Effects of Streptozotocin-Induced Hyperglycemiaon Brain Damage Following Transient Ischemia

Cai Li,*,† Ping-An Li,* Qing-Ping He,* Yi-Bing Ouyang,*,†

and Bo K. Siesjo**Center for the Study of Neurological Disease, The Queen’s Neuroscience Institute,The Queen’s Medical Center, Honolulu, Hawaii 96813; and †Institute of Preclinical Sciences,N. Bethune University of Medical Sciences, Changchun, Jilin, China

Received April 2, 1998; revised June 3, 1998; accepted for publication June 11, 1998

Hyperglycemia is known to aggravate ischemic brain damage. The present experiments wereundertaken to explore whether hyperglycemia caused by streptozotocin-induced diabetes exacer-bates brain damage following transient brain ischemia as it does in animals acutely infused withglucose. Experimental diabetes was induced by injection of streptozotocin in rats which weresubjected to 10 min of forebrain ischemia either 1 week (1-wk) or 4 weeks (4-wk) after the induction ofdiabetes. Normoglycemic rats exposed to the same duration of ischemia and sham-operated diabeticrats served as controls. The animals underwent evaluation of clinical outcome and histopathologicalanalysis of brain damage. Postischemic seizures developed in 35.3 and 42.1% of 1-wk and 4-wkdiabetic hyperglycemic animals, respectively. The incidence of seizure was not different between thetwo groups. None of the diabetic animals with plasma glucose concentrations below 12 mM exhibitedseizure activity. The extent and distribution of brain damage were similar between 1- and 4-wk diabeticanimals. In the CA1 and in the subicular regions of hippocampus, both diabetic hyperglycemic andnormoglycemic animals showed 70–80% cell death. Diabetic hyperglycemic animals had more severeneuronal necrosis in the parietal cortex than normoglycemic animals. In diabetic hyperglycemicanimals, neuronal damage involved additional brain structures, e.g., cingulate cortex, thalamus nuclei,substantia nigra, pars reticulata, and the hippocampal CA3 sector, i.e., structures in which neuronswere not affected in normoglycemic ischemic subjects at this duration of ischemia. These findingsdemonstrate that diabetic hyperglycemic animals frequently develop postischemic seizures and thatstreptozotocin-induced hyperglycemia results in exacerbated postischemic brain damage of the samedensity and distribution as in acutely glucose-infused animals. r 1998 Academic Press

Key Words: hyperglycemia; diabetes; brain damage; ischemia; seizure.

INTRODUCTION

Many studies have identified diabetes as an indepen-dent and significant risk factor for stroke (Abbott et al.,1987; Barrett-Connor & Khaw, 1988). There is a greaterincidence of cerebral ischemic insults and higher riskof death from stroke in persons with diabetes mellitusthan in those who do not have the disease (Pulsinelli etal., 1983; Bell, 1994; Tuomilehto et al., 1996). Preisch-emic hyperglycemia aggravates brain damage causedby transient global or forebrain ischemia (for reviews,see Siesjo, 1988; Siesjo et al., 1993, 1996). The cardinalfeatures of such accentuated damage are rapidly evolv-ing neuronal necrosis, gross edema, and postischemic

seizures. The question arises whether hyperglycemiaof longer duration, as is observed in diabetes, willexacerbate ischemic brain damage in the same way asobserved in glucose-infused animals. This question isrelevant to the clinical situation because hyperglyce-mia is a characteristic feature in poorly regulateddiabetes and the incidence of cerebral ischemic insultsin patients with diabetes is greater than that in nondia-betic patients.

