or slows the propagation of SD-like ADC transients is consistent with the notion that mild hypothermia reduces metabolic and excitotoxic stresses to the brain. The maximum ADC decrease within the SD-like ADC transients were lower among mild hypothermic brains even

216

M. A. Yenari et al. / Brain Research 885 (2000) 208 – 219

Fig. 5. SD-like ADC changes are temperature dependent. (a) ADC maps in the axial plane showing transient ADC decreases in a non-ischemic animal following KCl application. Arrows show spread of reduced ADC within the right (left side of image) hemisphere. Regions where ADC are decreased by 35% or more of the baseline are indicated in red. Time between images for the normothermic experiment is 3 s, while the time between images for the hypothermic experiment is 5 s (top of image is posterior brain; bottom of image is anterior brain). (b) A graph of representative ADC changes following topical KCl application with the cortex. The ADC decline and recovery times are prolonged with lower temperature, as well as the maximum ADC decrease (*ADC, corrected to a temperature of 388C).

M. A. Yenari et al. / Brain Research 885 (2000) 208 – 219

217

Table 3 Temperature dependence of SD-like ADC changes following transient focal cerebral ischemia a Temperature group (8C)

n

ADC cor Max (310 24 s / cm)

ADC decline (s)

ADC recovery (s)

Time to terminal ADC decrease (s)

Normo Hypo

3 3

5.660.3 4.460.1*

126624 360690*

162643 5286155*

162636 366690*

a

ADC, apparent diffusion coefficient; KCl, potassium chloride; ADC cor , ADC corrected to 388C; ADC cor Max, maximum ADC cor decrease; n, number of animals studied; normo, normothermia; hypo, mild hypothermia. *P,0.01 (338C vs. 388C).

after temperature correction in both the KCl and ischemia paradigms. The reasons for our observations are not clear, but lower temperatures are known to alter membrane fluidity [1] and ion channel function [25,47]. Our results could be explained in this light, particularly as temperature affects the function of proteins involved in regulating cellular fluid and ion homeostasis. Furthermore, our data suggest that these changes in ADC-detected fluid shifts are not necessarily damaging, since we observed robust protection with mild hypothermia. Other reports of temperature-dependent SD changes describe slowing and reduced numbers of propagated waves [9]; however, amplitude was not altered [55] and ADC during SD has not previously been measured. In our ischemia model, early terminal ADC decreases within the ipsilateral mildly hypothermic striata were also lower compared to normothermic striata. Yet, there was better recovery of the striatal ADC by 24 h, and reduction in DWI lesion size. Several groups have suggested that critically low ADCs measured early after ischemia onset might predict irreversible ischemic injury [11,16]. However, our data indicate that the extent of early ADC decreases does not necessarily predict the fate of the tissue. Clearly, more investigation into the source and significance of the ADC is needed. PWI is also gaining interest as a rapid means of imaging cerebral perfusion with the anatomic resolution of MRI [30,58,62]. In our hands and others [41,62], the bolus arrival delay is the most sensitive PWI parameter compared to cerebral blood volume, mean transit time and derived CBF, though not necessarily the most specific for actual blood flow. Jiang et al. [23] also examined striatal rCBF using an arterial spin labeling technique and found that rCBF recovered in the mildly hypothermic group, but not in the normothermic group. Using other methods to measure blood flow, others have found that rCBF increases with reperfusion, but is blunted by mild hypothermia [33,39]. Our results are somewhat in contrast to both of these findings in that we did not observe any changes in the areas of bolus arrival delay with mild hypothermia. Therefore, our data suggest that protection from mild hypothermia is independent of changes in cerebral perfusion. In summary, we show using DWI and PWI that intraischemic mild hypothermia is neuroprotective against focal cerebral ischemia, and that this protection is evident soon after ischemia onset, but is not associated with changes in cerebral perfusion. Mild hypothermia also slows the propa-

gation and further lowers the ADC of SD-like transients, but is associated with better ADC recovery by 24 h. As mild hypothermia is beginning to be studied at the clinical level for treatment of cerebral injury, DWI and PWI may be useful in monitoring the response to such therapy as well as non-invasively studying ischemic pathophysiology in human subjects.

