Evolution of Photochemically Induced Focal Cerebral Ischemia in the Rat
Magnetic Resonance Imaging and Histology
Background and Purpose Magnetic resonance imaging (MRI) is increasingly used to study the pathophysiological evolution of cerebral ischemia in humans and animals. We have investigated photochemically induced (rose bengal) focal cerebral ischemia, a relatively noninvasive, reproducible model for stroke, and compared the evolution of the ischemic response in vivo and postmortem with MRI and histology, respectively.
Methods MR images weighted for T2, diffusion, and T2* and parallel histological sections stained with cresyl fast violet (CFV) and for glial fibrillary acid protein were obtained from 34 adult male Hooded Lister rats at seven time points (3.75 to 196 hours) after bilateral ischemia induction. From CFV histology, lesion volumes and cell counts were calculated; from diffusion-weighted and T2-weighted images, apparent diffusion coefficients and lesion volumes were determined.
Results Both MRI and histology revealed a well-defined lesion at 3.75 hours after irradiation and a consistent pattern of temporal evolution; lesion apparent diffusion coefficients decreased significantly by 3.75 hours, increased significantly by day 2, and correlated strikingly with the decline in lesion CFV-positive cell numbers. After day 2, astrocytes and connective tissue cells invaded the infarct. Throughout the time course, lesion volumes determined in vivo and postmortem (after shrinkage correction) agreed well.
Conclusions MRI changes quantitatively reflect histopathology, revealing reproducible primary and secondary damage characteristics noninvasively. These changes essentially replicate those reported for other animal stroke models and clinically, emphasizing the value both of MRI and the photochemically induced focal cerebral ischemia model in stroke research.
Magnetic resonance imaging provides a unique method for investigating changes of brain water that occur during and after cerebral damage. When disruption of blood supply to the brain is sufficiently severe and prolonged as in stroke, neurons and other cellular components will die.1 In principle, it is possible to differentiate ischemia, edema, and infarction in a stroke lesion by the use of different MRI sequences, but unfortunately, MRI does not have adequate signal sensitivity or spatial resolution to visualize directly those cellular and biochemical components involved. Consequently, there is considerable interest in exploring a variety of forms of MR contrast in the hope that some of those will be sensitive to cellular and biochemical changes. Because it is unethical for those development studies to be made on human subjects, it is necessary to use animal models of cerebral ischemia, even though none of them will perfectly mimic the events that occur in clinical stroke.2
In the photochemically induced focal ischemia model,3 an in vivo photochemical reaction is used to induce a thrombosis leading to cerebral infarction. Reaction between the photosensitive dye rose bengal, administered intravenously, and light, via illumination of the external skull, produces highly reactive singlet oxygen molecules that react with endothelial cells to cause platelet aggregation and BBB breakdown.4 Among its positive features, this model is very reproducible, does not require craniotomy, and does not impair long-term survival. Furthermore, good control over the eventual size and location of the lesion can be achieved by manipulation of the intensity, duration, beam shapes, and position of the irradiating beam and the dose of the photosensitive dye. It is also relatively unaffected by macroscopic differences in the cerebral vascular architecture. Criticisms of the model include its end-arterial occlusive nature, apparent absence of a penumbra or rescuable region, and prominent microvascular injury resulting in early BBB opening and vasogenic edema.
This study was designed to compare changes in vivo (MRI) and postmortem (histology) resulting from photochemically induced focal cerebral ischemia in the rat. The aims were: (1) to provide an understanding of the evolution of the photochemically induced lesion; (2) to develop further insight into the cellular processes revealed by DW MRI; and (3) via the above, either to confirm objectively or rebut the numerous criticisms of this model.
We report during the first week of photochemically induced cerebral ischemia the temporal and spatial responses of MR images weighted for T2 (sensitive to static tissue water changes), diffusion (sensitive to translational water-proton motion but containing some T2 information), and T2* (susceptible to static field inhomogeneities, eg, to deoxyhemoglobin); from T2WI and DWI, the ADC (a measure of water-proton diffusibility) was mapped. The qualitative and quantitative correlations between these MRI protocols and histopathology are discussed.
Materials and Methods
Thirty-four male Hooded Lister rats (250 to 350 g; Olac Harlan, UK) were studied. Median lesion time points and the number of animals were 3.75 hours (3), 5.6 hours (4), 12.75 hours (6), 25.25 hours (5), 2 days (4), 4 days (4), and 7 days (4). Four animals were sham operated and imaged at 4 hours and 24 hours (see below). All animal procedures were performed in accordance with institutional guidelines and with legislative approval.
Each animal was placed in a stereotaxic frame (Stoelting Co). Anesthesia was induced and maintained with 2% halothane in oxygen via a purpose-built mask fitted over the stereotaxic frame. Rectal temperature was kept constant at 36.5±0.5°C with a homeothermic blanket. With the use of aseptic techniques, a midline scalp incision and pericranial tissue dissection were performed to expose the skull surface.
