Blood-brain barrier disruption in experimental focal ischemia: Comparison between in vivo MRI and...

Magnetic Resonance Imaging, Vol. 12, No. 3, pp. 403-41 I, 1994 Copyright 0 1994 Elsevier Science Ltd Printed in the USA. All rights reserved

0730-725X194 66.00 + .OO

0730-725X(93)EOO57-U

0 Original Contribution

BLOOD-BRAIN BARRIER DISRUPTION IN EXPERIMENTAL FOCAL ISCHEMIA: COMPARISON BETWEEN IN VIVO MRI AND IMMUNOCYTOCHEMISTRY

ENG H. Lo,* YI PAN,* KEIGO MATSUMOTO,* AND NEIL W. KOWALL~

*Center for Imaging and Pharmaceutical Research, Massachusetts General Hospital, Harvard Medical School, MGH-East Bldg 149, Charlestown, MA 02129, USA, and

tGeriatric Research Institute, Bedford VA Medical Center, Bedford, MA, USA

The definition of blood-brain barrier (BBB) damage in cerebral ischemia using contrast-enhanced MRI has not been clearly correlated to the spread of edema or other histological measures of barrier disruption. In this study, we used a rabbit model of focal cerebral ischemia to compare GdDTPA-enhanced MRI with spin-echo images of brain injury and immunocytochemical detection of BBB damage and vasogenic edema. After 4 h of transient ischemia followed by 6 h of reperfusion, in vivo TZW and T,W images were obtained in a 1.5 T magnet using a 3-inch surface coil. After MRI, the animals were sacrificed and anti-serum protein (IgG) monoclonal antibodies were used to detect regions of increased BBB permeability to serum proteins. Ischemic neuronal damage was confirmed with cresyl-violet histology. T2W scans showed focal regions of increased signal intensity in the ischemic hemisphere (17.0 f 4.1%) that primarily involved the cortex and strlatum. TIW scans showed corresponding regions of hypointensity but demonstrated, in general, smaller lesion sizes (10.1 f 2.9%). GdDTPAsnbanced images showed variable areas of BBB disruption that included regions of intense leakage as well as lesions that only showed subtle enhancement along the periphery of damaged tissue. It appeared that large and more severe lesions corresponded to peripheral enhancement whereas smaller lesions showed central parenchymal enhancement. The extent of MR contrast enhancement did not correlate well with immunocytochemlcal images of serum protein leakage. Anti-IgG stains demonstrated widespread regions of BBB damage corresponding with areas of damaged neurons that appeared pyknotlc on cresyl-violet sections. Severely damaged neurons also showed cellular IgG uptake. These data suggest that GdDTPA leakage in cerebral ischemla can be variable in a model of focal ischemia with reperfusion and does not precisely correlate with vasogenic edema defined by serum protein extravasation across the BBB.

Keywords: Blood-brain barrier; Cerebral ischemia; Cerebral blood flow; MRI; GdDTPA; Immunocytochemistry.

INTRODUCTION

In cerebral ischemia, changes in blood-brain barrier (BBB) permeability are likely to induce ischemic edema that further exacerbates the development of injury. MRI provides a sensitive means of detecting cerebral ischemic damage primarily due to alterations in proton relaxation parameters.‘,* Prolongation of T, and T2 have been correlated with the degree of edema in various experimental models.3-5 However, the underlying mechanisms and spatiotemporal correlations between the spread of edema in cerebral ischemia and the degree and extent of BBB disruption are not clearly understood.

Contrast-enhanced MRI provides a noninvasive means of detecting BBB disruptions in vivo. In the normal brain, gadolinium (Gd) DTPA does not cross the BBB. Thus, enhancement with GdDTPA has been associated with increased BBB permeability.6-8 How- ever, the utility of GdDTPA-enhanced MRI in cerebral ischemia has been questioned since variable patterns of enhancement have been obtained in clinical series of acute and subacute cerebral ischemia. Within the first week of stroke, GdDTPA images may include arterial enhancement of the affected vasculature, meningeal enhancement, and parenchymal enhancement?-‘* These changes may further progress or subside over the course of days to weeks. lo With clinical data, there is no pos-

RECEIVED 7/2/93; ACCEPTED lO/ 11/93.