Acknowledgements This work was supported in part by funds from NIH NINDS Grant RO1 NS 27292 (G.K.S.), NIH NINDS Grant K08 NS01860 (M.A.Y.), Bernard and Ronni Lacroute (G.K.S.) and the William Randolph Hearst Foundation (G.K.S.). The authors would like to thank Dr Christian Beaulieu and Mr David Kunis for expert technical assistance, and Ms Beth Houle for preparation of the figures.

References [1] M.T. Almeida, J. Ramalho-Santos, C.R. Oliveira, M.C. de Lima, Parameters affecting fusion between liposomes and synaptosomes. Role of proteins, lipid peroxidation, pH and temperature, J. Membr. Biol. 142 (1994) 217–222. [2] T. Back, M. Hoehn-Berlage, K. Kohno, K.A. Hossmann, Diffusion nuclear magnetic resonance imaging in experimental stroke. Correlation with cerebral metabolites, Stroke 25 (1994) 494–500. [3] A.E. Baird, A. Benfield, G. Schlaug, B. Siewert, K.O. Lovblad, R.R. Edelman, S. Warach, Enlargement of human cerebral ischemic lesion volumes measured by diffusion-weighted magnetic resonance imaging, Ann. Neurol. 41 (1997) 581–589. [4] C.J. Baker, A.J. Fiore, V.I. Frazzini, T.F. Choudhri, G.P. Zubay, R.A. Solomon, Intraischemic hypothermia decreases the release of glutamate in the cores of permanent focal cerebral infarcts, Neurosurgery 36 (1995) 994–1001. [5] E. Busch, M.L. Gyngell, M. Eis, M. Hoehn-Berlage, K.A. Hossmann, Potassium-induced cortical spreading depressions during focal cerebral ischemia in rats: contribution to lesion growth assessed by diffusion-weighted NMR and biochemical imaging, J. Cereb. Blood Flow Metab. 16 (1996) 1090–1099. [6] E. Busch, M. Hoehn-Berlage, M. Eis, M.L. Gyngell, K.A. Hossmann, Simultaneous recording of EEG, DC potential and diffusionweighted NMR imaging during potassium induced cortical spreading depression in rats, NMR Biomed. 8 (1995) 59–64. ´ M.D. [7] R. Busto, M.Y. Globus, W.D. Dietrich, E. Martinez, I. Valdes, Ginsberg, Effect of mild hypothermia on ischemia-induced release of neurotransmitters and free fatty acids in rat brain, Stroke 20 (1989) 904–910. [8] H. Chen, M. Chopp, A.M. Vande Linde, M.O. Dereski, J.H. Garcia,

218

[9]

[10]

[11]

[12]

[13]

[14]

[15]

[16]

[17]

[18]

[19]

[20] [21]

[22]

[23]

[24]

[25]