For skull illumination, standard equipment (Oriel Scientific Ltd) was used. A bifurcated glass-fiber light guide (diameter of 3.2 mm, Oriel part No. 77533, filtering ultraviolet light) from a 300 W xenon arc lamp (lamp No. 6259, housing No. 66066) was positioned on the skull surface at the bregma, 2.5 mm to the left and right of the midline. The lamp output was cooled with an in-line water filter (removing infrared wavelengths) to prevent burning and was insensible to touch; a thermistor placed on the skull surface did not vary by more than ±0.5°C. The light emitted from the light guide (measured with a broadband thermopile constructed by Dr Keith Langmach, Department of Medical Physics, Addenbrooke's Hospital, Cambridge, UK, and calibrated by the UK National Physical Laboratory) had an essentially flat spectrum (range, 400 to 1200 nm) with intensity 0.5 mW·cm−2·nm−1.
A dose (1 mL·kg−1) of rose bengal dye (Sigma Chemical Ltd; 20 mg/mL in sterile 0.9% NaCl) was injected slowly (10 to 20 seconds) via a lateral tail vein, and the skull was immediately illuminated for 5.0 minutes. The light was then interrupted, the scalp wound closed with sutures and tissue adhesive, and the animal allowed to recover. Sham-operated control animals were prepared with the same procedure; however, the rats were either not illuminated after receiving rose bengal (n=2) or sterile 0.9% NaCl was given with skull illumination (n=2). Recovery from anesthetic was uneventful, although it was prolonged in lesioned animals (≈30 min); having recovered fully, lesioned animals were indistinguishable from controls and did not suffer obvious behavioral or other deficits after return to their home cage. (We routinely use bilateral occlusion to detect the effects of neuroprotective agents, the conditions described providing reproducible results; this protocol was therefore also used in the present work.)
MRI: Animal Handling
For imaging, anesthesia was induced with 3% halothane in oxygen. The animal was intubated endotracheally and mechanically ventilated with a Harvard Miniature Ventilator (Harvard Apparatus Ltd; stroke rate, 45 min−1; stroke volume, 3.5 mL); anesthesia was maintained with 1.5% halothane in oxygen. Rectal temperature was kept constant at 36.5±1°C by use of a homeothermic blanket. Heart rate was continuously monitored by use of shielded subcutaneous electrodes. The rat head was mounted in a custom-built nonmagnetic stereotaxic frame. A 2.5-cm-diameter circular receive-only surface coil was mounted firmly onto the stereotaxic frame with the center of the coil located 8 mm in front of the ear line; the whole assembly was then placed inside a transmit-only volume resonator.
MRI was performed with the use of an Oxford Instruments 31-cm horizontal bore, 2.0-T superconducting magnet driven by a Bruker MSL 100 console operated with TOMIKON software (Bruker Medizintecknik GmbH). Custom-built 10-cm (ID) Golay type (X and Y directions) and Maxwell (Z direction) gradient coils were driven by a series pair of Techron 7570 amplifiers capable of giving a maximum gradient strength of 146 mT/m.
MRI: Image Acquisition
Fast gradient-echo pulse sequences with TR of 200 ms, TE of 7 ms, flip angle of 30° (FE200/7/30), and 2 excitations per scan (Nex 2) were used for pilot images, in the order of coronal, horizontal, and sagittal to provide reproducible orthogonal coronal target images. Contiguous 1-mm-thick coronal target slices were then positioned on the pilot sagittal image to cover the entire length of the infarct.
The target image matrix was 256×256 pixels (0.1-mm in-plane resolution). DWIs (TR=2500 ms, TE=49 ms, Nex 4) were acquired by use of a Stejskal-Tanner sequence5 with paired diffusion-sensitizing gradients of 67 mT/m applied 25 ms apart for 10 ms. T2WIs were acquired before DWIs by use of the same sequence but without the diffusion gradients. From the signal intensities in paired DWIs and T2WIs, the average values of the ADC of the water protons in each image voxel were calculated from the Stejskal-Tanner equation5 :\frac|<|\mathit|<|S|>||>||<|\mathit|<|S|>|_|<|0|>||>||<|=|>|\mathit|<|e|>|^|<|\mathit|<||<|-|>|b(ADC)|>||>|where\mathit|<|b|>||<|=|>||<|\gamma|>|^|<|2|>|\mathit|<|G|>|^|<|2|>||<|\delta|>|^|<|2|>|(|<|\Delta|>||<|-|>||<|\delta|>|/3)(S and S0 are signal intensities found respectively in the DWIs and T2WIs, obtained by use of fixed values of γ, δ, G, and Δ, respectively the proton gyromagnetic ratio, duration, strength, and interpulse timing). Quantitative DMs of the calculated ADC values were then constructed.
DWIs contain information from T2 as well as diffusion, making their interpretation difficult when significant tissue edema is present. Where diffusion is restricted (ie, ADC values are reduced), DWI signal intensity appears to increase because restricted water protons experience relatively less variation in the magnetic gradient and retain the coherence of their spins; conversely, increased diffusion in DWIs results in protons losing more spin coherence, thus reducing signal intensity. By contrast, in DMs, pixel intensity is determined solely by the ADC value itself, so reduced ADC values in the DMs correspond with reduced intensity.
T2*WIs were obtained by use of a gradient echo sequence (FE800/40/50, Nex 8). A 128×128 image matrix was used with the same image slice positioning as DWIs and T2WIs.
For further analysis, image data were transferred from the MR console by Ethernet to computers operating CaMReS image processing software (Dr N.J. Herrod) in Unix, C, and X-Window system environments.