403

Address correspondence to Eng H. Lo, PhD.

404 Magnetic Resonance Imaging 0 Volume 12, Number 3, 1994

sibility of correlating the in vivo MR image with histological measurements of BBB damage. Uncertainties regarding the time of stroke onset as well as variable rates of reperfusion further complicate the proper in- terpretation of these results.

In this study, we have chosen a model of focal ischemia in the rabbit brain to investigate the distribution of BBB disruption as defined by GdDTPA leakage. The spatial extent of Gd leakage was compared with regions of injury observed on T, and Tl spin-echo images. Ad- ditionally, after the MRI, the ischemic brains were processed for immunocytochemistry to detect the leakage of serum proteins that provide a direct measure of BBB disruption and vasogenic edema. We tested the hypothesis that visualization of BBB damage with GdDTPA- enhanced MRI may not depict the entire extent of vasogenic edema as indicated by serum protein leakage.

MATERIALS AND METHODS

Animal Model All procedures were approved by the MGH Subcom-

mittee for Research Animal Care. New Zealand White rabbits (n = 7) were induced with 4% halothane and tracheostomized. Thereafter, they were artificially ven- tilated with l-2% halothane in a 10: 1 air/oxygen mix- ture with a stroke volume of 30-40 ml at about 20 strokes/min. Muscle relaxants (pancuronium bromide 0.2 mg/kg) were given IV approximately every 45- 90 min or as needed. Four hours of focal cerebral ischemia was induced by selective transorbital occlusion of the left internal carotid artery (ICA), middle cerebral artery (MCA), and anterior cerebral artery (ACA) using techniques that have been previously described.‘3,14 Briefly, enucleation was achieved with electrocautery and a craniotomy adjacent to the optic foramen was performed under a stereo-operating microscope. Micro- aneurysm Yasargil clips were used to transiently occlude the ICA, distal MCA, and the distal ACA. Reperfu- sion was established by removing the clips after 4 h. Ar- terial blood pressure and blood gases were monitored during the entire experiment. Base deficits were cor- rected as necessary with IV sodium bicarbonate. Rec- tal core temperature was kept at about 38°C with a heating lamp. After 6 h reperfusion, the rabbits were scanned as described below. During MRI, anesthesia was continued with IV alpha-chloralose (50 mg/kg).

MRI After 6 h reperfusion, animals were scanned in a

1.5 T magnet (GE Signa System) using a 3-inch surface coil taped securely over the animal’s head. Scout T, W (TR 400 ms, TE 20 ms) images were obtained in a sag- ittal plane to determine coronal slice locations from

the olfactory lobes to the edge of the cerebellum. Three- millimeter thick coronal slices were used with no inter- slice spacing, resulting in 10 images per brain. T,W scans (TR 2500 ms, TE 35 x 2 ms) were performed fok- lowed by T, W scans (TR 400 ms, TE 20 ms) pre and postcontrast. GdDTPA (Magnevist) was injected (0.5 mmol/kg) via a catheter inserted into a marginal ear vein. Images were obtained with a 8 cm square field- of-view in a 256 x 256 matrix resulting in voxel sizes of 0.3 x 0.3 x 3 mm. The digital images were then an- alysed to quantitate T, and T, lesion volumes (DIP- Station, CA). After adjusting the window and level for maximal contrast, the area of damaged tissue was drawn by hand. In cases where the edge of the lesion was more difficult to define, we used a semi-objective method where a cortical region-of-interest was drawn for the contralateral hemisphere and mean and standard deviations were obtained for the signal intensity. An intensity threshold was then set at mean + 2 SDS to delineate the lesioned area in the ischemic hemisphere. In pilot studies, we found that lesion volumes obtained by the operator-drawn method was not sig- nificantly different from the intensity threshold tech- nique. Areas of damaged brain were integrated along the slice thickness and calculated as a percentage of the ischemic hemisphere. For correlations between individ- ual image slices, areas were compared between T,W, Tl W, and GdDTPA-enhanced lesions.