M. A. Yenari et al. / Brain Research 885 (2000) 208 – 219 K.M. Welch, The effects of post-ischemic hypothermia on the neuronal injury and brain metabolism after forebrain ischemia in the rat, J. Neurol. Sci. 107 (1992) 191–198. Q. Chen, M. Chopp, G. Bodzin, H. Chen, Temperature modulation of cerebral depolarization during focal cerebral ischemia in rats: correlation with ischemic injury, J. Cereb. Blood Flow Metab. 13 (1993) 389–394. M. Chopp, Y. Li, M.O. Dereski, S.R. Levine, Y. Yoshida, J.H. Garcia, Hypothermia reduces 72-kDa heat-shock protein induction in rat brain after transient forebrain ischemia, Stroke 23 (1992) 104–107. B.J. Dardzinski, C.H. Sotak, M. Fisher, Y. Hasegawa, L. Li, K. Minematsu, Apparent diffusion coefficient mapping of experimental focal cerebral ischemia using diffusion-weighted echo-planar imaging, Magn. Reson. Med. 30 (1993) 318–325. T. Els, J. Rother, C. Beaulieu, A. de Crespigny, M. Moseley, Hyperglycemia delays terminal depolarization and enhances repolarization after peri-infarct spreading depression as measured by serial diffusion MR mapping, J. Cereb. Blood Flow Metab. 17 (1997) 591–595. G. Gido, T. Kristian, B.K. Siesjo, Induced spreading depressions in energy-compromised neocortical tissue: calcium transients and histopathological correlates, Neurobiol. Dis. 1 (1994) 31–41. M.D. Ginsberg, L.L. Sternau, M.Y.-T. Globus, W.D. Dietrich, R. Busto, Therapeutic modulation of brain temperature: relevance to ischemic brain injury, Cerebrovasc. Brain Metab. Rev. 4 (1992) 189–225. O. Haraldseth, T. Gronas, T. Southon, L. Thommessen, G. Borchgrevink, P. Jynge, S.E. Gisvold, G. Unsgard, The effects of brain temperature on temporary global ischaemia in rat brain. A 31phosphorus NMR spectroscopy study, Acta Anaesthesiol. Scand. 36 (1992) 393–399. Y. Hasegawa, M. Fisher, L.L. Latour, B.J. Dardzinski, C.H. Sotak, MRI diffusion mapping of reversible and irreversible ischemic injury in focal brain ischemia, Neurology 44 (1994) 1484–1490. Y. Hasegawa, L.L. Latour, J.E. Formato, C.H. Sotak, M. Fisher, Spreading waves of a reduced diffusion coefficient of water in normal and ischemic rat brain, J. Cereb. Blood Flow Metab. 15 (1995) 179–187. Y. Hasegawa, L.L. Latour, C.H. Sotak, B.J. Dardzinski, M. Fisher, Temperature dependent change of apparent diffusion coefficient of water in normal and ischemic brain of rats, J. Cereb. Blood Flow Metab. 14 (1994) 383–390. J. Honkaniemi, B.A. States, P.R. Weinstein, J. Espinoza, F.R. Sharp, Expression of zinc finger immediate early genes in rat brain after permanent middle cerebral artery occlusion, J. Cereb. Blood Flow Metab. 17 (1997) 636–646. K.A. Hossmann, Periinfarct depolarizations, Cerebrovasc. Brain Metab. Rev. 8 (1996) 195–208. F.P. Huang, L.F. Zhou, G.Y. Yang, Effects of mild hypothermia on the release of regional glutamate and glycine during extended transient focal cerebral ischemia in rats, Neurochem. Res. 23 (1998) 991–996. T. Iijima, G. Mies, K.A. Hossmann, Repeated negative DC deflections in rat cortex following middle cerebral artery occlusion are abolished by MK-801: effect on volume of ischemic injury, J. Cereb. Blood Flow Metab. 12 (1992) 727–733. Q. Jiang, M. Chopp, Z.G. Zhang, J.A. Helpern, R.J. Ordidge, J. Ewing, P. Jiang, B.A. Marchese, The effect of hypothermia on transient focal ischemia in rat brain evaluated by diffusion- and perfusion-weighted NMR imaging, J. Cereb. Blood Flow Metab. 14 (1994) 732–741. A. Kader, V.I. Frazzini, C.J. Baker, R.A. Solomon, R.R. Trifiletti, Effect of mild hypothermia on nitric oxide synthesis during focal cerebral ischemia, Neurosurgery 35 (1994) 272–277. K. Katsura, H. Minamisawa, A. Ekholm, J. Folbergrova, B.K. Siesjo, Changes of labile metabolites during anoxia in moderately

[26]

[27]

[28]

[29] [30]

[31]

[32]

[33]

[34]

[35]

[36]

[37]

[38]

[39]

[40]

[41]

[42]

[43]

