LABORATORY I N V E S T I G A T I O N
infarction; stroke
Hemorrhagic Infarct Conversionin ExperimentalStroke From the Department of
Gabrielle M de Courten-Myers, MD*
Study objective: This study investigated the relations between hem-
Pathology, University o f
Maria Kleinholzt
Cincinnati;* and Department of
Pat Holm*
orrhagic infarction and occlusion, release, levels of glycemia, brain energy state, and lactate content after cerebrovascular occlusion.
Veterans Affairs Medical Center
Gary DeVoe*
Research Service, t Cincinnati,
Ohio. Received for publication May 22, 1991. Revision received September 30, 1991. Accepted
for publication October 17, 1991. Presented in part at the Society
for Academic Emergency
Gary Schmittt
Design: Prospective, controlled laboratory investigation.
Kenneth R Wagner, PhDt
Type of paJd;icipants: One hundred six pentobarbital-anesthetized cats.
Ronald E Myers, MD, PhDt
Interventions: The middle cerebral artery was occluded with a Yasargil clip transorbitally either temporarily (0.5, four, and eight hours) or permanently. Normoglycemic and hyperglycemic animals were closely monitored for eight hours. Brain pathology was assessed after two weeks' survival or at the time of spontaneous animal death. Topographic brain metabolite studies were carried out after four hours of middle cerebral artery occlusion.
Medicine Annual Meeting in San
Measurements and main results: Morphometric quantitation of
Diego, California, May 1989,
cerebral hemorrhage and infarction and fluorometric determinations of blood and brain tissue, glucose, glycogen, lactate, adenosine triphosphate, and phosphocreatinefrom 16 topographicbrain sites were carriedout. Twenty-one of 82 (25.6%) animals evaluated neuropathologically showed hemorrhagic infarcts. Occluding the artery in hyperglycemic animals caused fivefold more frequent and 25-fold more extensive hemorrhage into infarcts than in normoglycemic animals. Temporary occlusion with clip release after four hours in hyperglycemic animals caused the most extensive hemorrhage into infarcts. Most hemorrhages into infarcts (81%) took place in animals that died within a few hours after they experienced ischemia and that showed infarction and marked edema of the entire middle cerebral artery territory. Linear regression analyses demonstrated a close relation between hemorrhage into infarcts and near-total energy depletion (adenosine triphosphate, less than 0.3 #M/g; phosphocreatine, less than 0.5 #M/g)in brain sites that showed extremely high tissue lactate concentrations (more than 30 #M/g). The biochemical changes that correlated with hemorrhage into infarcts were more marked than those with infarcts without hemorrhage. I~
and the 15th International Joint
Conference on Stroke and Cerebral Circulation in Orlando,
Florida, February 1990. This study was supported by a grant from the National
Institutes of Health (NINDS) RO1NS21776 - and by Department of Veterans Affairs
Medical Research Service funds.
14/120
ANNALSOF EMERGENCYMEDICINE 21:2 FEBRUARY1992
INFARCT CONVERSION de Courten-Myers et al
Conclusion: Hyperglycemiaand restoration of blood flow to ischemic territories were strong risk factors for hemorrhagic infarct conversion. Concomitant tissue metabolic changes suggest that marked tissue energy depletion accompanied by acidosis damages brain vessels and renders them penetrable for edema fluid and, ultimately, red blood cell extravasation. [de Courten-MyersGM, Kleinholz M, Holm P,DeVoe G, Schmitt G, Wagner KR, Myers RE: Hemorrhagic infarct conversion in experimental stroke. Ann Emerg Med February 1992; 21:120-126.]