Available results do not give unequivocal informa-tion on this matter. Warner et al. (1992), studying ratswith 5–7 days of streptozotocin-induced diabetes,reported that such rats had damage to the CA1 sectorof the hippocampus and to the substantia nigra, pars

Neurobiology of Disease 5, 117–128 (1998)

Article No. NB980189

117

0969-9961/98 $25.00Copyright r 1998 by Academic PressAll rights of reproduction in any form reserved.

reticulata (SNPR) similar to that in animals madeacutely hyperglycemic. However, none of the diabeticrats exhibited postischemic convulsions, and damageto the cingulate cortex, the CA3 sector in the hippocam-pus, the thalamus, or the medial geniculate nuclei wasnot evaluated. A longer period of diabetes (3 months)by total pancreatectomy was studied in dogs subjectedto incomplete global ischemia of 20 min in duration(Sieber et al., 1996). Surviving diabetic dogs had thesame density of damage in the cerebellum, CA1,superior temporal gyrus, and caudate. However, onlyfour of eight diabetic dogs survived for 7 days. Inaddition, since the surviving diabetic dogs had a bloodglucose concentration of 7.0 6 2.8 mM, they were nothyperglycemic.

The purpose of the present study was to determinewhether diabetic hyperglycemia exacerbates brain dam-age and induces seizures following transient forebrainischemia as it does in acutely glucose-infused animals,whether the regional vulnerability of the brain to theischemic insult in diabetic hyperglycemia is differentfrom that in glucose-infused hyperglycemia, andwhether varying the duration of diabetes alters thedensity and distribution of ischemic brain damage.Experimental diabetes was induced with streptozoto-cin in rats. One week and 4 weeks after the inductionof diabetes, the animals were subjected to forebrainischemia of 10-min duration, the incidence of seizureswas recorded, and cerebral tissue damage was as-sessed with light microscopy after a 7-day recoveryperiod.

MATERIALS AND METHODS

Animals and Induction of Diabetes

Male Wistar rats of a special pathogen-free strain(Mollegaard’s Breeding Center, Copenhagen, Den-mark), weighing 330–370 g, were used for this study.The animals were fasted overnight before induction ofdiabetes but had free access to drinking water. Diabe-tes was induced by a single subcutaneous injection of40 mg/kg body weight of streptozotocin (STZ, SigmaChemical Co., St. Louis, MO), freshly dissolved in 0.1M citrate buffer at pH 4.5. The presence of hyperglyce-mia was confirmed 48–72 h after STZ administrationvia an Ames Glucometer II analysis after tailstick(Miles Laboratory Inc., Elkhart, IN). Before inductionof brain ischemia, arterial plasma glucose concentra-tions were measured on a Beckman Glucose Analyzer(Beckman Instrument Inc., Fulleron, CA). All animalswith plasma glucose levels exceeding 12 mM were

included in the hyperglycemic group. This bloodglucose concentration was chosen because, in a studyin glucose-infused animals with 10 min of brain isch-emia, Li et al. (1994, 1995) reported that in all thestructures examined except CA1, little or no damagecould be found in animals with plasma glucose concen-trations less than 12 mM, while extensive neuronalnecrosis was observed in animals with plasma glucoselevels higher than 12 mM. For the same reason,STZ-treated animals with plasma glucose concentra-tions of ,11.9 mM were excluded from the study.Age-matched rats were injected with an equal volumeof citrate buffer to serve as normoglycemic rats. Allanimals were allowed access to water and standard ratchow ad libitum. Prior to the ischemic insult or shamprocedure, the animals were fasted for 12 h with freeaccess to tap water.

Operative Procedures

The operative procedures applied were based on thetechniques described by Smith et al. (1984). The ratswere anesthetized with 3.5% halothane (Halothane,ICS Chemicals, England) in a mixture of N2O and O2