hypo- and hyperthermic rats: correlation to membrane fluxes of K 1 , Brain Res 590 (1992) 6–12. N. Kawahara, S.D. Croll, S.J. Wiegand, I. Klatzo, Cortical spreading depression induces long-term alterations of BDNF levels in cortex and hippocampus distinct from lesion effects: implications for ischemic tolerance, Neurosci. Res. 29 (1997) 37–47. S. Kobayashi, V.A. Harris, F.A. Welsh, Spreading depression induces tolerance of cortical neurons to ischemia in rat brain, J. Cereb. Blood Flow Metab. 15 (1995) 721–727. P. Kozlowski, A.M. Buchan, U.I. Tuor, D. Xue, Z.G. Huang, K.E. Chaundy, J.K. Saunders, Effect of temperature in focal ischemia of rat brain studied by 31 P and 1 H spectroscopic imaging, Magn. Reson. Med. 37 (1997) 346–354. J. Krivanek, V.I. Koroleva, Protein kinase C in the rat cerebral cortex during spreading depression, Neurosci. Lett. 210 (1996) 79–82. J. Kucharczyk, T. Roberts, M.E. Moseley, A. Watson, Contrastenhanced perfusion-sensitive MR imaging in the diagnosis of cerebrovascular disorders, J. Magn. Reson. Imaging 3 (1993) 241– 245. L.L. Latour, Y. Hasegawa, J.E. Formato, M. Fisher, C.H. Sotak, Spreading waves of decreased diffusion coefficient after cortical stimulation in the rat brain, Magn. Reson. Med. 32 (1994) 189–198. D. Le Bihan, J. Delannoy, R.L. Levin, Temperature mapping with MR imaging of molecular diffusion: application to hyperthermia, Radiology 171 (1989) 853–857. E.H. Lo, G.K. Steinberg, Effects of hypothermia on evoked potentials, magnetic resonance imaging, and blood flow in focal ischemia in rabbits, Stroke 23 (1992) 889–893. J. Maher, V. Hachinski, Hypothermia as a potential treatment for cerebral ischemia, Cerebrovasc. Brain Metab. Rev. 5 (1993) 277– 300. D.W. Marion, L.E. Penrod, S.F. Kelsey, W.D. Obrist, P.M. Kochanek, A.M. Palmer, S.R. Wisniewski, S.T. DeKosky, Treatment of traumatic brain injury with moderate hypothermia, N. Engl. J. Med. 336 (1997) 540–546. K. Matsushima, M.J. Hogan, A.M. Hakim, Cortical spreading depression protects against subsequent focal cerebral ischemia in rats, J. Cereb. Blood Flow Metab. 16 (1996) 221–226. S. Miettinen, F.R. Fusco, J. Yrjanheikki, R. Keinanen, T. Hirvonen, R. Roivainen, M. Narhi, T. Hokfelt, J. Koistinaho, Spreading depression and focal brain ischemia induce cyclooxygenase-2 in cortical neurons through N-methyl-D-aspartic acid receptors and phospholipase A2, Proc. Natl. Acad. Sci. USA 94 (1997) 6500– 6505. J. Mintorovitch, G.Y. Yang, H. Shimizu, J. Kucharczyk, P.H. Chan, P.R. Weinstein, Diffusion-weighted magnetic resonance imaging of acute focal cerebral ischemia: comparison of signal intensity with changes in brain water and Na 1 ,K( 1 )-ATPase activity, J. Cereb. Blood Flow Metab. 14 (1994) 332–336. E. Morikawa, M.D. Ginsberg, W.D. Dietrich, R.C. Duncan, S. Kraydieh, M.Y. Globus, R. Busto, The significance of brain temperature in focal cerebral ischemia: histopathological consequences of middle cerebral artery occlusion in the rat, J. Cereb. Blood Flow Metab. 12 (1992) 380–389. M.E. Moseley, Y. Cohen, J. Mintorovitch, L. Chileuitt, H. Shimizu, J. Kucharczyk, M.F. Wendland, P.R. Weinstein, Early detection of regional cerebral ischemia in cats: comparison of diffusion- and T2-weighted MRI and spectroscopy, Magn. Reson. Med. 14 (1990) 330–346. T.B. Muller, O. Haraldseth, R.A. Jones, G. Sebastiani, F. Godtliebsen, C.F. Lindboe, G. Unsgard, Combined perfusion and diffusionweighted magnetic resonance imaging in a rat model of reversible middle cerebral artery occlusion, Stroke 26 (1995) 451–457. K. Nakashima, M.M. Todd, Effects of hypothermia on the rate of excitatory amino acid release after ischemic depolarization, Stroke 27 (1996) 913–918. M. Nedergaard, Spreading depression as a contributor to ischemic brain damage, Adv. Neurol. 71 (1996) 75–83.

