Tobias Back, Weizhao Zhao, and Myron D. Ginsberg. (1890). With the advent of autoradiographic strategies for measuring local CBF (LCBF) (Reivich et

Transcription

1 Journal of Cerebral Blood Flow and Metabolism 15: The International Society of Cerebral Blood Flow and Metabolism Published by Raven Press, Ltd., New York Three-Dimensional Image Analysis of Brain Glucose Metabolism-Blood Flow Uncoupling and Its Electrophysiological Correlates in the Acute Ischemic Penumbra Following Middle Cerebral Artery Occlusion Tobias Back, Weizhao Zhao, and Myron D. Ginsberg Cerebral Vascular Disease Research Center, Department of Neurology, University of Miami School of Medicine, Miami, Florida, U.S.A. Summary: The relationship between local cerebral glucose utilization (LCMRglc) and local CBF (LCBF) is known to be disturbed in regions surrounding an acute focal ischemic lesion-areas that undergo repeated transient depolarizations. In this study, we evaluated the relationship between LCMRglc and LCBF in the acute focal ischemic penumbra to quantify metabolism-flow uncoupling, and we related these findings to local electrophysiological measurements. A novel strategy utilizing threedimensional (3D) autoradiographic image averaging yielded group 3D reconstructions of LCBF, LCMRg 1C' and the CMRlCBF ratio. The distal right middle cerebral artery of Sprague-Dawley rats was occluded by laserdriven photothrombosis following administration of the photosensitizing dye rose bengal; this was coupled with permanent ipsilateral and I-h contralateral common carotid artery occlusions. LCBF (n = 7) and LCMRglc (n = 7) were measured autoradiographically at 1.25 and h postocclusion, respectively, in matched animal groups. Within the ischemic penumbra (defined as having LCBF of 20-40% of control or ml g- J min - J), LCM Rglc showed a heterogeneous pattern with values ranging from near normal to markedly increased. The resulting CMRglc/CBF ratio in this zone was 234 ± 100 fl-mouloo ml (mean ± SD), representing a severe degree of metabolism-flow dissociation when compared with the CMRglcl CBF ratio of 51.0 ± 28.7 fl-molll00 ml of the contralateral (normal) hemisphere. Metabolism-flow uncoupling was confined to the ipsilateral cortex and was most prominent at the anterior and posterior coronal poles of the ischemic lesion. In the frontoparietal penumbra, where marked uncoupling was observed, sustained deflections of the DC potential were recorded, which increased significantly in duration over the initial 65 min postocclusion. Both the heterogeneous pattern of LCMR Ie and the widespread distribution of increased CMRgJCBF ratio in the ischemic penumbra are thought to reflect the metabolic consequences of periinfarct depolarizations. Analysis of averaged 3D autoradiographic data sets provides a powerful means for assessing metabolism-flow uncoupling surrounding an ischemic focus. Key Words: Autoradiography-Direct current potential-electrocorticogram Focal ischemia-image analysis-local cerebral blood flow-local cerebral glucose utilization. Coupling of cerebral metabolism and blood supply was first described by Roy and Sherrington (1890). With the advent of autoradiographic strategies for measuring local CBF (LCBF) (Reivich et Received September 2, 1994; final revision received November 4, 1994; accepted November 30, Address correspondence and reprint requests to Dr. M. D. Ginsberg at Department of Neurology (04-5), University of Miami School of Medicine, P.O. Box , Miami, FL 33101, U.S.A. Abbreviations us ed: CCA, common carotid artery; 3D, threedimensional; 2-DG, [J4Cj2-deoxY-D-glucose; LCBF, local CBF; LCMRglc, local CMRgJc; MCA, middle cerebral artery; ROI, region of interest; SO, spreading depression. ai., 1969; Sakurada et ai., 1978) and local cerebral glucose utilization (LCMRglc) by means of [14C]2_ deoxy-o-glucose (2-DG) (Sokoloff et ai., 1977), it was demonstrated that tight regional coupling of glucose metabolism and blood flow existed in various brain structures under normal conditions (Reivich, 1974; Ginsberg et ai., 1986) and in a variety of states including functional activation (Sokoloff, 1981; Ginsberg et ai., 1987), electrical brain stimulation (Iadecola et al., 1983), and metabolic acidosis (Kuschinsky et ai., 1981). In cerebral ischemia, however, this couple is severely disturbed. Pulsinelli et ai. (1982) reported a 566

2 METABOLISM-FLOW UNCOUPLING IN FOCAL ISCHEMIA 567 proportionately greater depression of postischemic CBF in comparison to glucose utilization in a model of transient forebrain ischemia. Following reversible hemispheral forebrain ischemia, Ginsberg et al. (1985) observed nearly normal LCMR g lc in the presence of depressed flow during the early postischemic period, which coincided spatially with the predominant locus of neuronal injury-the striatum. In experimental models of focal ischemia, a consistent pattern of LCMR g lc has been described in the acute stage: namely, a core area of greatly depressed LCMR g lc surrounded by a zone having increased deoxyglucose uptake (Ginsberg et ai., 1977; Welsh et ai., 1980; Hossmann et ai., 1985; Nedergaard et ai., 1986). Following transient occlusion of the middle cerebral artery (MCA), regions of marked metabolic depression exhibited a very severe reduction in blood flow (Tanaka et ai., 1985). In the peripheral MCA territory, regions with no histological damage showed parallel decreases in LCBF and LCMR g lc, indicating a preserved metabolism-flow couple, whereas regions having a similar degree of flow reduction but increased 2-DG accumulation developed slight ischemic damage (Tanaka et ai., 1985). Thus, metabolism-flow uncoupling in the ischemic borderzone is regarded as a factor contributing to infarct development. Nonetheless, the mechanisms underlying this uncoupling process are incompletely understood. In the penumbral region of acute MCA occlusion in the cat, Strong et a\. (1983) described transient elevations of pial surface potassium activity that resembled the synchronized periinfarct cell depolarization demonstrated by Nedergaard and Astrup (1986). These spontaneously occurring depolarizations bear an electrophysiological resemblance to spreading depression (SD) of cortical electrical activity originally described by Leao (1944). It is known that SD is associated with increased glucose consumption (Shinohara et ai., 1979), probably due to an increased energy demand of A TP-dependent ion pumps. In the energetically compromised penumbra, the