(70:30). Thereafter, the animals were endotracheallyintubated and connected to a respirator for artificialventilation. The halothane concentration was loweredto 1.5% for the operative period. The common carotidarteries were isolated via a middle neck incision andencircled with sutures for later clamping. A tail arteryand a tail vein were cannulated for blood pressuremonitoring, blood sampling, and drug infusion. A softSilastic catheter was inserted into the inferior cavalvein via the right jugular vein to allow rapid with-drawal of blood during the induction of brain isch-emia. Needle electrodes were placed in the temporalmuscles on the skull for electroencephalography (EEG)recording. Temperature probes were inserted into therectum and were placed subcutaneously on theskull bone to monitor core and head temperatures,respectively, which were maintained at 37°C by lampheating and a warm pad. After the surgical proce-dures, the halothane concentration was reduced to0.5–0.7%, the animals were immobilized with a 0.5 mgiv bolus dose of vecuronium bromide (Norcuron,Organon Teknika, Boxtel, Holland), followed by acontinuous infusion (2 mg/h). Heparin (90 IU/kg,Vitrum AB, Stockholm, Sweden) was injected iv beforethe first blood sample measurements. Ventilation andoxygen supply were adjusted to give a PaCO2 of 35–40mm Hg, a PaO2 of close to 100 mm Hg, and a pH ofclose to 7.4. Samples were analyzed for arterial blood

118 Li et al.

Copyright r 1998 by Academic PressAll rights of reproduction in any form reserved.

gases and pH on a standard microsample blood gasmonitor (ABL 300, Radiometer, Copenhagen Den-mark) and for plasma glucose concentration on aBeckman Glucose Analyzer.

Induction of Ischemia

After a steady-state period of 30 min, forebrainischemia of 10-min duration was induced by a combi-nation of bilateral carotid artery clamping and centralvenous exsanguination. Arterial blood pressure wasmaintained at 40–50 mm Hg during the period ofischemia. The EEG was monitored prior to, during,and after ischemia. Isoelectricity of EEG was consid-ered as the onset of ischemia. Ischemia was terminatedby reinfusion of the shed blood and removal of thecarotid clamps, and 0.5 ml of 0.6 M NaHCO3 solutionwas given intravenously to counteract systemic acido-sis. Halothane supply and vecuronium bromide infu-sion were discontinued at the end of ischemia. Thevenous catheter was removed and the neck incisionwas sutured. When animals regained spontaneousbreathing, they were extubated and disconnected fromthe respirator. The animals were then housed in cageswith free access to tap water and food. Over the 7-dayrecovery period, the animals were continuously ob-served for clinical manifestation of seizures and mortal-ity. Since the occurrence of tonic-clonic seizures couldbe observed only in the period from 0800 to 1800, thepossibility remains that unrecognized seizures oc-curred during the middle of the night. The forcefulmovements made by animals during a seizure result inthe bedding clips being disturbed and ejected thoughthe lid of the cage. Animals were kept isolated inindividual cages overnight and observed in the morn-ing to determine whether a seizure had occurred. Thebrains of surviving animals were perfusion-fixed after7 days of recovery.

Experimental Groups

Two major groups were studied, differing in theduration of diabetes, 1-wk and 4-wk, each of themcontaining sham-operated diabetic rats (Sham), normo-glycemic ischemic rats (NG), and diabetic hyperglyce-mic rats (HG).

Histological Preparation

After 7 days of recovery, the animals were reanesthe-tized with halothane and connected to a respirator. Thebrains were perfused via the ascending aorta with a

30-s flush of isotonic saline followed by 250 ml of 4%phosphate-buffered formaldehyde (pH 7.35). The brainswere stabilized at 4°C in situ for 1 day and then theywere removed from the skull and stored in coldfixative. The brains were cut coronally in 2.8-mm-thickslices and dehydrated in ethanol. The paraffin-embedded brain slices were subserially sectioned (5µm) and stained with a combination of celestine blueand acid fuchsin. The animals with postischemic sei-zures that survived for less than 7 days were immedi-ately perfusion-fixed after death.