iNTRODUCTION Recent methods of acute stroke treatment have focused on reestablishing blood flow to ischemic brain regions by either pharmacologic clot lysis or surgical recanalization. Howevei, these treatments have been restricted because of fears that restoration of blood flow to ischemic brain regions may provoke hemorrhage 1-6 and aggravate the patient's clinical state--particularly after the use of anticoagulant therapy.7, 8 To reduce risk of complications after the restoration of blood flow, the duration of stroke symptoms allowed in clinical studies of fibrinolysis is typically short. Thus, an ongoing clinical trial of efficacy of tissue plasminogen activator in ischemic stroke includes only those patients who have exhibited stroke symptoms for less than 90 minutes. Although setting such stringent time restrictions for the treatment of stroke is relatively safe, it excludes large segments of the stroke population from the potential benefits of treatment. 9 Why do infarcts develop hemorrhages, and what are the risk factors accounting for such bleeding? Hemorrhages into infarcts appear as multifocal, secondary points of bleeding that are mainly petechiae or areas of confluence of small hemorrhages. In autopsy studies, about 30% of recent brain infarcts show multifocal parenchymal hemorrhages, whereas brain areas damaged as a consequence of embolic strokes show still higher prevalences of bleeding, lo-ls However, autopsy studies probably bias the results in favor of hemorrhagic infarcts because hemorrhage into infarcts occurs most often in association with large infarcts. Computed tomography (CT) scans demonstrate an incidence of bleeding into brain infarcts of 5% to 43% depending on scan timing and infarct type.13,14 In cardioembolic stroke, hemorrhagic transformations are delayed for six to 12 hours and usually occur within 48 hours in anticoagulated patients.7, 8 The physiologic and brain chemical mechanisms that underlie hemorrhagic infarct conversion in stroke remain poorly understood. Better understanding of these mechanisms might provide more precise criteria with which to identify an appropriate time frame for therapeutic intervention.
FEBRUARY 1992
21:2
ANNALS OF EMERGENCY MEDICrNE
This study investigated mechanisms that may underlie hemorrhagic infarct conversion using an experimental stroke model involving middle cerebral artery (MCA) occlusion in cats. It determined the incidences of hemorrhagic infarction and the extents of hemorrhage into infarcts, and it correlated brain pathologic findings with physiologic and brain biochemical changes that take place during ischemia in normoglycemic and hyperglycemic cats exposed to temporary, or permanent MCA occlusion. Hyperglycemia during and after MCA occlusion in cats quadruples infarct size after permanent occlusionlS and increases by sevenfold the risk of fatal MCA territory edema occurring soon after four-hour MCA clip release. 16 MATERIAL AND METHODS We have detailed the surgical, brain pathologic, and biochemical techniques used in the present study elsewhere.iS-17 We exposed nine cat groups (Table) to temporary or permanent MCA occlusion while the animals were anesthetized with pentobarhita]. After placement of arterial and venous femoral vessel catheters advanced to the thoracic aorta and inferior vena cava, respectively, a transorbital approach was used to occlude the MCA using Yasargil aneurysm clips. We monitored multiple cardiovascular and blood compositional parameters in all cats for more than eight hours before, during, and after MCA occlusion. Parameters monitored continuously included blood pressure, heart rate, and core body temperature; those measured intermittently included arterial blood gases, pH, hematocrit, and glucose and pentoharbital concentrations. We excluded a number of animals because of incomplete MCA occlusion, intracranial bleeding from the operative site, or physiologic parameters lying outside of previously established mean values (2 _+SD). Postmortem India ink injections into the ipsilateral internal carotid artery in the animals of the groups with permanent MCA occlusion verified complete MCA occlusions in all instances in which the surgeon earlier was able to visually confirm a proper clip placement across the artery. The experimental protocol ~" Table. Animal groups studied
Duratiun of Vessel Occlusion Permanent Permanent 8-Hr temporary 8-Hr temporary 4-Hr temporary 4-Hr temporary 4-Hr temporary 4-Hr temporary O.5-Hrtemporary
Glycemia 20 mmol/L 6 mmol/L 20 mmol/L 6 mmol/L 20 mmol/t 6 mmol/L 20 mmol/L 6 mmol/L 20 mmoi/L
End Point Pathologic Pathologic Pathologic Pathologic Pathologic Pathologic Biochemic Biochemic Pathologic
N 12 13 I0 10 13 t2 14 10 12