SD-associated blood flow response (transient hyperperfusion) is abolished and repolarization delayed (Back et ai., 1994a). It is likely that the metabolic activation associated with transient depolarization is at least partially responsible for the observed increases in penumbral 2-DG uptake. As SD has been shown to lead to sustained metabolism-blood flow uncoupling in the intact brain (Lauritzen and Diemer, 1986), it is conceivable that periinfarct depolarizations would accentuate that process in the moderately ischemic penumbral (or borderzone) region of an evolving focal infarct. In the present study, we sought to quantify the ratio between LCBF and LCMR g lc in the acute period following MCA occlusion and to relate these findings to electrophysiological measurements in the infarct borderzone. From the studies cited, it appears that analysis of blood flow and glucose utilization, when carried out in individual animals and limited to two dimensions, may be limited in its ability to depict the full spatial extent and regional pattern of uncoupling in the infarct periphery. In the present article, we demonstrate ischemia-related metabolism-flow uncoupling in its three-dimensional (3D) representation by using a newly developed method of computer-based image analysis (Zhao et ai., 1993, 1994). Importantly, this method permits averaged 3D image data sets derived from replicate animal studies to be constructed for CMR g lc and CBF, and for a ratio data set of the CMR g lc/cbf couple to be generated (Zhao et ai., 1994). It is evident that these studies deal with a complex event requiring high spatial resolution and precision in the measurement of CMR g lc and CBF. We have chosen to avoid a double-tracer study because the precision of the CBF measurement is compromised and the double-study method necessitates that LCBF be measured during the final moment of the LCMR g lc study rather than at its epicenter (Ginsberg et a\., 1986). LCMR g lc and LCBF were measured in two separate groups of animals, at and at 1.5 h, respectively, after onset of MCA occlusion. A novel stroke model of distal MCA photothrombosis coupled with common carotid artery (CCA) occlusions was applied, which has been shown to produce a highly consistent pattern of infarction (Markgraf et ai., 1993; Yao et ai., 1993). Preliminary results of this investigation have been reported in abstract form (Back et ai., 1994c). MATERIALS AND METHODS Studies were carried out in 26 male Sprague-Dawley rats weighing g. All animals underwent distal MCA photothrombosis coupled with 1-h contralateral and permanent ipsilateral CCA occlusions. The rats were assigned in a randomized manner to experiments employing the measurement of either LCBF or LCMRglc centered at 1.5 h following MCA occlusion (n = 7 each). In a third animal group, EEG and DC potential were recorded in the infarct borderzone, in which CBF was moderately reduced (n = 8). In sham-operated animals (n = 4), electrophysiology was measured. One day later, sham animals underwent perfusion-fixation, and brains were paraffin embedded, stained with hematoylin-eosin, and examined histologically to confirm an absence of pathological changes. Surgical preparation Before surgery, rats were food-deprived for 12 h. Anesthesia was induced with 3% halothane and continued J Cereb Blood Flow Metab, Vol. 15, No.4, 1995

3 568 T. BACK ET AL. via a mask with 1.5% in a 7:3 mixture of nitrous oxide and oxygen. Both femoral arteries and one vein were cannulated with PE-50 tubing to record blood pressure continuously, to obtain arterial samples, and to administer drugs and isotopes. The rats were endotracheally intubated, immobilized with pancuronium bromide (0.3 mg/kg i. v.), and artificially ventilated at a rate of 50 breaths/min. The halothane concentration was reduced to 0.75%. Rectal and ipsilateral temporal muscle temperatures were both measured and were maintained at 36.9 ± 0.3 and 36.5 ± O.soC, respectively (means ± SD), using two feedback-controlled heating lamps placed over the animal's body and head. The surgical procedure was performed as described previously (Markgraf et a!., 1993; Yao et a!., 1993). Both CCAs were exposed by a ventral midline incision in the neck, dissected free, and surrounded by a loop of PE-lO tubing contained within 20-mm segments of bilumen Silastic tubing. After the rats were mounted in a stereotaxic headholder, a 1.5-cm vertical incision was made between the right eye and ear. The zygomatic and squamosal bones were exposed by separation and retraction of the temoralis muscle. With an operating microscope (model 50 T; Carl Zeiss, Thornwood, NY, U.S.A.), a burr hole -3 mm in diameter was made with a high-speed drill rostral to the anterior junction of the zygomatic and squamosal bones, leaving the dura intact. The distal segment of the right MCA was exposed above the rhinal fissure. Photochemically induced distal MCA thrombosis Three points of laser irradiation were directed onto the exposed MCA and were positioned predominantly at distal branching points according to the criteria described by Markgraf et a!. (1993). Irradiation was performed with an argon laser-activated dye laser (Coherent, Palo Alto, CA, U.S.A.) tuned to 562 nm at a power of 20 mw. The single laser beam was split into three beams by a Ronchi ruling and focused by a spherical lens of 25-cm focal length. The separated beams could be positioned individually by a mirror onto a designated point of the distal MCA. The right side of the stereotaxic frame was tilted slightly upward so that the laser would strike the vessel perpendicularly. The photosensitizing dye rose bengal (15 mg/ml, 1.34 mukg body wt) was administered intravenously over 1.5 min. Subsequently, the three points of the MCA were simultaneously irradiated for min until complete MCA occlusion was verified by microscopic observation. Five minutes after onset of irradiation, the snare ligatures around both CCAs were tightened for bilateral occlusion to restrict the extent of collateral blood flow to the ischemic MCA territory. The right CCA remained occluded, whereas the contralateral CCA occlusion was released after 1 h. Reperfusion was verified microscopically. Autoradiographic measurement of LCBF and glucose utilization To measure LCBF, 14C-labeled iodoantipyrine was injected 1.5 h after induction of focal ischemia. Twenty micro curies of the radioactive tracer (specific activity 40 /LCi/mmol; New England Nuclear) dissolved in 1 ml