Quantification of Brain Damage

The damaged neurons in the brains were evaluatedquantitatively in a blinded manner using light micros-copy at magnifications of 3400 by direct visual count-ing of acidophilic neurons or by visual estimation ofthe damaged area. In the hippocampal subiculum,CA1 and CA3 sectors, all necrotic and survivingneurons were counted in a coronal section at the levelof bregma 23.8 mm, and the percentages of deadneurons were calculated. For cerebral cortex, a coronalsection at the level of bregma 23.8 mm was selectedfor study and a modified method described by Mujsceet al. (1994) was used for quantification of the damagedneurons. In the parietal cortex, three equidistant pointsbetween the longitudinal fissure and the rhinal fissurewere marked on each side of the cerebral cortex. Threestrips of cortex adjacent to these points, from layers 1to 6, were studied on both sides using an eyepiecesquare grid calibrated for the 3400 microscopic objec-tive. The width of each strip equates one ocular grid inwidth at 3400 magnification. In the cingulate cortex,damaged neurons in the two strips were counted. Thenumber of damaged neurons/mm3 in the neocortexwas calculated. In the thalamus, in the medial genicu-late nuclei, and in substantia nigra, the degrees ofdamaged neurons were scored on a 4-grade scale,where grade 0 represents no observable histopathologi-cal changes; grade 1, ,10% damaged neurons; grade 2,11–50%; and grade 3, .50% of neurons damaged, orinfarction. In the caudoputamen the percentage ofdamaged neurons was estimated.

Statistics

Two-factor ANOVA followed by post hoc Scheffe’stest was used to detect significant differences in thephysiological parameters and pathological data. The

Diabetic Hyperglycemia and Ischemic Brain Damage 119

Copyright r 1998 by Academic PressAll rights of reproduction in any form reserved.

incidence of postischemic seizures was compared be-tween groups with the Fischer exact test. Gradedhistopathological data were analyzed by the nonpara-metric Mann–Whitney U test. Statistical significancewas considered to be present with P , 0.05.

RESULTS

The diabetic animals exhibited signs of polydipsia,polyphagia, and polyuria, and their body weightdecreased by 10–15% in the period prior to the opera-tive procedures. The physiological parameters andplasma glucose levels obtained from the groups beforethe onset of ischemia and at 10 min after recirculationare summarized in Table 1. Mean arterial blood pres-sure (MABP), PaCO2, PaO2, and pH values were in thephysiological ranges. PaO2 at 10 min after recirculationin 4-wk normoglycemic animals was higher than thatin 1-wk normoglycemic and 4-wk hyperglycemic ani-mals (see Table 1). This is unlikely to have influencedthe results. The plasma glucose concentrations indiabetic hyperglycemic and sham-operated diabeticgroups were similar and were significantly higher thanthose in the normoglycemic groups.

Postischemic Seizures

No animal with a plasma glucose concentration of,12 mM developed seizures, neither in the 1-wk nor inthe 4-wk diabetic groups. ‘‘Late’’ seizures occurredbetween 20 and 32 h of recirculation and ‘‘early’’seizures developed at 3 h after recovery in bothdiabetic hyperglycemic groups.

The relationship between the incidence of postisch-emic seizures and plasma glucose concentrations in1- and 4-wk diabetic hyperglycemic animals following10 min of brain ischemia is presented in Table 2. Noneof the normoglycemic ischemic animals exhibited sei-zure activity; thus their data are not included in Table 2.

As summarized in Table 2, in 1-wk diabetic hypergly-cemic animals, 6 of 17 developed clinical seizuresbetween 20 and 29 h (23.5 6 3.3, mean 6 SD) afterrecovery, thus displaying late seizures (Lundgren et al.,1991). The incidence of postischemic seizures in thisgroup was 35.3% and plasma glucose levels in theanimals with seizures were 19.6 6 2.9 (16.9–24.8) mM.Eight of 19 of the 4-wk diabetic hyperglycemic animalsexhibited clinical seizures between 3 and 32 h(20.9 6 11.6) after recirculation. Two of them showedearly postischemic seizures which started 3 h after

TABLE 1

Plasma Glucose Concentrations and Physiological Parameters before Ischemia and at 10 min after Recirculation

Groups n Glucose (mM) Temperature (°C) MABP (mm Hg) PaCO2 (mm Hg) PaO2 (mm Hg) Arterial pH