12 1/ 15
INFARCT
CONVERSION
de Courten-Myers et al
I
was approved by the Institutional Animal Care and Use committees of both the University of Cincinnati College of Medicine and the Department of Veterans Affairs Medical Center. We regulated the animals' serum glucose concentrations to constitute normoglycemic and hyperglycemic animal groups. All animals were food-deprived for 48 hours before occlusion to stabilize their serum glucose concentrations. The two glycemia levels were achieved by infusing animals with saline (0.9%) or 10% glucose in water DwW solutions beginning one hour before MCA occlusion, yielding serum glucose concentrations close to 6 or 20 mmol/L (106 or 360 mg/dL). We maintained these serum glucose concentrations for six hours after the clip was applied in the animals of groups with permanent or four-hour reversible occlusion. We maintained the animals of the group with clip release after eight hours of occlusion hyperglycemic for the entire eight hours of occlusion. All animals received the same volumes of fluids during the day of occlusion (10 mL/kg/hr for one hour before and 5 mL/kg/hr for eight hours after occlusion). Serum glucose concentrations were measured using a Beckman Glucose Analyzer 2 (Beckman, Fullerton, California) based on a glucose oxydase enzymatic method. We assessed brain pathologic outcome after animal survival for two weeks or after animals died spontaneously soon after clip application or release. All brains were perfusion-fixed in 10% formalin and examined grossly and microscopically after hematoxylin and eosin and Well staining whole mount sections. Infarct sizes were determined morphometrieally based on six representative, standard, coronal histologie brain sections and a computerassisted image analysis system (BioQuant, Nashville, Tennessee). The extents of hemorrhage into infarcts were quantified using a Magisean image analysis sytem (Magiscan, Gateshead, United Kingdom). Brain areas containing extravasated blood were estimated by digitizing the microscopic images, identifying areas of hemorrhage, and segmenting all images by selecting appropriate gray levels. Because this computer-assisted identification and segmentation process included RBCs contained within congested blood vessels and occasional nonhlood artifacts in addition to RBCs contained in pockets of hemonhage, the investigators edited the individual images until "all hemorrhages" and "only hemorrhages" were included in areas for quantitation. All brain regions showing histologie evidence of hemorrhage were measured in this fashion and summed to provide a quantitative estimate of the extent of bleeding in each animal. We assayed brain metabolite concentrations in other cats exposed to four hours of MCA occlusion. We fixed their brains for
16/122
biochemical analysis using in situ freezing a s previously described.X7,18 All frozen heads were sectioned into six representative coronal blocks that corresponded to the planes of the histological sections used for pathologic analysis. We sampled 16 ipsilateral and 16 contralateral brain sites (11 of cerebral cortex, one of head of caudate nucleus, and four of hemispheral white matter), 14 of which lie within the territory of the occluded MCA. We analyzed all tissue samples for adenosine triphospate (ATP), phosphocreatine (PCr), glucose, glycogen, and lactate; metabolite concentrations were expressed as micromoles per gram wet weight. We used standard perchloric acid extraction procedures and enzymatic-fluorometric assay methods to measure brain metabolite concentrations. 17 We tested the representativeness of the MCA territory sites chosen for topographic brain biochemical assays by comparing the actual infarct sizes in 59 cats as determined directly using microscopic morphometric techniques and as independently derived based on the percent of sample sites that showed infarction. The infarct sizes estimated using these two independent methods correlated well (regression, .66064; correlation, .905; P < .001). We correlated the findings of brain chemistry with those of brain pathology based on site-by-site linear regression analyses of group means. We also carried out multiple linear regressions using different threshold concentrations for ATP, PCr, and lactate to identify those concentrations that provided coefficients that approached 1.0, assuming that such concentrations should most closely correlate with observed pathologic changes (provided significant correlations do occur). We analyzed data using Fisher's exact test for 2 x 2 contigency tables, Student's t-tests, analysis of variance, and multiple linear regressions. The level of statistical significance was P < .05.