Immunocytochemisfry After the MRI, the rabbits were sacrificed with an

IV overdose of sodium pentobarbital followed by trans- cardiac perfusion and fixation with 4% buffered para- formaldehyde. This flushes out all serum proteins from the vascular space so that only proteins that have leaked across a perturbed BBB would remain in the paren- chyma and thus be detected. Anti-serum protein (IgG) immunocytochemistry was performed using standard procedures. l5 Briefly, all brains were placed in 200/o glycerol and 2% dimethylsulfoximide in 0.1 M phosphate buffer (pH 7.3) prior to sectioning into 50-pm sections with a freezing microtome. Free-floating sections were preincubated in absolute methanol and 0.3% hydrogen peroxide for 30 min, followed by three lo-min washes in phosphate-buffered saline (PBS). Sec- tions were then incubated with 10% normal goat serum for 1 h followed by peroxidase conjugated goat anti-rabbit IgG antibody at room temperature for 3 h. After another wash in PBS, sections were placed in a solution of 0.5 mg/ml3,3’-diaminobenzene tetrahydro- chloride, 0.01 M imidazole, and 0.02% hydrogen peroxide in 50 mM Tris-HCl buffer (pH 7.6) for 10 min. Adjacent sections were processed for cresyl-violet his- topathology for confirmation of ischemic neuronal

MRI of BBB disruption in cerebral ischemia 0 E.H. Lo ET AL. 405

damage. Selected sections were also counterstained to confirm the overlap of IgG staining and cresyl-violet changes. We did not attempt to quantify the areas or volumes of IgG staining or cresyl-violet damage be- cause it would be impractical to obtain a sufficient number of %-pm sections to cover the entire brain. Ap- proximately 4-5 representative histological sections per brain were examined.

RESULTS

Systemic parameters were within normal limits during ischemia and reperfusion (Table 1). Following arterial occlusion, it was visually confirmed that distal portions of the occluded blood vessels were collapsed and the brain region supplied by the occluded vasculature demonstrated the lack of pulsatility. Immediately upon removal of the clips, it was noted that the blood vessels showed some vasospasm at the sites of clip place- ment. However, after about 5-10 min of reperfusion, vessel integrity had recovered and, in fact, showed increased caliber that may be related to the phenomenon of postischemic hyperemia. Although the actual size

Table 1. Systemic parameters (mean + SE)

Ischemic Reperfusion

Arterial blood pressure (mm&) 69 + 3.1 76 + 4.9

Blood pH 7.396 + 0.02 7.373 + 0.02 PGZ (mmHg) 132 + 13 141 +9 PC% (mmHg) 32.0 + 0.69 34.3 + 2.23 Rectal temperature (“C) 38.9 + 0.37 37.7 + 0.24

of these vessel caliber changes could not be quantified, these observations confirmed that reperfusion was achieved after clip removal.

T2W scans demonstrated extensive regions of increased signal intensity (Figs. la and 2a) consistent with the expected distribution of damage in this model. The lesions were primarily located in the cortex and basal ganglia although, in two cases, damaged tissue extended posteriorly to include the hippocampus. Tl W images showed, in general, smaller regions of injury defined by decreased signal intensity (Figs. lb and 2b). GdDTPA-

Fig. 1. In vivo MRI of focal cerebral ischemia (4 h ischemia followed by 6 h reperfusion) showed (a) increased signal intensity in the ischemic lesion on T2 W scans and (b) decreased signal intensity on T, W scans. (c) GdDTPA-enhanced scans showed focal areas of leakage in central portions of the ischemic lesion (arrows), whereas (d) anti-IgG immunocytochemistry showed extensive leakage of serum proteins.


Fig. 2. In vivo MRI showed (a) hyperintensity on T2W and (b) hypointensity on T, W MR images, but in this case, (c) GdDTPA scans only showed contrast enhancement in the peripheral edges of the ischemic lesion (arrows); and (d) anti-IgG images showed extensive distribution of serum protein leakage.

enhanced scans demonstrated focal regions of increased intensity in the ischemic hemisphere. In five animals, leakage of the contrast agent was intense (Fig. lc). In the other two animals, enhancement was subtle and only observable in the peripheral rims of the damaged brain regions (Fig. 2~).