Under terminal anesthesia (halothane 3% in oxygen), brains were fixed for histology by transcardiac perfusion with ≈25 mL sodium phosphate buffer (0.1 mol/L, pH 7.4) containing sodium heparin (Multiparin; 25 U/mL), followed immediately by ≈250 mL sodium phosphate buffer containing paraformaldehyde (Sigma; 4% wt/vol) and glutaraldehyde (Sigma; G6257: 0.5% vol/vol). Perfusion fixation was done ≈50 minutes after starting the DW scan; thus, the histological time is later than the MRI timing, but for clarity, timing to the start of the DWI is usually quoted. Brains were immediately removed and kept in the fixation buffer at 4°C. They were subsequently cryostat sectioned (slice thickness, 20 μm) and stained (1) for GFAP (to assess tissue reaction to the ischemic insult) by use of rabbit anti-cow GFAP primary antibody (Dako Ltd) and a biotin-avidin–horseradish peroxidase kit (Vectastain; Vector Laboratories) and (2) with CFV, a Nissl stain for RNA (to assess cellular morphology and numbers).
In all animals, cell numbers in both (bilateral) lesions were counted. Three CFV-stained sections spanning the lesion length (front, middle, and rear) were visualized with a color camera with the use of image analysis software (Optimas; BioScan Inc). A FOV (0.21×0.16 mm; ×40 objective) was positioned within the center of the lesion and within the normal, temporal cortex in the same section. All cells within the FOV were counted manually. Cell counts were similar in all the chosen sections, and therefore, the means of the results for each animal were taken.
In serial, CFV-stained, 20-μm-thick sections, lesion areas were quantified with the use of image processing software (Quantimet 920, Leitz) after the lesion boundary was outlined manually. Lesion volumes were calculated from these areas by use of Simpson's rule.6
To compensate for image signal attenuation away from the surface coil, T2WIs were mathematically corrected to give a symmetrical signal-intensity profile across the image; the lesion volumes were then quantified by summating intensity-thresholded lesion areas. We measured DM lesion volumes by manually outlining the lesions, because their heterogeneity at later time points precluded the use of intensity thresholding.
To determine shrinkage due to histological processing, the areas of equivalent CFV-stained histological and T2WI brain slices from one animal per time point were measured.
Statistical differences were assessed by use of Student's t test and one-way ANOVA (for histology). Quoted values are mean±SEM.
In the control animals, no pathological changes were detected at 4 and 24 hours (the time points examined) with DW or T2W MRI or at 24 hours with histology. In contrast, the lesions were clearly visible in all test animals (receiving both illumination and rose bengal dye intravenously) at all time points. Because the illumination was known to be cold (by touch and measurement) and ultraviolet- and infrared-filtered, lesion development was due to dye-light interaction and not to burning or the actions of the light source alone. Within each group, images compared very well in terms of both size and development.
At 3.75 hours after illumination, the lesions were already large and clearly visible (Fig 1⇓). T2WIs (sensitive to static tissue water) revealed a hyperintense lesion, indicating edematous changes, with patches of isointensity; DM (the diffusion map of quantitative ADC values) at 3.75 hours revealed a uniformly hypointense region, indicating decreased water diffusion throughout the lesion. By ≈6 hours, T2WI intensity appeared to be further increased peripherally, and ADC was still reduced uniformly throughout the lesion at this time point.
At 12 to 24 hours after lesion induction, the lesion was still homogeneously hyperintense in T2WIs, although it had extended in size: by 12 hours, the hyperintense region included the corpus callosum in most animals. DM now revealed a lesion core with increased ADC, an observation more obvious at 24 hours, but a clearly demarcated hypointense rim also existed with decreased diffusion. The T2*WI (sensitive to blood “color” changes caused by altered deoxyhemoglobin levels) showed a similar pattern to T2WI but with a more heterogeneous appearance deriving from radial vein-like structures in the lesion.
Two days after lesion induction (Fig 2⇓), T2WI lesions had a hyperintense rim round the patchily hypointense or isointense lesion core. At this time point, DMs also showed a sharply delineated border, indicating decreased diffusion within it, but a core in which diffusion now exceeded normal values.
From day 4 (until day 7), the small dark patches in T2W lesion images grew in size and became increasingly well defined; these changes were more exaggerated in T2*WIs. When DMs were used, lesion heterogeneity was also visible, with patches of both decreased and increased diffusion. Some of the patches of decreased diffusion appeared to correspond to regions of decreased T2 intensity. The ventricles appeared enlarged, and this was especially obvious in DWIs (containing information from both the T2 and diffusion of water molecules) and DMs (constructed from ADC values alone).
Fig 3A⇓ and 3B⇓ shows that at 3.75 hours (timed to the DWI, corresponding to ≈4.5 hours after lesion induction), a loss of cytoplasmic Nissl staining had occurred throughout the lesion, clearly demarcating it. Cell morphology was strikingly changed, from a round, plump appearance in which the perinuclear cytoplasm appeared to take up CFV best to a triangulated morphology, reduced in size, in which cell processes were obvious and the cell body stained uniformly. A characteristic blebbing morphology and condensation of the nuclear material were observed in some cells. Cell counting showed that lesion cell numbers had declined only by ≈30% at 3.75 hours compared with the unlesioned temporal cortex, from 108.3±7.02 to 73.51±4.93 (×103), although the cell size reduction combined with cytoplasmic pallor suggested the cell count was much lower. Fig 4A⇓ shows that the cell numbers in the lesion continued to fall after 5.6 hours (when numbers apparently increased from 3.75 hours, reflecting interanimal variation rather than a real change) but did not reach a negligible count before day 4. GFAP, a sensitive marker for neuronal cytotoxicity,7 revealed activated astrocytes around the lesion and in the corpus callosum (not shown) but few in normal brain.