of isotonic saline was infused intravenously at a constant rate over 45 s via a Harvard infusion pump. Arterial blood was sampled at -2-s intervals from a freely flowing femoral arterial line. In the second animal group, LCMRglc was measured with 2-DG. An intravenous bolus of 20 /LCi 2-DG (specific activity /LCilmmol; New England Nuclear) dissolved in 0.1 ml saline was administered at 1 h 15 min post-mca occlusion. Arterial blood samples were drawn every 10 s during the first minute, at I-min intervals x 5 min, and at 2.5- to 5-min intervals for the remainder of the 45-min study period (the centroid of the 2-DG uptake period thus being close to the time point of the CBF study). Plasma samples were assayed for 2-DG by liquid scintillation counting and for plasma glucose by a glucose analyzer (YSI, Yellow Springs, OH, U.S.A.). Both autoradiographic studies were terminated by decapitation; brains were removed in <1.5 min and frozen in liquid nitrogen. In a cryostat at - 20 C, the brains were carefully embedded and serially cut into 20-/Lm-thick coronal sections at 100-/Lm intervals as described previously (Zhao et a!., 1994). Together with calibrated e4c]methylmethacrylate standards, the sections were exposed to Kodak Hyperfilm I3max film for 10 days. By means of a densitometric scanning system (Optronics Colors can C4lO0) interfaced to an image processor (Grinnell Systems) and a clustered MicroV AX computer system (Digital Equipment Corp.), the autoradiographic films were converted to digital images of optical density. The operational equations of the iodoantipyrine method (Sakurada et a!., 1978) and the 2-DG method modified for variable plasma glucose (Sokoloff et al., 1977; Savaki et al., 1980) were used to compute LCBF and LCMRg\c, respectively. Electrophysiological measurements In these rats (n = 8), in addition to the surgical procedure described, a right-sided trepanation of the frontoparietal skull was performed to place a miniature calomel electrode on the intact dura. The position, 2 mm lateral and mm anterior to bregma, was chosen to measure electrophysiology in a region of moderate flow reduction as shown in the LCBF study (20-50% of control). The reference electrode was placed on the contralateral skull. Connection between tissue and electrode was established by a saline-rinsed cotton wick. The signal was transduced to two differential amplifiers (BMA-931; CWE, Ardmore, PA, U.S.A.) and displayed as EEG and DC potential on a chart recorder (TA5000; Gould, Valley View, OH, U.S.A.). Both EEG and DC potential were measured before, during, and for 3 h after vascular occlusion. Computation of average 3D autoradiographic image data sets The reconstruction of 3D autoradiographic images was performed on the V AX computer system. The procedure is based on a novel image alignment algorithm whose theory and practical validation have been described previously in detail (Zhao et a!., 1993, 1994). A linear affine transformation was used to register sequential coronal sections of each brain, and transformation parameters were calculated by point-to-point disparity analysis. Decomposition of the transformation parameters revealed translation, rotation, scaling, and shearing factors between coronal sections (Zhao et a!., 1993, 1994). The alignment of each coronal section to its adjacent neighbor was performed by applying a reverse transformation with the estimated translation and rotation parameters. After alignment of individual LCBF or LCMRgl C images, corresponding coronal sections from all brains of the same experimental group were placed in register with one another using a common coronal reference level (bregma + 0.7) (Paxinos and Watson, 1982). One brain of each series served as a template, into whose contours J Cereb Blood Flow Me/ab, Vol. 15, No.4, 1995

4 METABOLISM-FLOW UNCOUPLING IN FOCAL ISCHEMIA 569 corresponding sections of the other brains were mapped at each coronal level, by means of an averaging procedure similar to that employed for alignment. In this manner, 3D reconstructions of quantitative LCBF and LCMRg1c average data sets were calculated and could be displayed on a video screen. Similarly, 3D image data sets representing standard deviation of the CBF and CMRglc images were also obtained. The CMRglc/CBF ratio image was calculated by applying the following equation: where Rk denotes the mean ratio value at pixel k, Pik is the LCMRglc reading at the kth pixel from the ith animal in the LCMRglc study group with a total of N (N = 7) animals, and Q j k is the LCBF reading at the kth pixel, geometrically corresponding to that of LCMRglc reading, from the jth animal in the LCBF study group with a total of M (M = 7) animals. The program used for image analysis allowed us to display every single coronal section of each 3D map separately, both on the average images and on the individual images from which they were derived. By means of a drawing tool, regions of interest (ROIs) could be delineated to measure the mean and standard deviation values of the respective variables (LCBF, LCMRglc). A thresholding tool allowed us to produce selective 3D reconstructions depicting only those pixels lying within predefined thresholds for LCMRglc, LCBF, and the CMRglcl CBF ratio. In addition, the volume of such 3D threshold images could be computed. To exclude from analysis those LCMRglc values lying within severely ischemic tissue regions (in which the quantitative 2-DG model cannot be rigorously applied), we elected to study specifically those pixels of the 3D image data sets having a range of moderately reduced blood flow (defined as 20-40% of contralateral values) to permit the study of the phenomenon of metabolism-flow uncoupling in the infarct borderzone. In light of the findings of Nedergaard et al. (1986) and Nakai et al. (1987), as well as our own observations using [14C]methylglucose at 1.5 h after photothrombotic MCA occlusion in this model (Yao et ai., 1995), brain glucose content, and hence the lumped constant of the 2-DG model, can be assumed to be virtually unchanged in borderzone regions (i.e., CBF?20% of control). Data analysis and statistics Values are expressed as means ± SD. Statistical differences between experimental groups were assessed by analysis of variance, including repeated-measures analysis of variance followed by Tukey's honestly significant (1) difference test. For ROI analysis, the z-test for means with known variance was applied to test for significant differences between the two hemispheres. Statistical significance was accepted at the p < 0.05 level unless otherwise stated. RESULTS Physiological variables As shown in Table I, the physiological variables were kept within the normal range throughout the experiments. No significant differences were detected between experimental groups or during the course of the experiment (the latter data not shown). 