Anti-serum protein immunocytochemistry showed extensive vasogenic edema in the ischemic hemisphere. IgG leakage was noted as increased interstitial staining in all seven animals (Figs. Id and 2d). This was true in all rabbits including the two animals with peripheral GdDTPA enhancement. In general, the leakage of serum proteins paralleled the distribution of ischemic neuronal damage observed with cresyl-violet histology and was more extensive than the areas of GdDTPA enhancement on MRI. It was also noted that, in regions of particularly severe injury within the core of the isch- emit lesion, IgG absorption into damaged neurons occurred (Fig. 3). However, there was no strict correlation between cellular absorption of IgG and the severity of GdDTPA leakage in the MRI scans.

Ischemic injury (mean + SEM) calculated as percent volume of the left hemisphere was 17.0 + 4.1% (T,W scans) and 10.1 + 2.9% (7’r W scans). The areas of in-

jury in each image slice on Ti W and T,W scans were highly correlated (r = 0.895, p < .Ol), although the lesion areas were slightly larger on T2 W compared with T1 W images (Fig. 4a). There were no significant correlations between the sizes of Tl W or T,W lesions with the extent of GdDTPA leakage (r = 0.496 and 0.429 respectively); (Figs. 4b and 4~).

DISCUSSION

Following an ischemic insult to the brain, a rapid cascade of pathophysioiogic events occur that lead to perturbed energy states, altered ionic gradients and neu- rotransmitter dynamics, cellular damage, and parenchymal infarction. I6 Recently, these molecular mechanisms of damage have been elucidated and shown to include abnormal activation of glutamergic transmission and calcium fluxes.16 In addition to these neurochemical perturbations, cerebrovascular events should also play a crucial role in the development of damage since cerebral ischemia is basically derived from a primary def- icit in flow. These include cellular swelling (cytotoxic edema), increase in BBB permeability with the associated passage of serum proteins and increased tissue wa-

MRI of BBB disruption in cerebral ischemia 0 E.H. LO ET AL.

Fig. 3. In central areas of severe ischemic damage, in addition to leakage of serum proteins through a disrupted BBB, tl also appeared to be cellular absorption of InG in nvknotic neurons (x32). This was probably related to increased membr permehility in damaged neurons.

ter content (vasogenic edema).i7 The spread of edema and swelling remains one of the most detrimental con- sequences of ischemia and is primarily related to BBB permeability alterations. Contrast-enhanced in vivo imaging should, therefore, be a useful way of visualizing the spread of edema within the ischemic lesion.

Previous studies of contrast-enhanced MRI in experimental cerebral ischemia suggested that contrast- enhanced MRI may demonstrate regions of BBB disruption during certain phases of infarction. GdDTPA in a cat model of focal ischemia demonstrated that acute phases of ischemia did not enhance but intense and rapid GdDTPA leakage was noted at 16-24 h post- ischemia followed by slow enhancement at 3-4 days.iE Another study using a rat model of photothrombosis suggested that GdDOTA improved the delineation of the cerebral injury primarily in older lesions.8 These studies, however, did not correlate the contrast- enhanced MR images with other direct measurements of BBB integrity.

Our study showed that 4 h of focal ischemia fol-

lere .ane

lowed by 6 h of reperfusion led to well-defined lesions detectable on spin-echo MRI. T2W scans were the most reliable in delineating the lesions; T, W lesions were, in general, smaller in volume. Significant correlations were found between areas of T,W injury and T, W injury. The prolongation of T2 and Tl is probably caused by the increased water content in the edem- atous brain. Regions of injury on MRI were found to correspond to pyknotic and shrunken neurons on sub- sequent histological analysis. However, there was no strict correlation between the distribution of the TZW lesions and the extent of GdDTPA leakage on contrast- enhanced scans. Two different patterns of enhancement were noted. In five animals, intense parenchymal leakage was observed in the central portions of damaged tissue. In two animals, the enhancement was subtle and restricted to the peripheral rims of the lesion volumes. Ischemic lesions in these two animals were sig- nificantly larger (30.8% on T2W scans compared to 12.1% for the other five animals) and appeared more severe on cresyl-violet histology, suggesting that more