RESULTS Eighty-two animals were studied for brain pathologic end points focusing on the presence of hemorrhagic infarcts and the extent of parenchymal bleeds. Twenty-one (25.6%) showed hemorrhagic infarcts (a typical example is illustrated in Figure 1). The incidences and average hemorrhage sizes observed among the animals of the different experimental groups are described (Figure 2). As can be seen, hemorrhagic infarcts occurred most ot~en and most extensively in hyperglycemic animals. Thus, among 35 normoglycemic and 35 hyperglycemic animals subjected to four- or eight-hour temporary or permanent MCA occlusion, hyperglycemia increased the hemorrhagic conversion rate more than fivefold (17 of 35 vs three of 35, P < .01). Hyperglycemic •
ANNALS OF EMERGENCY MEDICINE
21:2
FEBRUARY1992
INFARCT CONVERSION de Courten-Myers et al
animals also showed hemorrhages into infarcts that were 25-fold larger than in normoglycemic animals (mean + SD, 6.92 + 18.45 vs 0.22 + 0.91 mm, respectively, in hyperglycemic and normoglycemic cats; P < .05). Restoring blood pressure and blood flow to the area of ischemia by clip removal also affected the incidence and extent of hemorrhages into infarction. Thus, although only three of 25 (12%) animals with permanent MCA clip application showed hemorrhagic infarct conversions, 17 of 45 (38%) of those with the clip removed showed conversions. This more-than-threefold increase of hemorrhagic conversions occurring after clip release is statistically significant (P < .05). Furthermore, clip release was associated with an average aggregated size of hemorrhages into infarcts ninefold larger than in otherwise comparable cases where the clip was retained (mean + SD, 5.28 + 16.48 versus 0.49 + 1.83 mm, respectively; P < .10). Releasing the clip after four hours of MCA occlusion in hyperglycemic cats provoked especially large hemorrhages into infarcts (Figure 2). All 17 animals that died with infarction of the entire MCA territory accompanied by marked edema also showed focal hemorrhages throughout the areas of damage. These findings confirm a strong association between large infarcts and hemorrhagic conversion. Twenty-one animals bled into areas of infarct. Of physiologic parameters measured in these animals, only arterial blood pressure correlated with extent of bleeding. However, these animals'
linear correlation regarding this parameter fell short of statistical significance (P < .06). The timing of bleeding into areas of tissue injury in these animals remains unknown. However, cross sections taken through the frozen brains of the animals used for biochemical study showed no visible bleeding despite four hours of maintained MCA occlusion. Thus, most hemorrhagic conversions (17 of 21) probably developed some time after the clip was released but before the animal died one to 36 hours later or before the cats were eleetively killed after two weeks of survival. Some animals killed after two weeks' survival showed hemosiderin deposits demonstrable by iron stain within the hemorrhagic infarcts. A positive hemosiderin reaction at two weeks suggests that the bleeding took place during the first week after MCA occlusion because approximately seven days are required for this reaction to become positive. Topographic ATP, PCr, lactate, glucose, and glycogen analyses were carried out in the brains of normoglyeemic and hyperglycemic eats exposed to four hours of MCA occlusion. The concentrations of these metabolites in these animals were then correlated with the incidences of hemorrhagic infarcts affecting these same sites but as determined from parallel neuropathologie studies carried out in other animals. That is, for each of the 14 MCA territory sites sampled for chemistry, we correlated the metabolite group mean with the corresponding hemorrhagic infarct frequencies that occurred after clip release after the same duration (four hours) of MCA occlusion (12 and 13, •
Figure 1.
Figure 2.
Hemorrhagic infarcts in cats with MCA occlusion. Left: Coronal brain section showing ill-defined right MCA territory infarct with edema causing a midline shift. The injured MCA terriorty shows areas of confluent hemorrhage affecting the head of the caudate nucleus and the fronto-orbital cortex. Four hours of MCA occlusion in a hyperglycemic cat; death occurred at 36 hours. ]light: Coronal brain section at the level of the anterior third ventricle showing hemorrhages at the periphery of the MCA territory infarct. This region, probably supplied by ana~stomotic channels, was hemorrhagic in a hyperglycemic eat with permanent MCA occlusion that was killed after two weeks" survival.
The incidence of hemorrhagic infarcts (left pane]) after MCA occlusion is higher in hyperglycemic than normoglycemic animals and is maximal after blood flow is restored afier foar and eight hours of reversible occlusion. The largest hemorrhages are found in hyperglycemic cats in which MCA blood flow was restored afier four hours of ncclusion (right panel). The glucose infusions maintaining hyperglycemia were stopped at six hours in all animals, except those in the group with eight-hour clip release. In the latter group, the glucose infusions were continued for eight hours, so all animals were equally hyperglycemic at the time of clip release.
[///J
Normoglycemic
r.,'////~ Hyperglycemic
2fl "18
o.2.