3D reconstruction of LCBF Mean CBF of the contralateral nonischemic hemisphere, as evaluated in the averaged 3D data set, was 1.17 ± 0.48 ml g - I min - I. This value corresponded closely to resting CBF values determined in our laboratory in intact animals under identical experimental conditions (1.20 ± 0.24 ml g- I min- I ) (T. Back et ai., unpublished) and was taken to represent 100% of control. In this hemisphere, no focal areas of reduced flow were detected. The pattern of flow reduction in the ischemic right hemisphere was characterized by a marked flow gradient from cingulate to frontolateral cortex, with LCBF ranging from 0.01 to 0.80 ml g- I min - I (Figs. 1-3A; Table 2). The averaged 3D LCBF map revealed the most severe flow reduction to be located in the lateral portion of the frontoparietal cortex, extending posteriorly beyond the level of hippocampus (Figs. 2 and 3A). More moderate flow reductions covered the remainder of the distal MCA territory but also spread, to a lesser degree, to adjacent parts of neighboring vascular territories (medial striatum, cingulate cortex, basal temporal cortex). Following the criterion of Tyson et al. (1984), we chose 20% of control level as the CBF threshold for ischemic injury in normotensive Sprague-Dawley rats to define the ischemic core. In the ischemic hemisphere, a volume of 117 ± 55 mm 3 exhibited flow values below 0.23 ml g- I min- I (20% of control). This volume corresponds closely to infarct volumes previously reported in this model (Markgraf et ai., 1993). Borderzone flow values TABLE 1. Physiological variables ph Pe02 P02 Glucose Rectal temp. Head temp. MABP Exp. group (units) (mm Hg) (mm Hg) (mg/dl) CC) ec) (mm Hg) CBF (n = 7) 7.40 ± ± ± ± ± ± ± 14 CMRglc (n = 7) 7.40 ± ± ± ± ± ± ± 14 DC pot. (n = 8) 7.42 ± ± ± ± ± ± ± 20 Mean values ± SD. CBF, CMRglc, DC pot.: groups studied for local CBF, CMRglc, or direct current potential, respectively. J Cereb Blood Flow Metab, Vol. 15, No. 4, 1995

5 570 T. BACK ET AL. Level A (at bregma) left right left Level B (2 mm posterior to bregma) right FIG. 1. Topography of regions of interest (ROls) selected for analysis of local CBF (LCBF), local CMR g lc (LCMR g 1c) ' and the CMR g 1JCBF ratio (data shown in Table 2). The two representative coronal levels A and B are displayed as anatomic templates with numbered ROls. Level A is located in the infarct center and corresponds to three-dimensional image section no. 30. Level B is -2 mm more posterior (section no. 47 of the three-dimensional reconstruction). The arrow in template A marks the site of electrophysiological recording. were thus defined as lying between 20 and 40% of control ( ml g- I min - I ) and were present in an ipsilateral volume of 136 ± 30 mm 3 (see Figs. 2 and 4). 3D reconstruction of cerebral glucose utilization Mean CMR g Jc in the left hemisphere was 39.6 ± 12.5 flmol 100 g- I min- I. In the ischemic right hemisphere, metabolic suppression below 50% of the contralateral value was observed in a volume of 85 ± 43 mm 3 Pixels with those values were located within the ischemic core of the infarct (Fig. 2). Within the infarct borderzone (defined, as previously, by the CBF criterion of ml g- I min -1), glucose utilization was remarkably preserved (Table 2; Figs. 3A and 5). Individual LCMR g Jc maps revealed focal hypermetabolism in four of seven animals, in which CMR g lc increased to a maximum of 160 flmol 100 g- I min - I (Fig. 5). Due to interanimal variation in the topography and extent of LCMR g Jc in the ischemic borderzone, however, marked hypermetabolism was not equally reflected in the averaged LCMR g lc image. The relationship between hemispheric brain volumes affected by CBF reduction and metabolic suppression is shown in Fig. 4. For comparable degrees of reduction, significantly larger volumes of the ischemic hemisphere were affected by reduction of LCBF than by depression of LCMR g lc' This disparity increased markedly at higher flow ranges (Fig. 4). 3D reconstruction of cerebral glucose metabolismlblood flow ratio The mean CMR g lc/cbf ratio in the left nonischemic, hemisphere, as determined in the averaged ratio image, was 51.0 ± 28.7 flmol/loo ml. All ratio (A) (8) FIG. 2. Three-dimensional reconstructions of averaged image data set for local CBF (n = 7; A) and local CMR g lc (n = 7; 8). Each three-dimensional stack is based upon coronal sections taken every 100 J..Lm and separated in the image by 200 J..Lm. The lower, middle. and upper arrows in the left panel denote coronal levels of + 2, -1, and - 4 mm with respect to bregma, respectively. J Cereb Blood Flow Metab, Vol. /5, No.4, /995

6 METABOLISM-FLOW UNCOUPLING IN FOCAL ISCHEMIA 571 (A) FIG. 3. A: Local CBF (LCBF), local CMR g,c (LCMR g,c), and CMR g 'c/cbf ratio images derived from three-dimensional reconstruction of averaged data sets. Quantitative data are displayed only for the image pixel subset having LCBF values of 20-40% of control (Le., ml g -1 min -1 ), corresponding to the ischemic borderzone. The upper pseudocolor calibration bars apply to LCBF (left), LCMR g,c (middle), and the CMR g 1c/CBF ratio (right), respectively. Top image row: Averaged LCBF images, showing the ischemic borderzone of the right hemisphere at five coronal planes (levels in mm are bregma +2.0, +0.7, -1.1, -4.0, and -7.3) (Paxinos and Watson, 1982). [White denotes pixels having LCBF >40% of control, and black denotes pixels in which LCBF is in the ischemic core range «20% of control).] Middle image row: Averaged LCMR g,c images of the ischemic borderzone (same levels as the top row). Third image row: Averaged CMR g 'c/cbf ratio images of ischemic borderzone at the same coronal levels. B: Three-dimensional reconstruction of CMR g 1c/CBF ratio for those pixels having LCBF of 20-40% of control (see text). This reconstruction is based upon coronal sections taken every 100 IJ-m and separated in the image by 200 IJ-m. [Note that subcortical ratio pixels are visible throughout the ischemic core ("see-through" effect) at sites where the surface pixels have zero values owing to the abovementioned LCBF threshold criterion.] (B) J Cereb Blood Flow Metab. Vol. 15. No