% a

rea

Gd

enh

ance

men

t 5%

area

G

d e

nhan

cem

ent

9% ar

ea T

lW

MR

I le

sion

s lu

P

a,

N

P

U

J N

P

0)

0

0 0

0 0

0 0

0 0

0 0

0

0

e e Q

Q

W

0 0 0

Q


extensive ischemic damage is related to subtle peripheral Gd enhancement without central parenchymal leakage. The reason for this is presently unclear but it is possible that severe lesions have very little residual blood flow, thus delivery of the contrast agent to the site of injury is limited to the periphery. It is, therefore, possible that the severity of ischemic damage is, in fact, inversely related to the degree of MR contrast enhancement. This intriguing possibility warrants further investigation.

In contrast, anti-IgG immunocytochemistry demonstrated widespread regions of leakage that paralleled the T2 W lesions and the distribution of pyknotic neurons seen on cresyl-violet histology. The volumes of IgG leakage were not quantified since it was impractical to obtain sufficient 50-pm sections to cover the entire brain. However, all immunocytological sections examined (4-5 representative sections per brain) clearly showed more extensive leakage than GdDTPA enhancement. Although we only looked at IgG leakage, other studies have shown that there was no difference between the distributions of IgG leakage and that of all other serum proteins in ischemic edema.15 Regions of severe damage also showed cellular absorption of IgG. The absorption of IgG has been noted in a global model of ischemia,” and is probably caused by damaged cellular membranes allowing internalization of the IgG molecules. The lack of correlation between GdDTPA and IgG leakage may be due to various factors. The different molecular sizes of GdDTPA and IgG may lead to different penetration rates through the BBB. However, this is unlikely since GdDTPA, at a molecular weight of 590, is smaller than IgG at 200 kDa and should theoretically penetrate more easily if permeability was based on size alone. Another possibility is the differential diffusion rates through the brain pa- renchyma once the agents have leaked across the BBB. This is also an unlikely scenario, since IgG and other serum proteins, having more potential sites of interac- tion in various cerebral membranes, would diffuse less than the inorganic Gd ion. The most likely explanation of the limited enhancement shown by GdDTPA is the complex temporal development of BBB damage during reperfusion. Studies in feline models of cerebral ischemia have suggested that a biphasic pattern exists after ischemia.” Immediately following reperfusion, a transient increase in permeability occurs. This first phase is probably related to a reactive hyperemia that has been shown in several models.20~2’ After this, the barrier is reestablished and cellular damage develops over the course of several hours dependent on the severity of the initial ischemia. After 4-6 h, a delayed phase of BBB injury is expressed.20 In our study, se-

rum protein leakage may have occurred throughout the entire reperfusion period whreas GdDTPA leakage was probably limited to the secondary phase of BBB opening at the time of contrast injection.

It is important to realize that GdDTPA leakage was restricted to only a single timepoint during MR scan- ning whereas serum protein leakage occurred throughout the entire experiment. It is likely that if GdDTPA was infused throughout the reperfusion period, the distribution of enhancement would more closely approx- imate the spread of IgG. However, standard MRI protocols involve only single contrast administrations probably in the subacute phase similar to the timing of the MRI scans in our experiment. Our results suggest that contrast-enhanced MRI at single timepoint may not completely depict the spread of BBB damage and vasogenic edema that has occurred over the course of the stroke. Depending on when the GdDTPA is admin- istered, it may or may not catch the BBB at its leakiest. We have purposely chosen not to compare GdDTPA leakage with another externally applied tracer but rather we have decided to compare GdDTPA enhancement directly with the edema process itself. Serum protein is not an external marker that “predicts” BBB damage and vasogenic edema. It is an endogenous sub- stance that plays a central role in the pathophysiology of stroke. Therefore, by comparing GdDTPA enhancement with serum protein immunocytochemistry, we are directly comparing contrast-enhanced MRI with the central mechanism of BBB disruption and edema.