> 16 ~
infarction; stroke
Hemorrhagic Infarct Conversionin ExperimentalStroke From the Department of
Gabrielle M de Courten-Myers, MD*
Study objective: This study investigated the relations between hem-
Pathology, University o f
Maria Kleinholzt
Cincinnati;* and Department of
Pat Holm*
orrhagic infarction and occlusion, release, levels of glycemia, brain energy state, and lactate content after cerebrovascular occlusion.
Veterans Affairs Medical Center
Gary DeVoe*
Research Service, t Cincinnati,
Ohio. Received for publication May 22, 1991. Revision received September 30, 1991. Accepted
for publication October 17, 1991. Presented in part at the Society
for Academic Emergency
Gary Schmittt
Design: Prospective, controlled laboratory investigation.
Kenneth R Wagner, PhDt
Type of paJd;icipants: One hundred six pentobarbital-anesthetized cats.
Ronald E Myers, MD, PhDt
Interventions: The middle cerebral artery was occluded with a Yasargil clip transorbitally either temporarily (0.5, four, and eight hours) or permanently. Normoglycemic and hyperglycemic animals were closely monitored for eight hours. Brain pathology was assessed after two weeks' survival or at the time of spontaneous animal death. Topographic brain metabolite studies were carried out after four hours of middle cerebral artery occlusion.
Medicine Annual Meeting in San
Measurements and main results: Morphometric quantitation of
Diego, California, May 1989,
cerebral hemorrhage and infarction and fluorometric determinations of blood and brain tissue, glucose, glycogen, lactate, adenosine triphosphate, and phosphocreatinefrom 16 topographicbrain sites were carriedout. Twenty-one of 82 (25.6%) animals evaluated neuropathologically showed hemorrhagic infarcts. Occluding the artery in hyperglycemic animals caused fivefold more frequent and 25-fold more extensive hemorrhage into infarcts than in normoglycemic animals. Temporary occlusion with clip release after four hours in hyperglycemic animals caused the most extensive hemorrhage into infarcts. Most hemorrhages into infarcts (81%) took place in animals that died within a few hours after they experienced ischemia and that showed infarction and marked edema of the entire middle cerebral artery territory. Linear regression analyses demonstrated a close relation between hemorrhage into infarcts and near-total energy depletion (adenosine triphosphate, less than 0.3 #M/g; phosphocreatine, less than 0.5 #M/g)in brain sites that showed extremely high tissue lactate concentrations (more than 30 #M/g). The biochemical changes that correlated with hemorrhage into infarcts were more marked than those with infarcts without hemorrhage. I~
and the 15th International Joint
Conference on Stroke and Cerebral Circulation in Orlando,
Florida, February 1990. This study was supported by a grant from the National
Institutes of Health (NINDS) RO1NS21776 - and by Department of Veterans Affairs
Medical Research Service funds.
14/120
ANNALSOF EMERGENCYMEDICINE 21:2 FEBRUARY1992
INFARCT CONVERSION de Courten-Myers et al
Conclusion: Hyperglycemiaand restoration of blood flow to ischemic territories were strong risk factors for hemorrhagic infarct conversion. Concomitant tissue metabolic changes suggest that marked tissue energy depletion accompanied by acidosis damages brain vessels and renders them penetrable for edema fluid and, ultimately, red blood cell extravasation. [de Courten-MyersGM, Kleinholz M, Holm P,DeVoe G, Schmitt G, Wagner KR, Myers RE: Hemorrhagic infarct conversion in experimental stroke. Ann Emerg Med February 1992; 21:120-126.]