Preischemic1-wk

Sham-operated 5 20.3 6 1.7** 37.0 6 0.2 119 6 6 37.1 6 1.2 110 6 10 7.42 6 0.02Normoglycemic 9 8.1 6 0.8 36.9 6 0.2 122 6 11 37.8 6 1.9 105 6 9 7.44 6 0.12HG, seizure 1 5 19.6 6 3.0** 36.8 6 0.1 125 6 15 38.0 6 1.9 111 6 16 7.44 6 0.02HG, seizure 2 11 17.8 6 3.8** 36.9 6 0.2 122 6 9 38.8 6 1.5 110 6 11 7.45 6 0.02

4-wkSham-operated 4 20.5 6 1.1** 36.9 6 0.2 123 6 12 36.1 6 1.6 104 6 12 7.44 6 0.05Normoglycemic 10 7.9 6 0.5 37.0 6 0.1 126 6 12 39.3 6 2.0 115 6 13 7.42 6 0.04HG, seizure 1 7 19.4 6 5.4** 36.9 6 0.2 127 6 9 37.9 6 1.8 114 6 13 7.44 6 0.02HG, seizure 2 11 17.9 6 4.7** 37.0 6 0.1 122 6 10 38.8 6 2.5 106 6 9 7.44 6 0.04

Postischemic1-wk

Sham-operated 5 37.0 6 0.1 127 6 10 36.5 6 1.6 107 6 4 7.43 6 0.03Normoglycemic 9 37.0 6 0.1 135 6 10 38.3 6 2.1 102 6 10 7.45 6 0.03HG, seizure 1 5 37.1 6 0.1 143 6 9 39.5 6 2.2 108 6 8 7.43 6 0.04HG, seizure 2 11 37.1 6 0.1 134 6 7 38.9 6 1.6 104 6 8 7.44 6 0.03

4-wkSham-operated 4 37.1 6 0.2 138 6 4 38.7 6 2.8 106 6 5 7.42 6 0.06Normoglycemic 10 37.0 6 0.1 137 6 9 40.4 6 2.3 119 6 6* 7.41 6 0.02HG, seizure 1 7 37.0 6 0.1 136 6 6 39.3 6 2.1 111 6 10 7.42 6 0.07HG, seizure 2 11 37.0 6 0.1 130 6 12 39.3 6 2.7 103 6 8 7.43 6 0.03

Note. Data are expressed as means 6 SD. MABP, mean arterial blood pressure; HG, hyperglycemic.*P , 0.05, compared with 1-wk normoglycemic and 4-wk diabetic hyperglycemic groups (two-factor ANOVA followed by Scheffe’s test).**P , 0.01, compared with 1- and 4-wk normoglycemic groups.

120 Li et al.

Copyright r 1998 by Academic PressAll rights of reproduction in any form reserved.

recovery, and the other 6 animals developed seizuresbetween 21 and 32 h (20.1 6 12.3) after recirculation.The incidence of postischemic seizures in this groupwas 42.1% and plasma glucose levels were 20.0 6 5.2(12.7–25.0) mM (Table 2).

As shown in Table 2, in the 1-wk diabetic hypergly-cemic group, no obvious signs of postischemic seizureswere observed in 11 rats despite their having plasmaglucose concentrations (17.8 6 3.8 mM) similar to thosein animals with clinical seizures. In the 4-wk diabetichyperglycemic group, 11/19 animals did not exhibitseizure activity. In these 11 rats, the blood glucoselevels (18.2 6 4.8 mM) were similar to those in animalshaving seizures. There was no significant difference inseizure incidence and plasma glucose levels between1- and 4-wk diabetic hyperglycemic animals withseizures. Likewise, plasma glucose concentrations weresimilar in 1- and 4-wk diabetic hyperglycemic animalswithout seizures.