The relationship between BBB disruption and parenchymal cellular damage is still unclear. Some studies suggest that there is a threshold for BBB disruption; only tissue that suffered flow deficits below 10 ml/100 g per minute will show BBB damage.20 Others have suggested that ischemic damage to neurons alone is insuf- ficient and only fully expressed infarction including damage to blood vessels and astrocytes will cause BBB injury. 22 Regardless of the underlying mechanisms, it is important to recognize that the time scale of leakage is probably not as crucial as the spatial extent of leakage. Once serum proteins have crossed the barrier, an osmotic gradient is present and water influx will take place.

In clinical studies, Elster and Moody’ have described different classes of MRI enhancement including intravascular enhancement, meningeal enhancement, mixed compartment enhancement, and parenchymal enhancement. There was no clear pattern of temporal progression of these changes. Within the acute/subacute phase, intravascular enhancement may be due to sluggish flow rates in affected arteries. Arterial enhancement was most likely seen within the first 24 h and


disappearance of this vascular enhancement is most often associated with worsening parenchymal enhancement.10*‘2 The degree of parenchymal enhancement is variable and may include both reversible and irrevers- ible clinical symptoms. lo It is important to recognize that both the severity of the BBB disruption as well as the presence of adequate delivery of the contrast agents via some collateral supply may influence the degree of contrast accumulation. Some series have found that pa- tients with early (<24 h) intense enhancement may have a better prognosis.”

Recent advances in brain imaging have allowed in vivo measurements of cerebral damage in various neu- rologic diseases. However, the correlatioris between these in vivo images and more traditional indices of brain injury are often not well understood. As part of the overall effort in our Center to correlate in vivo imaging parameters with biochemical and histological endpoints, we compared contrast-enhanced MRI with immunocytochemical mapping of BBB disruptions in a model of transient focal ischemia in the rabbit brain. It was found that GdDTPA leakage was not well correlated with leakage of serum proteins. These differ- ences were likely to be related to the fact that leakage of serum proteins occurred throughout the entire experiment whereas GdDTPA extravasation was limited to the time of injection during reperfusion. Since the continued presence of serum proteins may induce osmotic gradients and water influx regardless of the time of onset, the temporal pattern of BBB disruption may not be as crucial as the spatial spread of leakage. This study is the first to directly demonstrate the potential limitations of contrast-enhanced MRI compared to his- tochemistry. Since vasogenic edema is strictly defined as a disruption in BBB to serum proteins that induce the formation of edema, 23 these data suggest caution in the use of GdDTPA-enhanced MRI at a single timepoint to estimate the extent of BBB injury and vaso- genie edema in cerebral ischemia.

Acknowledgments-The authors thank Drs. Leena Hamberg, Hong Jiang, and Gerald Wolf (MGH-Harvard) and Dr. Daniel Luttinger (Sterling-Winthrop) for helpful discussions. Pamela Taylor helped with the figures. Research supported by a grant from Sterling- Winthrop Pharmaceuticals.

REFERENCES

Brant-Zawadzki, M.; Weinstein, P.; Bartkowski, H.; Moseley, M. MR imaging and spectroscopy in clinical and experimental ischemia: A review. AJR 148:579-588; 1987. Matthews, V.P.; Barker, P.B.; Bryan, R.N. Magnetic resonance evaluation of stroke. Magn. Reson. Q. 8:245- 263; 1992. Naruse, S.; Horikawa, Y.; Tanaka, C.; Hirakawa, H.;

4.

5.

6.

7.

8.

9.

10.

11.

12.

13.

14.

15.

16.

17.

18.