iNTRODUCTION Recent methods of acute stroke treatment have focused on reestablishing blood flow to ischemic brain regions by either pharmacologic clot lysis or surgical recanalization. Howevei, these treatments have been restricted because of fears that restoration of blood flow to ischemic brain regions may provoke hemorrhage 1-6 and aggravate the patient's clinical state--particularly after the use of anticoagulant therapy.7, 8 To reduce risk of complications after the restoration of blood flow, the duration of stroke symptoms allowed in clinical studies of fibrinolysis is typically short. Thus, an ongoing clinical trial of efficacy of tissue plasminogen activator in ischemic stroke includes only those patients who have exhibited stroke symptoms for less than 90 minutes. Although setting such stringent time restrictions for the treatment of stroke is relatively safe, it excludes large segments of the stroke population from the potential benefits of treatment. 9 Why do infarcts develop hemorrhages, and what are the risk factors accounting for such bleeding? Hemorrhages into infarcts appear as multifocal, secondary points of bleeding that are mainly petechiae or areas of confluence of small hemorrhages. In autopsy studies, about 30% of recent brain infarcts show multifocal parenchymal hemorrhages, whereas brain areas damaged as a consequence of embolic strokes show still higher prevalences of bleeding, lo-ls However, autopsy studies probably bias the results in favor of hemorrhagic infarcts because hemorrhage into infarcts occurs most often in association with large infarcts. Computed tomography (CT) scans demonstrate an incidence of bleeding into brain infarcts of 5% to 43% depending on scan timing and infarct type.13,14 In cardioembolic stroke, hemorrhagic transformations are delayed for six to 12 hours and usually occur within 48 hours in anticoagulated patients.7, 8 The physiologic and brain chemical mechanisms that underlie hemorrhagic infarct conversion in stroke remain poorly understood. Better understanding of these mechanisms might provide more precise criteria with which to identify an appropriate time frame for therapeutic intervention.
FEBRUARY 1992
21:2
ANNALS OF EMERGENCY MEDICrNE
This study investigated mechanisms that may underlie hemorrhagic infarct conversion using an experimental stroke model involving middle cerebral artery (MCA) occlusion in cats. It determined the incidences of hemorrhagic infarction and the extents of hemorrhage into infarcts, and it correlated brain pathologic findings with physiologic and brain biochemical changes that take place during ischemia in normoglycemic and hyperglycemic cats exposed to temporary, or permanent MCA occlusion. Hyperglycemia during and after MCA occlusion in cats quadruples infarct size after permanent occlusionlS and increases by sevenfold the risk of fatal MCA territory edema occurring soon after four-hour MCA clip release. 16 MATERIAL AND METHODS We have detailed the surgical, brain pathologic, and biochemical techniques used in the present study elsewhere.iS-17 We exposed nine cat groups (Table) to temporary or permanent MCA occlusion while the animals were anesthetized with pentobarhita]. After placement of arterial and venous femoral vessel catheters advanced to the thoracic aorta and inferior vena cava, respectively, a transorbital approach was used to occlude the MCA using Yasargil aneurysm clips. We monitored multiple cardiovascular and blood compositional parameters in all cats for more than eight hours before, during, and after MCA occlusion. Parameters monitored continuously included blood pressure, heart rate, and core body temperature; those measured intermittently included arterial blood gases, pH, hematocrit, and glucose and pentoharbital concentrations. We excluded a number of animals because of incomplete MCA occlusion, intracranial bleeding from the operative site, or physiologic parameters lying outside of previously established mean values (2 _+SD). Postmortem India ink injections into the ipsilateral internal carotid artery in the animals of the groups with permanent MCA occlusion verified complete MCA occlusions in all instances in which the surgeon earlier was able to visually confirm a proper clip placement across the artery. The experimental protocol ~" Table. Animal groups studied
Duratiun of Vessel Occlusion Permanent Permanent 8-Hr temporary 8-Hr temporary 4-Hr temporary 4-Hr temporary 4-Hr temporary 4-Hr temporary O.5-Hrtemporary
Glycemia 20 mmol/L 6 mmol/L 20 mmol/L 6 mmol/L 20 mmol/t 6 mmol/L 20 mmol/L 6 mmol/L 20 mmoi/L
End Point Pathologic Pathologic Pathologic Pathologic Pathologic Pathologic Biochemic Biochemic Pathologic
N 12 13 I0 10 13 t2 14 10 12