At 12 to 24 hours, cells were further reduced in number in CFV-stained sections, and the survivors appeared smaller; the blebbing morphology appeared more common. Polymorphonuclear-like cells started to appear on the lesion surface and in some superficial vessels.
By day 2 (Fig 3C and 3D⇑⇑), very few neuronal cell bodies (10%) remained in the lesion; polymorphonuclear-like cells were abundant but became replaced by round, mononuclear cells with undifferentiated CFV-staining structure. The boundary region of the lesion and the neighboring corpus callosum contained many well-defined astrocytes with long processes. At the later time points, astrocytosis was even more marked, the lesions were filled with mononuclear cells, and angiogenesis was apparent; amorphous deposits of fibrinlike material could also be visualized.
Lesion Volume Changes
With the use of equivalent brain slices visualized histologically and with MRI, it was possible to calculate the shrinkage due to histological processing. Slice shrinkage was very reproducible (between time points, whether comparing lesion or whole-brain image-slice areas) with a MRI:CFV brain area ratio of 1.42±0.064 (n=7 animals). Assuming the histologically fixed brain shrinks equivalently in all three dimensions, the CFV histology lesion volumes were appropriately corrected (×1.69=1.423/2). Fig 5⇓ shows the corrected CFV postmortem data and in vivo T2W and DM lesion volume in absolute and time-course terms; they agree well. The lesion volume increased to a maximum at 12 to 24 hours before declining thereafter. By the end of the time course under study, the lesions were considerably reduced in size, perhaps because of the reduction in edema and infarct retraction.
There was a significant decrease (P<.05) in lesion ADC at the earliest time point (3.75 hours), indicating an acute reduction in water-proton diffusibility. However, as Fig 4B⇑ shows, from 24 hours onward, ADC significantly increased (P<.05): values (×10−3 mm2·s−1) were 0.43±0.01 (3.75 hours) and 0.41±0.01 (5.6 hours), increasing to 0.9±0.03 (day 7). The ADC for the corresponding lesion-free temporal cortex was 0.70±0.06×10−3 mm2·s−1, and in control animals, 0.71±0.03×10−3 mm2·s−1. ADC values in the range 0.6 to 0.8×10−3 mm2·s−1 thus lie within 1 SD of normal, and values outside this range can be considered abnormal.
Plotting the proportion of the lesion volume (calculated from the DM) with ADC values below or above those limits illustrates lesion evolution: although the proportion of the lesion with increased ADC (ie, increased water-proton diffusibility) is low at the earliest time point, the proportion increases as the lesion progresses (Fig 4C⇑). By contrast, the proportion of the lesion with reduced ADC values (ie, reduced diffusion) is high early on but decreases with time. The lesion ADC values were inversely and very closely related (r=.98) in a linear manner to the lesion cell count in CFV-stained sections (Fig 4D⇑).
This time-course study compared in vivo (MRI) and postmortem (histology) changes resulting from photochemically induced focal cerebral ischemia in the rat brain. The results reveal notable and striking correlations between the two methods, both quantitatively and qualitatively.
Early Changes in ADCs and T2WIs
We used MR images weighted for T2, diffusion, and T2*. In T2WIs, image contrast is known to reflect static (“tumbling” rather than translational) changes in tissue water content (eg, through vasogenic edema).8 However, it has been reported that early after temporary or permanent MCAO, T2WIs fail to clearly demonstrate brain injury.9 10 DWI is based on the natural sensitivity of MRI to the diffusibility of water protons, but DWIs also contain T2 information. DWI is sensitive to early ischemic changes and has been widely studied in various occlusion models9 10 11 12 and in the photo-induction model,13 14 but the T2 content is unhelpful for studying diffusional change when tissue edema also develops. When paired T2W and DW images are acquired by use of pulse sequences differing only in their degree of diffusion weighting (see Equations 1 and 2), a quantitative DM solely of brain ADC values can be generated. T2*WIs are sensitive to susceptibility effects caused by static field inhomogeneities, in particular deoxyhemoglobin, which, unlike oxyhemoglobin, is paramagnetic; thus, contrast in T2*WIs is blood oxygenation level (or blood “color”) dependent.15 16 This pulse sequence has been used to study changes in regional cerebral blood flow and oxygen consumption depending on regional brain activation.
At the earliest time point studied (3.75 hours), the ischemic lesion was readily identifiable in T2WIs, indicating acute vasogenic edema due to BBB breakdown.17 18 This was observed also by Lanens et al,19 who showed BBB opening with postcontrast MRI, and van Bruggen et al,13 who showed BBB disruption using Evans Blue dye shortly after lesion induction. Furthermore, the prompt reduction in ADC after ischemia observed in the present study (Figs 1C and 4B⇑⇑) is similar in magnitude to reports for all other models regardless of etiology.9 13 14 20 It may be asked how a net reduction in water diffusion can occur when vasogenic edema is increased; there appears to be no simple, unique answer, and a number of mechanisms probably contribute.