Nishikawa, H.; Yosizaki, K. Proton magnetic resonance studies on brain edema. J. Neurosurg. 56:747-752; 1982. Kato, H.; Kogure, K.; Ohmoto, 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. 6:212-221; 1986. Fatourus, P.P.; Marmarou, A.; Kraft, K.A.; Inao, S.; Schwarz, F.P. In vivo brain water determination by TI measurements: Effect of total water content, hydration fraction and field strength. Magn. Reson. Med. 17:402- 413; 1991. Runge, V.M.; Price, A.C.; Wehr, C.J.;Atkinson, J.B.; Tweedle, M.F. Contrast enhanced MRI: Evaluation of a canine model of osmotic blood-brain barrier disruption. Invest. Radiol. 20:830-844; 1985. Virapongse, C.; Mancuso, A.; Quisling, R. Human brain infarcts: Cd-DTPA enhanced MR imaging. Radiology 161:785-794; 1986. Van de Vyver, EL.; Peersman, G.V. Experimentalcere- bra1 infarction in the rat: Utility of GdDOTA for MR imaging. Magn. Reson. Imaging 8:333-340; 1990. Elster, A.D.; Moody, D.M. Early cerebral infarction: Gadopentate dimeglumine enhancement. Neuroradiol- ogy I77:627-632; 1990. Crain, M.R.; Yuh, W.T.C.; Greene, G.M.; Loes, D.J.; Ryals, T.J.; Sato, Y.; Hart, M.N. Cerebral ischemia: evaluation with contrast-enhanced MR imaging. AJNR 12: 631-639; 1991. Sato, A.; Takahashi, S.; Soma, Y.; Ishii, K.; Kikuchi, Y.; Watanabe, T.; Sakamoto, K. Cerebral infarction: Early detection by means of contrast-enhanced cerebral arteries at MR imaging. Radiology 178:433-439; 1991. Yuh, W.T.C.; Crain, M.R.; Loes, D.J.; Greene, G.M.; Ryals, T.J.; Sato, Y. MR imaging of cerebral ischemia: Findings in the first 24 hours. AJNR 12:621-629; 1991. Lo, E.H.; Steinberg, G.K. Effects of dextromethorphan on regional cerebral blood flow in focal cerebral ischemia. J. Cereb. BIood Flow Metab. 11:803-809; 1991. Lo, E.H.; Steinberg, G.K. Effects of hypothermia on evoked potentials, MRI and cerebral blood flow in focal ischemia. Stroke 23:889-893; 1992. Scmidt-Kastner, R.; Szymas, J.; Hossman, K.A. Immu- nohistochemical study of glial reaction and serum protein extravasation in relation to neuronal damage in rat hippocampus after ischemia. Neuroscience 38:527-540; 1990. Siesjo, B.K.; Bengtsson, F. Calcium fluxes, calcium an- tagonists and calcium-related pathology in brain ischemia, hypoglycemia and spreading depression: A unifying hypothesis. J. Cereb. Blood Flow Metab. 9: 127- 140; 1989. Klatzo, I. Concepts of ischemic injury associated with brain edema. In: Y. Iaba, I. Klatzo, M. Spatz (Eds). Brain Edema. New York: Springer-Verlag; 1985: pp. l-5. McNamara, M.T.; Brant-Zawadzki, M.; Berry, I.; Pereira, B.; Weinstein, P.; Derugin, N.; Moore, S.; Kucharczyk, W.; Brasch, R.C. Acute experimental ce-


rebral &hernia: MR enhancement using GdDTPA. Ra- diology 158:701-705; 1986.

19. Williams, C.; Jenkins, L.; Povlishock, J. Postischemic neuronal flooding with IgG: An early predictor of hip- pocampal cell death. Sot. Neurosci. Abstr. 15805; 1989.

20. Kuroiwa, T.; Ting, P.; Martinez, H.; Klatzo, I. The biphasic opening of the blood-brain barrier to proteins following temporary middle cerebral artery occlusion. Acta Neuropathol. 68:122-129; 1985.

21. Welsh, EA. Role of vascular factors in regional ischemic injury. Prog. Brain Res. 63:19-27; 1985.

22. Petito, C.K.; Pulsinelli, W.A.; Jacobsen, G. ; Plum, F. Edema and vascular permeability in cerebral ischemia: Comparison between ischemic neuronal damage and infarction. J. Neuropathol. Exp. Neural. 41~423-436; 1982.

23. Betz, A.L.; Iannotti, F.; Hoff, J.T. Brain edema: A clas- sification based on blood-brain barrier integrity. Cere- brovasc. Brain Metab. Rev. 1:133-154; 1989.

Blood-brain barrier disruption in experimental focal ischemia: Comparison between in vivo MRI and...

Documents

Transcript of Blood-brain barrier disruption in experimental focal ischemia: Comparison between in vivo MRI and...