12 1/ 15
INFARCT
CONVERSION
de Courten-Myers et al
I
was approved by the Institutional Animal Care and Use committees of both the University of Cincinnati College of Medicine and the Department of Veterans Affairs Medical Center. We regulated the animals' serum glucose concentrations to constitute normoglycemic and hyperglycemic animal groups. All animals were food-deprived for 48 hours before occlusion to stabilize their serum glucose concentrations. The two glycemia levels were achieved by infusing animals with saline (0.9%) or 10% glucose in water DwW solutions beginning one hour before MCA occlusion, yielding serum glucose concentrations close to 6 or 20 mmol/L (106 or 360 mg/dL). We maintained these serum glucose concentrations for six hours after the clip was applied in the animals of groups with permanent or four-hour reversible occlusion. We maintained the animals of the group with clip release after eight hours of occlusion hyperglycemic for the entire eight hours of occlusion. All animals received the same volumes of fluids during the day of occlusion (10 mL/kg/hr for one hour before and 5 mL/kg/hr for eight hours after occlusion). Serum glucose concentrations were measured using a Beckman Glucose Analyzer 2 (Beckman, Fullerton, California) based on a glucose oxydase enzymatic method. We assessed brain pathologic outcome after animal survival for two weeks or after animals died spontaneously soon after clip application or release. All brains were perfusion-fixed in 10% formalin and examined grossly and microscopically after hematoxylin and eosin and Well staining whole mount sections. Infarct sizes were determined morphometrieally based on six representative, standard, coronal histologie brain sections and a computerassisted image analysis system (BioQuant, Nashville, Tennessee). The extents of hemorrhage into infarcts were quantified using a Magisean image analysis sytem (Magiscan, Gateshead, United Kingdom). Brain areas containing extravasated blood were estimated by digitizing the microscopic images, identifying areas of hemorrhage, and segmenting all images by selecting appropriate gray levels. Because this computer-assisted identification and segmentation process included RBCs contained within congested blood vessels and occasional nonhlood artifacts in addition to RBCs contained in pockets of hemonhage, the investigators edited the individual images until "all hemorrhages" and "only hemorrhages" were included in areas for quantitation. All brain regions showing histologie evidence of hemorrhage were measured in this fashion and summed to provide a quantitative estimate of the extent of bleeding in each animal. We assayed brain metabolite concentrations in other cats exposed to four hours of MCA occlusion. We fixed their brains for
16/122
biochemical analysis using in situ freezing a s previously described.X7,18 All frozen heads were sectioned into six representative coronal blocks that corresponded to the planes of the histological sections used for pathologic analysis. We sampled 16 ipsilateral and 16 contralateral brain sites (11 of cerebral cortex, one of head of caudate nucleus, and four of hemispheral white matter), 14 of which lie within the territory of the occluded MCA. We analyzed all tissue samples for adenosine triphospate (ATP), phosphocreatine (PCr), glucose, glycogen, and lactate; metabolite concentrations were expressed as micromoles per gram wet weight. We used standard perchloric acid extraction procedures and enzymatic-fluorometric assay methods to measure brain metabolite concentrations. 17 We tested the representativeness of the MCA territory sites chosen for topographic brain biochemical assays by comparing the actual infarct sizes in 59 cats as determined directly using microscopic morphometric techniques and as independently derived based on the percent of sample sites that showed infarction. The infarct sizes estimated using these two independent methods correlated well (regression, .66064; correlation, .905; P < .001). We correlated the findings of brain chemistry with those of brain pathology based on site-by-site linear regression analyses of group means. We also carried out multiple linear regressions using different threshold concentrations for ATP, PCr, and lactate to identify those concentrations that provided coefficients that approached 1.0, assuming that such concentrations should most closely correlate with observed pathologic changes (provided significant correlations do occur). We analyzed data using Fisher's exact test for 2 x 2 contigency tables, Student's t-tests, analysis of variance, and multiple linear regressions. The level of statistical significance was P < .05.