7 572 T. BACK ET AL. TABLE 2. Region-oj-interest (ROJ) analysis oj local CBF (LCBF), glucose utilization (LCMRg[c), and CMRglclCBF ratio Jollowing right middle cerebral artery occlusion LCBF LCMRglc LCMRglc/LCBF ratio (ml g-i min-i) ().Lmol 100 g-i min-i) ().LmoI!100 ml) ROI no. Left Right Left Right Left Right Level A ± ± ± ± ± ± 48b ± ± 0.14" 39 ± ± ± ± 202b ± ± 0.07" 40 ± ± ± ± 304b ± ± 0.2Sb 40 ± ± ± ± ± ± ± ± ± ± ± ± 0.19h 58 ± ± ± 9 82 ± 16b ± ± 0.18b 41 ± ± ± ± 25 Level B ± ± ± ± ± ± ± ± O.13b 43 ± ± ± ± 248b ± ± 0.10" 45 ± ± ± ± 20 lb ± ± ± ± ± ± 52b ± ± ± II 25 ± 7 49 ± ± 42 I ± ± ± ± I3 78 ± ± 39 Values are given as means ± SD. ROJs were analyzed in the averaged three-dimensional (3D) data sets of the variables. SD for LCBF and LCMRglc was obtained from the respective 3D data sets and reflects interanimal variation. SD of CMRglc/CBF ratio values represents interpixel variation inside the chosen ROJ of the averaged data set. For ROJ no., see templates in Fig. I. "p < 0.01, bp < 0.05, significantly different from contralateral side (z-test for means). values exceeding the (upper) 95% confidence limit of this mean ratio, i.e., 108 flmollloo ml, were considered to be significantly increased. The nonischemic hemisphere showed no focal areas with significantly increased CMR g lc/cbf ratio. In the affected hemisphere, significant uncoupling was confined to the cortex. Within the infarct borderzone (defined as LCBF 20-40% of control), the mean CMR g lcl CBF ratio increased nearly fivefold to 234 ± 100 flmoll100 ml, ranging between 108 and 490 flmoll 100 ml (Fig. 3). These findings indicate that, at 1.5 h after MCA occlusion, CMR g lc was well maintained in the infarct border despite CBF values close to the threshold of ischemic injury. At flow values above 0.30 ml g - 1 min - 1, varying degrees of glucose hypermetabolism (ranging up to a fourfold elevation) accounted for the considerable increase of the CMR g lc/cbf ratio (Figs. 3 and 5). 450 * p < * 400 * 350 * i * CBF CMRgl FIG. 4. Relationship between reduction of blood flow or metabolism and hemispheric volumes having local CBF (LCBF) or local CMR g lc (LCMR g lc) values below certain thresholds of the respective variable. Threshold volumes were determined in individual animals of groups studied for either LCBF or LCMR g lc' Interanimal variation is displayed as SO error bars at every level of CBF or CMR g lc reduction (ranging from 5 to 50% of control). Means were tested for statistically significant intergroup differences by analysis of variance followed by Tukey ' s test, which corrects for multiple comparisons. CBF threshold volumes that were significantly greater than the corresponding CMR g lc threshold volumes are indicated by asterisks. Note the increasing discrepancy between CBF and CMR g lc volumes, especially at flow values higher than 20% of control.. ol- == -- o CBF. CMRgl ra. of control] J Cereb Blood Flow Metab, Vol. 15, No.4, 1995

8 METABOLISM-FLOW UNCOUPLING IN FOCAL ISCHEMIA 573 FIG. 5. Coronal autoradiographic sections of local CMR g lc (LCMR g 1c) at the level of midstriatum. The left (arrow) and middle panels of the top row show the averaged and SD images, respectively, derived from image averaging of seven replicate studies. The remaining panels show LCMR g lc autoradiograms in the seven individual brains from which the averaged image data set was derived. Heterogeneously configured zones of hypermetabolism are evident in the individual autoradiograms. EEG and DC potential measurements In eight experiments, a total of 46 negative DC deflections (4-9/animal) was recorded in the frontoparietal borderzone of the infarcts, where marked uncoupling was observed [cf. Figs. 1 (template A) and 3A]. Figure 6 shows a typical recording with repeated waves of periinfarct depolarization. Induction of ischemia and onset of DC negativation were accompanied by a reduction in EEG amplitude. The amplitude of the epidurally measured DC transients was 7.3 ± 3.6 my. In seven of eight rats, the initial depolarization occurred within 5 min following MCA irradiation (i.e., before bilateral CCA occlusion) with a latency of 2.1 ± 0.5 min. In eight animals, a total of 35 DC shifts was detected during the first 65 min and showed a significant increase in duration over this period (Fig. 6; Table 3). The correlation between the duration of DC shifts and their FIG. 6. Original recordings of EEG and DC potential measured in the ischemic penumbra of a rat undergoing photothrombotic middle cerebral artery (MCA) occlusion. The arrow on the left indicates the time of rose bengal injection. Right MCA occlusion (MCAO) was achieved by laser irradiation at three distal sites of the vessel. Both common carotid arteries were subsequently ligated (CCAO). One hour later, the contralateral (left) CCA was released. Note the increasing duration of depolarizations during the initial 30 min postocclusion. DC Repetitive ciepolu'izatiolll in the infarct bonier.i I '.1,l1li1;, ,... II.'1... ]loo.v _CCAO 1/ --A- IofICCA'-,.. J 'OmV J Cereb Blood Flow Metab, Vol. 15, No.4, 1995

9 574 T. BACK ET AL. TABLE 3. Number and duration of periinfarct DC shifts following photothrombotic MCA occlusion coupled with J-h bilateral and permanent ipsilateral eca occlusion Occurrence Sequence No. of rats (min Duration of DC shifts (DC shifts) postocclusion) (min) 0-65 min post-mcao 1st ± ± 0.5 2nd ± ± 3.0 3rd ± ± 2.5 4th ± ± 4.7a 5th ± ± 4.9b min post-mcao Subsequent 5 (JOC) III ± ± 2.8 MCAO, middle cerebral artery occlusion; CCA, common carotid artery. a p < 0.05, bp < 0.01, significantly different from duration of initial depolarization (analysis of variance for repeated measures, Tukey honestly significant difference test for unequal sample sizes). C Ten DC shifts were measured in five animals; one anoxic depolarization was not included. time point of occurrence (n = 35, measured 0-65 min post-mca irradiation) was highly significant (r = 0.58, p < 0.001) (Fig. 7). After release of the contralateral CCA, the frequency of negative DC transients was diminished from 4.0 to 0. 