M. A. Yenari et al. / Brain Research 885 (2000) 208 – 219 [44] M. Nedergaard, A.J. Cooper, S.A. Goldman, Gap junctions are required for the propagation of spreading depression, J. Neurobiol. 28 (1995) 433–444. [45] T.P. Obrenovitch, E. Zilkha, Inhibition of cortical spreading depression by L-701,324, a novel antagonist at the glycine site of the N-methyl-D-aspartate receptor complex, Br. J. Pharmacol. 117 (1996) 931–937. [46] J.C. Plumier, J.C. David, H.A. Robertson, R.W. Currie, Cortical application of potassium chloride induces the low-molecular weight heat shock protein (Hsp27) in astrocytes, J. Cereb. Blood Flow Metab. 17 (1997) 781–790. [47] B.M. Rodriguez, D. Sigg, F. Bezanilla, Voltage gating of Shaker K 1 channels. The effect of temperature on ionic and gating currents, J. Gen. Physiol. 112 (1998) 223–242. [48] J. Rother, A.J. de Crespigny, D.A. H, K. Iwai, M.E. Moseley, Recovery of apparent diffusion coefficient after ischemia-induced spreading depression relates to cerebral perfusion gradient, Stroke 27 (1996) 980–986. [49] J. Rother, A.J. de Crespigny, D.A. H, M.E. Moseley, MR detection of cortical spreading depression immediately after focal ischemia in the rat, J. Cereb. Blood Flow Metab. 16 (1996) 214–220. [50] R. Saito, R. Graf, K. Hubel, J. Taguchi, G. Rosner, T. Fujita, W.D. Heiss, Halothane, but not alpha-chloralose, blocks potassium-evoked cortical spreading depression in cats, Brain Res. 699 (1995) 109– 115. [51] S. Schwab, S. Schwarz, M. Spranger, E. Keller, M. Bertram, W. Hacke, Moderate hypothermia in the treatment of patients with severe middle cerebral artery infarction, Stroke 29 (1998) 2461– 2466. [52] M. Shimazawa, H. Hara, T. Watano, T. Sukamoto, Effects of Ca 21 channel blockers on cortical hypoperfusion and expression of c-Foslike immunoreactivity after cortical spreading depression in rats, Br. J. Pharmacol. 115 (1995) 1359–1368. [53] G.R. Sutherland, H. Lesiuk, P. Hazendonk, J. Peeling, R. Buist, P. Kozlowski, A. Jazinski, J.K. Saunders, Magnetic resonance imaging and 31 P magnetic resonance spectroscopy study of the effect of

[54]

[55]

[56]

[57]

[58]

[59]

[60]

[61]

[62]

219

temperature on ischemic brain injury, Can. J. Neurol. Sci. 19 (1992) 317–325. K. Takano, L.L. Latour, J.E. Formato, R.A. Carano, K.G. Helmer, Y. Hasegawa, C.H. Sotak, M. Fisher, The role of spreading depression in focal ischemia evaluated by diffusion mapping, Ann. Neurol. 39 (1996) 308–318. S. Takaoka, R.D. Pearlstein, D.S. Warner, Hypothermia reduces the propensity of cortical tissue to propagate direct current depolarizations in the rat, Neurosci. Lett. 218 (1996) 25–28. T. Tatlisumak, K. Takano, M.R. Meiler, M. Fisher, A glycine site antagonist, ZD9379, reduces number of spreading depressions and infarct size in rats with permanent middle cerebral artery occlusion, Stroke 29 (1998) 190–195. Y. Tohyama, K. Sako, Y. Yonemasu, Hypothermia attenuates hyperglycolysis in the periphery of ischemic core in rat brain, Exp. Brain Res. 122 (1998) 333–338. D.C. Tong, M.A. Yenari, G.W. Albers, M. O’Brien, M.P. Marks, M.E. Moseley, Correlation of perfusion- and diffusion-weighted MRI with NIHSS score in acute (,6.5 hour) ischemic stroke, Neurology 50 (1998) 864–870. M. Ueda, N. Watanabe, Y. Ushikubo, K. Kasai, T. Tsuzuki, K. Aoki, Y. Yamazaki, H. Samejima, The effect of hypothermia on CSD propagation in rats, No Shinkei Geka 25 (1997) 523–528. S. Warach, J.F. Dashe, R.R. Edelman, Clinical outcome in ischemic stroke predicted by early diffusion-weighted and perfusion magnetic resonance imaging: a preliminary analysis, J. Cereb. Blood Flow Metab. 16 (1996) 53–59. M. Yenari, J. Palmer, G. Sun, A. de Crespigny, M. Moseley, G. Steinberg, Time-course and treatment response with SNX-111, an N-type calcium channel blocker, in a rodent model of focal cerebral ischemia using diffusion-weighted MRI, Brain Res. 739 (1996) 36–45. M.A. Yenari, A. de Crespigny, J.T. Palmer, S. Roberts, S.L. Schrier, G.W. Albers, M.E. Moseley, G.K. Steinberg, Improved perfusion with rt-PA and hirulog in a rabbit model of embolic stroke, J. Cereb. Blood Flow Metab. 17 (1997) 401–411.