Distribution of Brain Damage

Normoglycemic and Sham-Operated Animals

In both the 1- and the 4-wk normoglycemic groups,the animals subjected to 10 min of brain ischemiashowed a similar distribution of brain damage in allstructures evaluated. The hippocampal CA1 (Fig. 1A)and subicular regions were heavily affected. Anotherstructure involved was the parietal cortex. No discern-ible neuronal damage was found in other brain regionsstudied (see Figs. 1B and 1C). In the sham-operatedanimals with diabetic hyperglycemia, the cerebralstructures examined did not show any histopathologi-cal abnormalities.

Diabetic Hyperglycemic Animals

The distribution of the brain damage presentedbelow is that in diabetic hyperglycemic animals sub-

jected to forebrain ischemia of 10-min duration, asdescribed under Materials and Methods.

Neocortex. All 11 animals in the 1-wk diabetichyperglycemic group showed acidophilic neuronalnecrosis in the parietal cortex. Three of the 11 animalsin this group displayed hypercellular infarcts.

The pattern of neuronal damage in the parietalcortex in 4-wk diabetic hyperglycemic animals wassimilar to that in 1-wk diabetic hyperglycemic animals.Likewise, 3 of 11 animals developed cortical infarcts inthis group.

In the cingulate cortex, neuronal damage was seenin 1- and 4-wk diabetic hyperglycemic animals (11 ratsin each) subjected to 10 min of brain ischemia. Microin-farcts in this structure appeared in 6 and 3 animals inthe 1- and 4-wk-diabetic hyperglycemic groups, respec-tively. In the remaining animals the damaged neuronswere scattered in the cingulate cortex (Fig. 1E).

Hippocampus. In the subicular and CA1 (Fig. 1D)regions of hippocampus, all 11 animals in the 1- and4-wk diabetic hyperglycemic groups exhibited acido-philic neuronal damage, affecting 83% of the cells.Bilateral symmetrical neuronal necrosis was fre-quently seen. The damaged neurons occurred withroughly equal frequency along the length of the CA1and subicular cell bands.

Five of 11 animals in the 1-wk diabetic hyperglyce-mic group showed neuronal damage in the hippocam-pal CA3 and CA4 regions. Similarly, 2 of 11 animals inthe 4-wk diabetic hyperglycemic group showed dam-age to the hippocampal CA3 and CA4 regions.

Thalamus. Acidophilic neuronal damage wasfound in 8 of 11 animals in the 1-wk diabetic hypergly-cemic group. Of the 8 animals 3 exhibited infarcts orsponginess of neuropil in the thalamus. Mild neuronaldamage in the medial geniculate nuclei was found in 4of 11 animals in this group.

Among 11 animals in the 4-wk diabetic hyperglyce-mic group, 9 showed acidophilic neuronal damage inthe thalamus. Three and one of the 9 animals displayed

TABLE 2

Clinical Seizures Following 10 min of Brain Ischemia in 1-wk and 4-wk Diabetic Hyperglycemic Rats

Seizure Seizure-free

N Pl. glucose (mM) Seizure onset time (h) Incidence (%) N Pl. glucose (mM) Incidence (%)

1-wk 6 19.6 6 2.9 23.5 6 3.3 35.3 11 17.8 6 3.8 64.7(16.9–24.8) (20–29) (12.7–25.2)

4-wk 8 20.0 6 5.2 20.9 6 11.6 42.1 11 18.2 6 4.8 57.9(12.7–25.0) (3–32) (12.1–25.1)

Note. Data are expressed as means 6 SD. Pl. glucose, plasma glucose concentrations.

Diabetic Hyperglycemia and Ischemic Brain Damage 121

Copyright r 1998 by Academic PressAll rights of reproduction in any form reserved.

FIG

.1.