8 DC shifts/h for the remaining period of observation. The duration of DC shifts from 65 min postocclusion onward was indistinguishable from the shortlasting initial depolarization. In one animal, a transition from repeated DC transients into terminal anoxic depolarization was observed 95 min following MCA occlusion. Sham-operated animals displayed normal electrophysiology; brain histology 1 day after operation was also normal. DISCUSSION To the best of our knowledge, the present investigation is the first to utilize 3D reconstruction of averaged autoradiographic images to demonstrate acute focal ischemia-related metabolism-flow uncoupling. In this model of distal MCA occlusion, giving rise to widespread cortical ischemia, we were able to demonstrate marked metabolism-flow uncoupling in the cortical borderzone of the ipsilateral hemisphere-a region that underwent repeated sustained cell depolarizations. The morphological consequences of this model of distal MCA photothrombosis coupled with I-h contralateral and permanent ipsilateral CCA occlusion have been under extensive investigation in our laboratory (Markgraf et ai., 1993; Yao et ai., 1993). A comparison between the CBF data of the present study and the infarct pattern previously observed in this model (Markgraf et ai., 1993) reveals a spatial coincidence between the volume of the low-flow region «0.23 ml g - 1 min -1) of the present study (117 ± 55 mm 3 ) and the volume of overt infarction 3 days after the induction of stroke in the previous study (131 ± 40 mm 3 ) (Markgraf et ai., 1993). The slight difference in volumes may be explained by the fact that the threshold of ischemic injury increases with increasing duration of ischemia (Heiss and Rosner, 1983). We defined the ischemic borderzone as having flows 20 Y r P < _ 12 co. co 10.2! '" c FIG. 7. Correlation between time point of occurrence and duration of 35 DC potential shifts. The DC transients were measured in eight animals during the initial 65 min following middle cerebral artery (MCA) laser irradiation. Distal MCA occlusion was achieved by laser-driven photothrombosis in conjunction with rose bengal administration. It was coupled with permanent ipsilateral and 1-h contralateral common carotid artery occlusions Time of QCCUnence [min post MCA Irradlatlonl J Cereb Blood Flow Metab, Vol. 15, No.4, 1995

10 METABOLISM-FLOW UNCOUPLING IN FOCAL ISCHEMIA 575 between 0.23 and 0.47 ml g-1 min -1 in accordance with the findings of Tyson et al. (1984), who observed in MCA occlusion in the same rat strain that LCBF in regions with histological damage was always <0.25 ml g-1 min-i, whereas in regions with flow values between 0.25 and 0.35 ml g -1 min -1, no consistent correlation between LCBF and histological changes was evident. These threshold values are considerably lower than the ones reported by Jacewicz et al. (1992), who studied spontaneously hypertensive rats-a strain known to be more susceptible to large infarcts than normotensive strains. In the present stroke model, the severity of collateral flow impairment is reflected in the large volume of tissue (136 mm3) having LCBF values between 20 and 40% of control. By contrast, the tissue compartment showing CMRglc suppression below 50% of the contralateral value was smaller than the volume of the ischemic core (defined as having CBF below 20% of control). In the ischemic borderzone (i.e., CBF 20-40% of control), LCMRglc was either only slightly impaired or, in circumscribed areas, showed elevations ranging up to fourfold. Apart from the early time point of the autoradiographic study, one explanation for this surprising maintenance of glucose metabolism despite moderate to severe degrees of ischemia is the fact that, owing to collateral channels, the transport capacity for glucose is not as readily exhausted as the oxygen transport capacity of the blood. This also explains why tissue glucose content is preserved within small MCA lesions even 7 h after MCA occlusion (Back et ai., 1994b). This consideration is of importance for the calculation of LCMRglc in ischemic brain because depletion of tissue glucose alters the lumped constant of the operational equation for local glucose utilization (Sokoloff et ai., 1977). There is evidence that 1 h after MCA occlusion, the rate of glucose utilization is still unchanged (Nakai et ai., 1987). In the cortical periinfarct zone, the lumped constant was found to be near normal (N edergaard et ai., 1986). It has been suggested that the 2-DG model may produce erroneous values of LCMRglc when blood flow falls below 0.10 ml g-1 min -1, and that without correction an 8% underestimation of LCMRglc occurs_ at flow rates of 0.05 ml g -1 min -1 (Tanaka et ai., 1985; Ginsberg, 1990). Although e4c]methylglucose studies in the present model of focal ischemia failed to show changes of glucose content within the ischemic focus at either the or the h time points studied (Yao et ai., 1995), we decided to exclude from further image analysis those CMRglc1 CBF ratio values that were based on LCMRglc data obtained from the ischemic core. In most animals studied for LCMRglc, marked hypermetabolism was observed in the ischemic borderzone, the degree of which was attenuated in the averaged image data set because a precise spatial overlap of these hypermetabolic foci was unlikely to occur. This disadvantage of the averaging procedure was partially compensated for by the method chosen to compute the CMRglc/CBF ratio for the average 3D reconstruction. In this process, every image pixel within the ischemic border zone of an individual 2-DG data set was divided by the corresponding flow value of every individual LCBF study (see Eq. 1). The fact that marked hypermetabolism in the ischemic borderzone was not observed in three of seven animals may be due in part to the severity of the ischemic insult, which may have hindered metabolic activation. Alternatively, the incidence of periinfarct depolarizations as a major cause of activation of glucose metabolism may have varied among the individual experimental animals, thus leading to varying degrees of 2-DG phosphorylation. Such depolarizations, associated with large transient negative shifts of the DC potential, are known to stimulate glucose utilization (Shinohara et ai., 1979), probably due to the transient release of intracellular potassium into the extracellular space, thereby increasing ion transport. In the infarct rim of MCA-occluded rats, hyperglycemia was able to prevent both the potassium transients and the increased 2-DG phosphorylation (Nedergaard and Astrup, 1986). In the present series, periinfarct depolarizations were measured that showed characteristics similar to those reported in previous publications (Nedergaard and Astrup, 1986; Back et ai., 1994a). Interestingly, the