First, it is suggested that the significant reduction in water diffusion during early ischemia is related to cytotoxic edema9 21 22 : the cessation of blood supply leads to an acute deprivation of oxygen and energy that compromises cellular components metabolically; this results in the loss of cellular osmoregulation, leading to a shift of water from extracellular to intracellular compartments with lower ADC values. In normal brain tissue, the ECS occupies 20% of the total brain volume, but during the early stages of cerebral ischemia, this is reduced to only 10% as cellular swelling occurs. Benveniste et al22 showed that in the rat MCAO model, changes in the ratio of ECS to ICS (80:20 to 90:10) could account almost entirely for the change in ADC from 0.76×10−3 to 0.43×10−3 mm2·s−1 during cytotoxic edema.
Although the reduction in ADC may represent a relative transfer of water from extracellular to intracellular compartments22 in which diffusion is restricted by intracellular components (eg, mitochondria and endoplasmic reticulum23 ), the movement of cytoplasm through the neuropil, which is also partly energy dependent, may be reduced too. The lack of perfusion and hydrostatic pressure, or perhaps extravasation of protein (into a relatively protein-free cerebrospinal fluid), may also decrease the net mobility of water.
Nevertheless, the increased T2W signal intensity may still reflect an alteration of the ECS compared with normal tissue due to vasogenic edema24 through microvascular injury.17 However, acute lesion pathology (as witnessed by the reduction in cell count) may also change the macromolecular environment of the ECS (which affects T2): the necrotic release of intracellular enzymes could degrade the extracellular matrix and thus alter the behavior of extracellular lesion water molecules (and T2W changes in cerebral ischemia are thought to indicate irreversible cellular pathology).
Although it is known that a 1°C decrease in temperature decreases the brain ADC by 0.013×10−3 mm2·s−1,21 local hypothermia is unlikely to be sufficient to account totally for the magnitude of the observed reduction. Moreover, unlike global and MCAO models, the lesion volume in the model used in the present study was small and discrete and thus would have its temperature maintained by the surrounding area that was still receiving normal blood flow.
Thus, a number of different mechanisms must contribute to the changes in T2WIs and ADC observed in stroke models and clinically; that MRI can reveal such pathophysiology noninvasively emphasizes its value, even though the underlying mechanisms are still debated and require further investigation.
Later Changes in Histopathology and MRI
Histopathological findings (Fig 3A⇑) revealed that in the short term, neurons appeared to react “defensively” to photothrombosis: their morphology was changed, but they still stained with CFV (a quasi “vital” stain selective for perinuclear ribosomal RNA, which in vivo is rapidly recycled) rather than lysing immediately. This fits with evidence that neuroprotection is possible several hours after ischemia induction.25
By 12 hours, edema in the corpus callosum is visible in all three MR sequences. Increased diffusion in the corpus callosum is expected because fluid movement from the lesion will tend to occur along the fiber tracts, which offer a path of least resistance. At about this time point, diffusion within the core of the lesion becomes increased compared with the lesion periphery. The degree of increase varies between animals. The increasing lesion ADC values, which show an inverse trend to histological cell count (Fig 4D⇑), indicate cell loss; thus, increased ADC in the core may be due to incipient necrosis so that the water molecules are no longer confined to the ICS but diffuse freely. In the lesion border, the reduction in ADC may continue, in part, due to fresh secondary damage caused by the migration of various potentially harmful free fatty acid by-products through the interstitial fluid.3 17 The presence of activated astrocytes (staining GFAP-positive) could also retard water movement. The radial lines observed in T2*WIs may perhaps reflect a susceptibility artifact caused by the paramagnetic property of deoxyhemoglobin in the blood vessels.
From day 2 on, the hypointense patches on T2WIs with correspondingly decreased diffusion probably reflect a brisk gliosis, as observed histologically. Some of the patches of hypointensity seen on T2WIs, at this and later time points but not earlier, may also be due to angiogenesis and possibly hemorrhagic transformation at the lesion boundary. The presence of blood pooling or perhaps hematoma is supported by their larger appearance in T2*WIs, suggesting a marked magnetic susceptibility, but because the brains were perfused with a large volume of fixative immediately postmortem, very few vessels within the brain contained sufficient cells to confirm definitely either possibility. In some animals, the lesion core contained amorphous, fibrinlike deposits. Thus, at this later stage, the lesion progresses into a classic wound-healing phase in which inflammatory cells invade the infarct and replace the necrotic tissue with a vascularized connective tissue matrix. DWI no longer clearly delineates the lesion beyond 2 days, except for a few isolated patches of edematous foci with increased diffusion. Clearance of vasogenic edema fluid is thought to be through the ventricular wall into the cerebrospinal fluid.26 This may explain the enlarged ventricles with increased diffusion. Scar tissue development may cause the lesion to contract, also resulting in ventricular enlargement.
When lesion area is calculated, the volumes derived from T2WI and DM correlate well with histological findings but may differ in their clarity depending on the time after infarction. By 4 hours, T2WI and DWI are both equally good at depicting the lesion, then, at later stages, lesions in T2WI become more clearly defined. T2WI and T2*WI provide essentially similar contrast, but at the later time points (2 days and beyond), T2*WIs are too prone to gross magnetic susceptibility effects to give reliable volume measurements, although they can highlight areas with angiogenic or hemorrhagic changes. Generally, the characteristic MR changes observed in vivo reflected the underlying histological changes. The good correlation, observed here for the first time, between lesion ADC changes in vivo and cell loss investigated postmortem suggests the possibility that ADC measurements may indicate cell viability, and hence their importance in studying ischemic evolution.