Rep

rese

ntat

ive

phot

omic

rogr

aphs

show

ing

hist

opat

holo

gica

ldam

age

inth

ehi

ppoc

ampa

lCA

1(A

,D),

cing

ulat

eco

rtex

(B,E

),an

dth

alam

us(C

,F)7

day

saf

ter

10m

inof

brai

nis

chem

iain

norm

ogly

cem

icra

ts(A

,B

,C

)an

dd

iabe

tic

hype

rgly

cem

ican

imal

s(D

,E

,F)

.T

here

isan

alm

ost

com

plet

ene

uron

alne

cros

isin

the

hipp

ocam

pal

CA

1re

gion

sin

norm

ogly

cem

ic(A

)and

dia

beti

chy

perg

lyce

mic

(D)r

ats.

All

neur

ons

inth

eci

ngul

ate

cort

ex(B

)and

inth

eth

alam

us(C

)app

eare

dno

rmal

inno

rmog

lyce

mic

rats

,whi

lem

any

necr

otic

neur

ons

inth

eci

ngul

ate

cort

ex(E

)an

din

the

thal

amus

(F)

wer

eob

serv

edin

dia

beti

chy

perg

lyce

mic

anim

als.

Aan

dB

wer

eob

tain

edfr

om1-

wk

norm

ogly

cem

ican

dd

iabe

tic

hype

rgly

cem

icra

ts;o

ther

pane

lsw

ere

obta

ined

from

4-w

kno

rmog

lyce

mic

and

dia

beti

chy

perg

lyce

mic

rats

,res

pect

ivel

y.C

eles

tine

blue

/ac

idfu

chsi

nst

ain;

bar,

50µm

.

122

sponginess (Fig. 1F) and infarcts with surroundingvacuolization, respectively, in this structure. Eight of 11animals in this group showed damage to the medialgeniculate nuclei but the lesion was usually mild.

Substantia nigra. Sponginess of neuropil and in-farcts with peripheral sponginess were seen in theSNPR in 7 of 11 animals in the 1-wk diabetic hypergly-cemic group. Of the 7 animals 3 showed symmetricalbilateral lesions in the SNPR. In the 4-wk diabetichyperglycemic group, 6 of 11 animals exhibited spongi-ness of neuropil in the SNPR. In these 6 animals 3showed bilateral symmetrical lesions in the SNPR. Theinfarcts observed in the SNPR were characterized byhypercellular infiltration of macrophages. The lesion inthe SNPR was chiefly localized in the lateral aspect ofthe nucleus and never extended into the adjacent parscompacta of substantia nigra.

Caudoputamen. Neither sham controls nor normo-glycemic ischemic animals showed damage in thecaudate putamen, while acidophilic neurons could befound in 3 of 11 animals in the 1-wk and 4 of 11 animalsin the 4-wk diabetic hyperglycemic groups. The dam-aged neurons were mainly localized in the lateral partsof this structure.

Animals with Postischemic Seizures

Among the 1- and the 4-wk diabetic hyperglycemicanimals with seizures, 2/6 and 1/8 animals, respec-tively, died after seizure onset. The brains of theanimals were not perfusion-fixed and histological ex-amination was not performed on these brains. Accord-ingly, only 4 and 7 brains from the 1- and 4-wk diabetichyperglycemic groups, respectively, were evaluatedhistopathologically after 7 days of recovery. Theseanimals thus had transient seizures of the early andlate types.

In the 1-wk diabetic hyperglycemic group, neuronaldamage was seen in the parietal and cingulate cortex,the hippocampal CA1, CA3, and subicular regions, thethalamus, and the SNPR in all 4 animals with seizures,and in the caudoputamen and the medial geniculatebody in 3 of the 4 animals. All 7 animals developingseizures in the 4-wk diabetic hyperglycemic groupshowed acidophilic neuronal damage in the parietaland cingulate cortex, the hippocampal CA1, CA3, andsubicular regions, the thalamus, and the SNPR, as wellas the medial geniculate nuclei. Six of 11 animalsexhibited mild or moderate neuronal damage in thecaudoputamen.