duration of recurrent depolarizations increased significantly with time during the first hour, i.e., under conditions of grossly reduced collateral circulation produced by bilateral CCA occlusion. In brain slices, the prevention of irreversible loss of function correlated closely with a shortening of the time spent in SD (Somjen et ai., 1990). The observation of reduced frequency and duration of DC transients after release of the contralateral CCA (i.e., 15 min prior to initiating the LCMRglc study or 30 min prior to measuring LCBF in the parallel series) supports the view that restoration of glucose (and oxygen) supply to the ischemic penumbra inhibits the generation of depolarizations and unburdens local metabolism. Since periinfarct SD-like DC transients are known to propagate to cortical regions distant from their origin (Dietrich et' ai., 1994), both the heterogeneity of LCMRglc in the borderzone and the widespread distribution of significantly increased CMRglc/CBF ratio values can J Cereb Blood Flow Metab, Vol. 15, No.4, 1995

11 576 T. BACK ET AL. be explained by the metabolic consequences of periinfarct DC potential shifts. The average CMRglc/CBF ratio determined in the hemisphere contralateral to MCA occlusion corresponded closely to baseline ratio values of previous reports (Ginsberg et ai., 1986, 1987). In the ischemic penumbra, the CMRglc/CBF ratio was elevated nearly fivefold, reflecting that uncoupling is a gradual process that occurs at various flow rates, with correspondingly variable levels of LCMRglc ranging from near-normal values to fourfold elevations. This complex relationship could be demonstrated in its 3D representation by directly comparing the 3D reconstructions of LCMRglc, LCBF, and the averaged CMRglc/CBF ratio. Thus, traditional difficulties in interpreting two-dimensional data sets for LCMRglc and LCBF evident in previous investigations (Pulsinelli et ai., 1982; Hossmann et ai., 1985; Tanaka et al., 1985) could be elegantly circumvented. The approach of 3D reconstruction is particularly valuable as alterations occur at variable distances surrounding the infarct core and in an irregular configuration. Topographic analysis of the steep flow gradient across the ischemic focus also benefits from the improvements available through 3D image analysis. In addition, 3D reconstruction makes it possible to apply threshold analysis to brain volumes rather than two-dimensional brain sections, providing a novel means of assessing metabolism-flow correlations. The comparison of ipsilateral brain volumes affected by comparable degrees of flow reduction versus metabolic depression (Fig. 4) clearly shows that disparity in these two volumes is most pronounced at flow values above 15% of control. As LCMRglc is likely to be underestimated at very low flow rates (Tanaka et ai., 1985; Ginsberg, 1990), this strengthens the notion that metabolism-flow uncoupling is a phenomenon particularly evident in the infarct borderzone. Since the infarct rim is characterized histologically as a transition zone of scattered neuronal cell loss (Mies et ai., 1983), the amount of which can be attenuated by hyperglycemia (Nedergaard and Diemer, 1985), uncoupling of metabolism and blood flow may be regarded as a major pathophysiological event leading to this type of ischemic injury. This is in agreement with the findings of Tanaka et al. (1985), who observed (at the periphery of the ischemic focus in MCA occlusion) a lack of histological damage when the metabolism-flow couple was preserved, but scattered neuronal injury in regions with disturbed coupling. Recent investigations employing noninvasive nuclear magnetic resonance techniques to monitor the evolution of ischemic infarcts during the first hours after onset have shown that the ischemic penumbra is a highly dynamic structure that moves toward the periphery while the infarct core grows over time (Kohno et ai., 1994). It is likely that, during this dynamic process, the uncoupling of metabolism and flow reaches a degree of dissociation that leads to irreversible ischemic damage. However, further investigations are needed to correlate the extent and distribution of metabolism-flow uncoupling to morphologic alterations within the evolving ischemic lesion, Acknowledgment: These studies were supported by USPHS grants NS and NS and by a research grant of the Deutsche Forschungsgemeinschaft (Ba 1390/ 1-1) to Dr. Back. We are deeply indebted to Brant D. Watson, Ph.D., for his expert assistance with the laserbased technique of photothrombosis; to Ms. J. Y. Loor Estades for her skillful preparation of the auto radiographic material; to Ms. I. Valdes, Ms. O. F. Alonso, and Raul Busto, B.S., for their superb assistance in conducting the autoradiographic studies; and to Mr. David W. Smith for indispensable computer system support. Ms. Helen Valkowitz helped to prepare the typescript. REFERENCES Back T, Kohno K, Hossmann K-A (1994a) Cortical negative DC deflections following middle cerebral artery occlusion and KCI-induced spreading depression: effect on blood flow, tissue oxygenation and electroencephalogram. J Cereb Blood Flow Metab 14: Back T, Hoehn-B erlage M, Kohno K, Hossmann K-A (1994b) Diffusion NMR imaging in experimental stroke: correlation with cerebral metabolites. Stroke 25 : Back T, Ginsberg MD, Zhao W, Loor JY, Valdes I, Alonso OF, Watson BD, Busto R (1994c) Three-dimensional image analysis of local metabolism-flow uncoupling in photothrombotic MCA occlusion in rats. Soc Neurosci Abstr 20 :1761 Dietrich WD, Feng Z-C, Leistra H, Watson BD, Rosenthal M (1994) Photothrombotic infarction triggers multiple episodes of cortical spreading depression in distant brain regions. J Cereb Blood Flow Metab 14: Ginsberg MD (1990) Local metabolic responses to cerebral ischemia. Cerebrovasc Brain Metab Rev 2:58-93 Ginsberg MD, Reivich M (1979) Use of the 2-deoxyglucose method of local cerebral glucose utilization in the abnormal brain: evaluation of the lumped constant during ischemia. Acta Physiol Scand 60(suppl 72): Ginsberg MD, Reivich M, Giandomenico M, Greenberg JH (1977) Local glucose utilization in acute focal cerebral ischemia: local dysmetabolism and diaschisis. Neurology 27: Ginsberg MD, Graham DI, Busto R (1985) Regional glucose utilization and blood flow following graded forebrain ischemia in the rat: correlation with neuropathology. Ann Neurol18: Ginsberg MD, Smith DW, Wachtel MS, Gonzalez-Carvajal M, Busto R (1986) Simultaneous determination of local cerebral glucose utilization and blood flow by carbon- 14 double-label autoradiography: method of procedure and validation studies in the rat. J Cereb Blood Flow Metab 6: Ginsberg MD, Dietrich D, Busto R (1987) Coupled forebrain increases of local cerebral glucose utilization and blooo flow during physiologic stimulation of a somatosensory pathway in the rat: demonstration by double-label autoradiography. Neurology 37: Heiss W-D, Rosner G (1983) Functional recovery of cortical neu- J Cereb Blood Flow Metab. Vol. 15, No.4, 1995