Comparisons With Other Stroke Models
Forsting et al27 suggested that the changes observed in “occlusion” models (eg, MCAO) closely replicate human brain ischemia and are therefore preferred for stroke research. However, because no animal model is able to exactly replicate the pathology and biochemistry of human stroke, it is important to place the current results in the context of other ischemia studies. Here, we have demonstrated that the early precipitous fall in lesion ADC is similar to that observed in occlusion models.9 10 28 Although the photochemically induced lesion has rather early microvascular injury, BBB breakdown, and vasogenic edema, cellular changes are still evolving and follow a similar trend to those of the occlusive models. This suggests that the underlying neurochemical pathology, notably, calcium- and excitatory amino acid–mediated events, continues to play an important role in subsequent neuronal death,29 30 31 just as it does in occlusion models. This is supported by various studies demonstrating that neuroprotective compounds acting on this common pathway reduce or reverse the extent of the lesion in the photochemically induced ischemia model.25 32
The other criticism of the photochemically induced ischemic lesion is that it has little or no rescuable penumbral region and is therefore not susceptible to neuroprotective agents. However, animal models of stroke appear to be somewhat strain dependent in their behavior,33 34 perhaps accounting for some negative findings, and we have found photothrombotic ischemia to be treatable.32 Indeed, the results presented here clearly demonstrate that the lesion size increases by 80% between 4 and 24 hours (Fig 5⇑). Furthermore, DWI also clearly shows a core and periphery with distinct ADC values between 12 and 24 hours. This may be due to the development of delayed oxidative and reactive free radical damage35 36 37 typically prevalent in ischemic tissue and may be useful for evaluating a wide range of neuroprotective agents against secondary damage. The involvement of endothelial injury and platelet aggregation also indicates the potential of this model for evaluating antiplatelet and thrombolytic agents.38
Given the difficulty of using any form of animal pathology as a model for human brain ischemia, it is important to integrate the current results with those of previous MRI studies13 14 18 19 39 in photochemically induced ischemia and to delineate our additional findings. We have confirmed that (1) DWI reveals a lesion core and periphery with distinct ADC values,13 14 (2) the damaged area increases significantly between 4 to 24 hours19 and declines thereafter,39 and (3) MRI can visualize the evolution of this lesion and provide a good correlation with histological findings.14 18 In addition, we observed that (4) hyperemic vasogenic edema tends to occur rapidly, but at the acute stage, most of the cellular components are still intact; (5) the lesion ADC changes in vivo closely reflect the lesion cell loss, suggesting that DW MRI reflects lesion cellular status and hence may predict ischemia outcome40 ; and (6) by using a larger number of animals than previous studies, we have demonstrated that the photochemically induced lesion size and characteristics are reproducible throughout the lesion evolution. Taken together, these features provide further support for the use of MRI, and this particular model, for evaluating a wide range of putative neuroprotective therapeutic agents. Finally, we note that the photochemically induced ischemia model has excellent ergonomics: it is simple to perform, is reproducible, and is relatively noninvasive, and, despite the fact that it is considered not to reflect well the ischemic events observed clinically, the ADC and cellular changes reported here are very similar to those observed after MCAO in animals and stroke in humans.41 42
Selected Abbreviations and Acronyms
|ADC||=||apparent diffusion coefficient|
|CFV||=||cresyl fast violet|
|DWI||=||diffusion-weighted image (imaging)|
|FOV||=||field of view|
|GFAP||=||glial fibrillary acid protein|
|MCAO||=||middle cerebral artery occlusion|
|T2WI||=||T2-weighted image (imaging)|
|T2*WI||=||T2*-weighted image (imaging)|
SmithKline Beecham Pharmaceuticals contributed toward the running costs of the Herchel Smith Laboratory during this study to allow the work to be performed. We thank Dr Herchel Smith for his generous endowment of the laboratory (Prof Hall and Dr Carpenter) and studentships (N.G. Burdett and V.M. Lee), and the Overseas Research Students Award Scheme for additional financial support (V.M. Lee). We also wish to thank Dr A.R.C. Gates for helpful suggestions when initiating this work, Jeremy K. Brown for counting cells, and Emma J. Williams (Herchel Smith Laboratory) and Dr Keith Langmach (Department of Medical Physics, Addenbrooke's Hospital, Cambridge) for kindly measuring the light source irradiance. Current affiliations: Department of Physiology, University of Leeds (P.S. Pambakian), and Herchel Smith Laboratory, Cambridge (N.I. Wood), UK.
- Received November 27, 1995.
- Revision received May 30, 1996.
- Accepted June 7, 1996.
- Copyright © 1996 by American Heart Association
Burst JCM. Cerebral circulation: stroke. In: Kandel ER, Schwartz JH, Jessell TM, eds. Principles of Neural Science. New York, NY: Elsevier Scientific; 1991:1041-1049.
Jeffrey A. Numerical analysis. In: Mathematics for Engineers and Scientists. 2nd ed. Walton-on-Thames, United Kingdom: Nelson; 1979:634-673.
Horikawa Y, Naruse S, Tanaka C, Hirakawa K, Nishikawa H. Proton NMR relaxation times in ischemic brain edema. Stroke. 1986;17:1149-1152.
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.
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.