RESULTS Eighty-two animals were studied for brain pathologic end points focusing on the presence of hemorrhagic infarcts and the extent of parenchymal bleeds. Twenty-one (25.6%) showed hemorrhagic infarcts (a typical example is illustrated in Figure 1). The incidences and average hemorrhage sizes observed among the animals of the different experimental groups are described (Figure 2). As can be seen, hemorrhagic infarcts occurred most ot~en and most extensively in hyperglycemic animals. Thus, among 35 normoglycemic and 35 hyperglycemic animals subjected to four- or eight-hour temporary or permanent MCA occlusion, hyperglycemia increased the hemorrhagic conversion rate more than fivefold (17 of 35 vs three of 35, P < .01). Hyperglycemic •
ANNALS OF EMERGENCY MEDICINE
21:2
FEBRUARY1992
INFARCT CONVERSION de Courten-Myers et al
animals also showed hemorrhages into infarcts that were 25-fold larger than in normoglycemic animals (mean + SD, 6.92 + 18.45 vs 0.22 + 0.91 mm, respectively, in hyperglycemic and normoglycemic cats; P < .05). Restoring blood pressure and blood flow to the area of ischemia by clip removal also affected the incidence and extent of hemorrhages into infarction. Thus, although only three of 25 (12%) animals with permanent MCA clip application showed hemorrhagic infarct conversions, 17 of 45 (38%) of those with the clip removed showed conversions. This more-than-threefold increase of hemorrhagic conversions occurring after clip release is statistically significant (P < .05). Furthermore, clip release was associated with an average aggregated size of hemorrhages into infarcts ninefold larger than in otherwise comparable cases where the clip was retained (mean + SD, 5.28 + 16.48 versus 0.49 + 1.83 mm, respectively; P < .10). Releasing the clip after four hours of MCA occlusion in hyperglycemic cats provoked especially large hemorrhages into infarcts (Figure 2). All 17 animals that died with infarction of the entire MCA territory accompanied by marked edema also showed focal hemorrhages throughout the areas of damage. These findings confirm a strong association between large infarcts and hemorrhagic conversion. Twenty-one animals bled into areas of infarct. Of physiologic parameters measured in these animals, only arterial blood pressure correlated with extent of bleeding. However, these animals'
linear correlation regarding this parameter fell short of statistical significance (P < .06). The timing of bleeding into areas of tissue injury in these animals remains unknown. However, cross sections taken through the frozen brains of the animals used for biochemical study showed no visible bleeding despite four hours of maintained MCA occlusion. Thus, most hemorrhagic conversions (17 of 21) probably developed some time after the clip was released but before the animal died one to 36 hours later or before the cats were eleetively killed after two weeks of survival. Some animals killed after two weeks' survival showed hemosiderin deposits demonstrable by iron stain within the hemorrhagic infarcts. A positive hemosiderin reaction at two weeks suggests that the bleeding took place during the first week after MCA occlusion because approximately seven days are required for this reaction to become positive. Topographic ATP, PCr, lactate, glucose, and glycogen analyses were carried out in the brains of normoglyeemic and hyperglycemic eats exposed to four hours of MCA occlusion. The concentrations of these metabolites in these animals were then correlated with the incidences of hemorrhagic infarcts affecting these same sites but as determined from parallel neuropathologie studies carried out in other animals. That is, for each of the 14 MCA territory sites sampled for chemistry, we correlated the metabolite group mean with the corresponding hemorrhagic infarct frequencies that occurred after clip release after the same duration (four hours) of MCA occlusion (12 and 13, •
Figure 1.
Figure 2.
Hemorrhagic infarcts in cats with MCA occlusion. Left: Coronal brain section showing ill-defined right MCA territory infarct with edema causing a midline shift. The injured MCA terriorty shows areas of confluent hemorrhage affecting the head of the caudate nucleus and the fronto-orbital cortex. Four hours of MCA occlusion in a hyperglycemic cat; death occurred at 36 hours. ]light: Coronal brain section at the level of the anterior third ventricle showing hemorrhages at the periphery of the MCA territory infarct. This region, probably supplied by ana~stomotic channels, was hemorrhagic in a hyperglycemic eat with permanent MCA occlusion that was killed after two weeks" survival.
The incidence of hemorrhagic infarcts (left pane]) after MCA occlusion is higher in hyperglycemic than normoglycemic animals and is maximal after blood flow is restored afier foar and eight hours of reversible occlusion. The largest hemorrhages are found in hyperglycemic cats in which MCA blood flow was restored afier four hours of ncclusion (right panel). The glucose infusions maintaining hyperglycemia were stopped at six hours in all animals, except those in the group with eight-hour clip release. In the latter group, the glucose infusions were continued for eight hours, so all animals were equally hyperglycemic at the time of clip release.
[///J
Normoglycemic
r.,'////~ Hyperglycemic
2fl "18
o.2.
> 16 ~