12 METABOLISM-FLOW UNCOUPLING IN FOCAL ISCHEMIA 577 rons as related to degree and duration of ischemia. Ann Ne u rol 14: Hossmann K-A, Mies G, Paschen W, Csiba L, Bodsch W, Rapin JR, Le Poncin-Lafitte M, Takahashi K (1985) Multiparametric imaging of blood flow and metabolism after middle cerebral artery occlusion in cats. J Ce re b Blood Flow Me tab 5: Iadecola C, Nakai M Mraovitch S, Ruggiero DA, Tucker LW, Reis DJ (1983) Global increase in cerebral metabolism and blood flow produced by focal electrical stimulation of dorsal medullary reticular formation in rat. Brain Re s 272: Jacewicz M, Tanabe J, Pulsinelli WA (1992) The CBF threshold and dynamics for focal cerebral infarction in spontaneously hypertensive rats. J Ce re b Blood Flow Me tab 12: Kohno K, Hoehn-Berlage M, Mies G, Back T, Hossmann K-A (1995) Relationship between diffusion-weighted magnetic resonance images, cerebral blood flow and energy state in experimental brain infarction. Mag n Res Im ag 13:73-80 Kuschinsky W, Suda S, Sokoloff L (1981) Local cerebral glucose utilization and blood flow during metabolic acidosis. Am J Phys iol 24 1: H772-H777 Lauritzen M, Diemer NH (1986) Uncoupling of cerebral blood flow and metabolism after single episode of cortical spreading depression in the rat brain. Brain Re s 370: Leao AAP (1944) Spreading depression of activity in the cerebral cortex. J Ne urophys iol 7: Markgraf CG, Kraydieh S, Prado R, Watson BD, Dietrich WD, Ginsberg MD (1993) Comparative histopathologic consequences of photothrombotic occlusion of the distal middle cerebral artery in Sprague-Dawley and Wistar rats. Stroke 24: Mies G, Auer LM, Ebhardt G, Traupe H, Heiss W-D (1983) Flow and neuronal density in tissue surrounding chronic infarction. Stroke 14:22-27 Nakai H, Matsuda H, Takara E, Diksic M, Meyer E, Yamamoto YL (1987) Simultaneous in vivo measurement of lumped constant and rate constants in experimental cerebral ischemia using F- 18 FDG. Stroke 18: Nedergaard M, Astrup J (1986) Infarct rim: effect of hyperglycemia on direct current potential and [14C]2-deoxyglucose phosphorylation. J Ce re b Blood Flow Me tab 6: Nedergaard M, Diemer NH (1985) Influence of hyperglycemia on ischemic brain damage after middle cerebral artery (MCA) occlusion in the rat. J Ce re b Blood Flow Me tab 5(suppl 1): Nedergaard M, Gj edde A, Diemer NH (1986) Focal ischemia of the rat brain: autoradiographic determination of cerebral glucose utilization, glucose content, and blood flow. J Ce re b Blood Flow Me tab 6: Paxinos G, Watson C (1982) The Rat Brain in Ste re otax ic Coordinates. New York, Academic Press Pulsinelli WA, Levy DE, Duffy TE (1982) Regional cerebral blood flow and glucose metabolism following transient forebrain cerebral ischemia. Ann Ne urol 1 1: Reivich M (1974) Blood flow metabolism couple in brain. In: Brain Dysf unction in Me tabolic D isorde rs (Plum F, ed), New York, Raven Press, pp Reivich M, Jehle J, Sokoloff L, Kety SS (1969) Measurement of regional cerebral blood flow with antipyrine-14c in awake cats. J Ap p l Phys iol 27: Roy CS, Sherrington CS (1890) On the regulation of the blood supply of the brain. J Phys iol (L ond) 11 : Sakurada 0, Kennedy C, Jehle J, Brown JD, Carbin GL, Sokoloff L (1978) Measurement of local cerebral blood flow with iodoc4c)antipyrine. Am J Phys iol 234:H59-H66 Savaki HE, Davidsen L, Smith C, Sokoloff L (1980) Measurement of free glucose turnover in brain. J Ne urochem 35: Shinohara M, Dollinger B, Brown G, Rapoport S, Sokoloff L (1979) Cerebral glucose utilization: local changes during and after recovery from spreading cortical depression. Scie nce 203 : Sokoloff L (198 1) Review. Localization of functional activity in the central nervous system by measurement of glucose utilization with radioactive deoxyglucose. J Ce re b Blood Flow Me tab 1 :7-36 Sokoloff L, Reivich M, Kennedy C, Des Rosiers MH, Patlak CS, Pettigrew KD, Sakurada 0, Shinohara M (1977) The [14C]deoxyglucose method for the measurement of local cerebral glucose utilization: theory, procedure, and normal values in the conscious and anesthetized rat. J Ne urochem 28: Somjen GG, Aitken PG, Balestrino M, Herreras 0, Kawasaki K (1990) Spreading depression-like depolarization and selective vulnerability of neurons. A brief review. Stroke 21(suppl 1 I) : III 179-III l 83 Strong AJ, Venables GS, Gibson G (1983) The cortical ischaemic penumbra associated with occlusion of the middle cerebral artery in the cat: 1. Topography of changes in blood flow, potassium ion activity, and EEG. J Ce re b Blood Flow Me tab 3:86-96 Tanaka K, Greenberg JH, Gonatas NK, Reivich M (1985) Regional flow-metabolism couple following middle cerebral artery occlusion in cats. J Ce re b Blood Flow Me tab 5: Tyson GW, T'easdale GM, Graham DJ, McCulloch J (1984) Focal cerebral ischemia in the rat: topography of hemodynamic and histopathological changes. Ann Ne urol 15: Welsh FA, Greenberg JH, Jones SC, Ginsberg MD, Reivich M (1980) Correlation between glucose utilization and metabolic levels during focal ischemia in cat brain. Stroke 11:79-84 Yao H, Ginsberg MD, Watson BD, Prado R, Dietrich WD, Kraydieh S, Busto R (1993) Failure of MK-801 to reduce infarct volume in thrombotic middle cerebral artery occlusion in rats. Stroke 24: Yao H, Ginsberg MD, Eveleth DD, LaManna JC, Watson BD, Alonso OF, Loor JY, Foreman JH, Busto R (1995) Local cerebral glucose utilization and cytoskeletal proteolysis as indices of evolving focal ischemic injury in core and penumbra. J Ce re b Blood Flow Me tab 15: Zhao W, Young TY, Ginsberg MD (1993) Registration and threedimensional reconstruction of autoradiographic images by the disparity analysis method. IEEE Trans Me d Im ag 12: Zhao W, Ginsberg MD, Smith DW (1995) Three-dimensional quantitative autoradiography by disparity analysis: theory and application to image-averaging of local cerebral glucose utilization. J Ce re b Blood Flow Me tab 15: J Cereb Blood Flow Metab, Vol. 15, No. 4, 1995