Gill R, Sibson NR, Hatfield RH, Burdett NG, Carpenter TA, Hall LD, Pickard JD. A comparison of the early development of ischemic damage following permanent middle cerebral artery occlusion in rats as assessed using magnetic resonance imaging and histology. J Cereb Blood Flow Metab. 1995;15:1-11.
van Bruggen N, Cullen BM, King MD, Doran M, Williams SR, Gadian DG, Cremer JE. T2- and diffusion-weighted magnetic resonance imaging of a focal ischemia lesion in rat brain. Stroke. 1992;23:576-582.
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.
Kharlamov A, Guidotti A, Costa E, Hayes R, Armstrong D. Semisynthetic sphingolipids prevent protein kinase C translocation and neuronal damage in the perifocal area following a photochemically induced thrombotic brain cortical lesion. J Neurosci. 1993;13:2463-2494.
Klatzo I. Brain oedema following brain ischaemia and the influence of therapy. Br J Anaesth. 1985;57:18-22.
Benham CD, Brown TH, Cooper GD, Evans ML, Harries MH, Herdon HJ, Meakin JE, Murkitt KL, Patel SR, Roberts JC, Rothal AL, Smith SJ, Wood N, Hunter AJ. SB 201823-A, a neuronal Ca2+ antagonist, is protective in two models of cerebral ischaemia. Neuropharmacology. 1993;32:1249-1257.
Markgraf CG, Kraydieh S, Prado R, Watson BD, Dietrich WD, Ginsberg MD. Comparative histopathologic consequences of photothrombotic occlusion of the distal middle cerebral artery in Sprague-Dawley and Wistar rats. Stroke. 1993;24:286-293.
Sauter A, Rudin M. Strain-dependent drug effects in rat middle cerebral artery occlusion model of stroke. J Pharmacol Exp Ther. 1995;274:1008-1013.
Hayashi N, Prado R, More J, Bunge R, Green BA. Regional changes of free radicals in photochemically induced ischemic injury in the central nervous system. Lasers Life Sci. 1991;4:153-159.
Watson BD, Ginsberg MD. Mechanisms of lipid peroxidation potentiated by ischemia in the brain. In: Halliwell B, Marshall DJ, Tacker MM, eds. Oxygen Radicals and Tissue Injury: Proceedings of a Brook Lodge Symposium, Augusta, Mich, April 27-29, 1987. Augusta, Mich: FASEB for the Upjohn Co; 1988:81-91.
Lee VM. Magnetic resonance imaging of cerebral ischaemia models. University of Cambridge, United Kingdom; 1995. Thesis.
Hoehn-Berlage M, Norris DG, Kohno K, Mies G, Leibfritz D, Hossman KA. Evolution of regional changes in apparent diffusion coefficient during focal ischemia of rat brain: the relationship of quantitative diffusion NMR imaging to reduction in cerebral blood flow and metabolic disturbances. J Cereb Blood Flow Metab. 1995;15:1002-1011.
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.
In this study Lee and coauthors have provided detailed evidence of the utility of MRI and processing techniques in following noninvasively the progression of a cerebral ischemic lesion in vivo. The method of inducing a cortical lesion by means of photochemically induced thrombosis of small cortical vessels is quite well suited for this demonstration, as the authors have explained, owing to its relative noninvasiveness, infarct reproducibility, and excellent survival rate. On the other hand, primary occlusion of small vessels is unknown in clinical stroke, but this feature ensures reproducible lesions because the density of cerebral microvasculature is constant and independent of arterial distribution.1R Yet, this is the principal reason that this approach has been perceived as artificial. The authors' goal of confirming the legitimacy of this model by correlating its physical and histopathological development with conventional models of cerebral ischemia and with clinical observations is welcomed inasmuch as this method of lesion induction has been widely used in brain research and in several other experimental contexts, including spinal cord injury, since its detailed disclosure in 1985.2R
In the present work, the authors acquired the fundamental T2-, diffusion-, and T2*-weighted MR images, but combined this information to deduce the spatial distribution of ADCs as the lesion evolved and expanded in its cortical location. Construction of a DM, ie, the distribution of ADC as seen on a coronal section, allowed more direct observation of edematous changes occurring in and around the lesion. Several striking results have emerged from this work. A precipitous fall in lesion ADC just after lesion induction agrees with MRI results from conventional animal models and assessments of clinical stroke, as the authors have stated, and implicates cytotoxic edema as the cause. This presents an interesting and complicated conundrum, which the authors have taken much care to explain, as to how the MRI representation of profuse cytotoxic edema at early times can dominate the MRI representation of vasogenic edema, a prominent and well-documented acute effect of rose bengal dye–mediated vascular photochemistry first reported in 1987.3R The lesion ADC soon begins to increase, however, with the death of cells and concomitant release of sequestered water (Fig 4B⇑), and the subsequent growth of the infarct can be related to the increase in lesion ADC (Fig 4C⇑), which is in turn inversely related to the density of viable cells (Fig 4D⇑). The authors have shown also (Fig 5⇑) that MRI can be correlated acutely with the histological development of the lesion.
Another important consideration that the authors addressed is whether a penumbra accompanies a cortical photochemical lesion. The authors have previously demonstrated an affirmative answer4R by showing that this lesion can indeed be treated by drugs which interrupt commonly accepted pathways of neuronal degeneration in stroke and which in fact mitigate the extent of lesion expansion.4R 5R In the present work, the authors maintain that the penumbra in this model is visible on a DM at 12 to 24 hours as a region of low ADC values surrounding the lesion core, distinguished by high ADC values. These observations support the suggestion that the penumbra should be regarded as that tissue region which is susceptible by proximity to ischemic damage but which is amenable to treatment.6R With regard to acute treatment of the core itself, this is usually considered fruitless in view of the extensively thrombosed vasculature therein. However, partial reperfusion of the core via apparent thrombolysis has been reported,7R but in general this is extremely difficult if the platelet-rich, fibrin-poor thrombi have been formed in response to endothelial damage mediated by rose bengal photochemistry.8R
A possible factor contributing to the authors' detection of a penumbra is their use of a minimal beam intensity to obtain a lesion. With a minimum but effective intensity, the rate of lesion expansion due to vasogenic edema may be slowed to the point at which the penumbra can be formed and persist long enough to be detected, as the authors have done. A caveat is in order, however. In general, use of an optically unfiltered white-light source such as they have employed is not to be encouraged, because irradiation of cortical tissue at wavelengths not absorbed by the photosensitive dye may deposit heat into the tissue, thereby unnecessarily superimposing the tissue-damaging effects of temperature elevation onto those wrought by ischemia.9R However, the intensity of either of their irradiating beams incident on the cortical surface can be calculated to be only 83 mW/cm2 in the 400- to 800-nm wavelength range (accounting for a beam loss of 50% through the skull in conjunction with the optical properties and position of the fiber). This intensity will not noticeably heat the cortical tissue, as the authors have apparently verified (although optimally the temperature probe should be placed beneath the skull surface), and it is much less than that originally used,2R which necessitated forced air cooling of the skull surface.
Salvage of cortical tissue by means of drugs introduced via patent collateral channels is the scenario most commonly envisioned clinically, and development and characterization of an animal model in which such effects can be reproducibly studied is the necessary analogue desired experimentally. A possible alternative candidate in the photochemical context is the laser-induced “ring” model of cortical infarction.10R This method produces a cortical lesion consisting of a ring-shaped locus of occluded microvasculature that is nonetheless penetrated by patent distal branches of main feeder arteries. In the absence of intervention, the cortical tissue enveloped by the ring is destined to die within a time dependent on the irradiation parameters. The central region thus likely experiences a progression of toxic changes in common with a conventional penumbra, except for the unusual feature that the compromised tissue is inside the initially ischemic zone, therefore accounting for its reproducible volume.
Selected Abbreviations and Acronyms
|ADC||=||apparent diffusion coefficient|
|CFV||=||cresyl fast violet|
|DWI||=||diffusion-weighted image (imaging)|
|FOV||=||field of view|
|GFAP||=||glial fibrillary acid protein|
|MCAO||=||middle cerebral artery occlusion|
|T2WI||=||T2-weighted image (imaging)|
|T2*WI||=||T2*-weighted image (imaging)|
Watson BD, Dietrich WD, Prado R, Nakayama H, Kanemitsu H, Futrell NN, Yao H, Markgraf CG, Wester P. Concepts and techniques of experimental stroke induced by cerebrovascular photothrombosis. In: Ohnishi ST, Ohnishi T, eds. Central Nervous System Trauma: Research Techniques. Boca Raton, Fla: CRC Press; 1995:169-194.
Watson BD, Dietrich WD, Busto R, Wachtel MS, Ginsberg MD. Induction of reproducible brain infarction by photochemically initiated thrombosis. Ann Neurol.. 1985;17:497-504.
Dietrich WD, Watson BD, Busto R, Bethea JR, Scheinberg P, Ginsberg MD. Photochemically induced cerebral infarction, 2: edema and blood-brain barrier disruption. Acta Neuropathol (Berl). 1987;72:326-334.
Benham CD, Brown TH, Cooper GD, Evans ML, Harries MH, Herdon HJ, Meakin JE, Murkitt KL, Patel SR, Roberts JC, Rothau AL, Smith SJ, Wood N, Hunter AJ. SB 201823-A, a neuronal Ca2+ antagonist, is neuroprotective in two models of cerebral ischaemia. Neuropharmacology. 1993;32:1249-1257.
De Ryck M. Animal models of cerebral stroke: pharmacological protection of function. Eur Neurol. 1990;30(suppl 2):21-27.
Memezawa H, Smith M-L, Siesjo BK. Penumbral tissues salvaged by reperfusion following middle cerebral artery occlusion in rats. Stroke.. 1992;23:552-559.
Watson BD, Dietrich WD, Busto R, Ginsberg MD, Hoyng CF. Influence of tissue-type plasminogen activator on blood flow during evolution of thrombotic stroke. Soc Neurosci Abstr. 1985;11(part I):433.
Watson BD, Prado R, Dietrich WD, Busto R, Scheinberg P, Ginsberg MD. Mitigation of evolving cortical infarction in rats by recombinant tissue plasminogen activator following photochemically induced thrombosis. In: Raichle ME, Powers WJ, eds. Proceedings of the Fifteenth Princeton Conference on Cerebrovascular Diseases. New York, NY: Raven Press Publishers; 1987:317-330.
Wester P, Watson BD, Prado R, Dietrich WD. A photothrombotic ‘ring’ model of rat stroke-in-evolution displaying putative penumbral inversion. Stroke. 1995;26:444-450.