The histopathological alterations noted in animalswith seizures were as follows. First, of four animals

with seizures in the 1-wk diabetic hyperglycemicgroup, two animals showed acidophilic neurons in thehippocampal CA4 sector. Similarly, among seven ani-mals developing seizures in the 4-wk diabetic hypergly-cemic group, acidophilic neurons were found in theCA4 in three animals. Second, in animals with seizuresin the 1-wk diabetic hyperglycemic group, two out offour animals showed sponginess in SNPR, and oneanimal showed a unilateral SNPR lesion. Of sevenanimals with seizures in the 4-wk diabetic hyperglyce-mic group, six displayed sponginess in SNPR. Third, inthe animals which developed seizures, preacidophilicneurons were seen frequently in the CA1 and subicularregions of hippocampus. A possible explanation is thatthe survival times of the animals with seizures wereshorter than those of seizure-free animals, hence neuro-nal necrosis in the dorsal hippocampus was not al-lowed to mature.

Extent of Brain Damage

To assess the severity of brain damage, the animalswith and without seizures were first combined intoone group, referred to as diabetic hyperglycemic Group2. The extent of damage in the brain regions examinedin this group was compared with that in the othergroups (see Figs. 2–5). Then, the diabetic hyperglyce-mic animals with and without seizures were separatedinto two subgroups, i.e., groups with and withoutseizures. The severity of brain damage was evaluatedin the two subgroups (see Fig. 6).

1-wk Diabetic Hyperglycemic Animals

As illustrated in Figs. 2 and 3, for parietal cortex,cingulate cortex, thalamus, and SNPR, the animals inhyperglycemic Group 2 showed greater neuronal dam-age than normoglycemic and sham-operated animals(Figs. 2 and 3). In the hippocampal CA1 and subicularregions, the severity of the neuronal injury was similarbetween hyperglycemic Group 2 and the normoglyce-mic group. For the hippocampal CA3 region, caudopu-tamen, and medial geniculate nuclei, the animals inhyperglycemic Group 2 showed numerically greaterneuronal damage than normoglycemic animals, butthere was no statistically significant difference.

4-wk Diabetic Hyperglycemic Animals

As demonstrated in Figs. 4 and 5, in the parietalcortex, cingulate cortex, thalamus, and SNPR, the

Diabetic Hyperglycemia and Ischemic Brain Damage 123

Copyright r 1998 by Academic PressAll rights of reproduction in any form reserved.

extent of neuronal damage in hyperglycemic Group 2was more severe than that in normoglycemic andsham-operated animals (Figs. 4 and 5). In the hippocam-pal CA1 and subicular regions, the extent of theneuronal damage was similar between hyperglycemicGroup 2 and the normoglycemic group. For the hippo-campal CA3 region, caudoputamen, and medial genicu-late nuclei, the animals in hyperglycemic Group 2seemed to show more severe damage than normogly-cemic animals, but statistical significance was notreached.

The extent of neuronal damage in the parietal andcingulate cortex in animals without seizures was com-pared with that in animals with seizures. As Fig. 6demonstrates, in 1- and 4-wk diabetic hyperglycemicanimals, the extent of damage to the parietal andcingulate cortex in animals with seizures was moresevere than that in seizure-free animals (Fig. 6). Theseverity of the neuronal damage in other brain regions

studied was not significantly different between theanimals with and without seizures.

DISCUSSION

As stated in the Introduction, the present study wasdesigned to explore whether diabetic hyperglycemiaexacerbates brain damage and induces postischemicseizures following transient forebrain ischemia,whether the pathological features of brain damage indiabetic hyperglycemic animals were different fromthose seen in glucose-infused hyperglycemic animals,and whether the duration of diabetes influences thedistribution and density of brain damage. The presentresults demonstrated that, as is observed in glucose-infused hyperglycemic animals, some diabetic hyper-glycemic animals showed postischemic seizures andadditionally aggravated brain damage after cerebralischemia. Furthermore, the pathological characteristics

FIG. 2. Histopathological outcome in the parietal and cingulatecortex following 10 min of ischemia in each experimental group at 1week after the induction of diabetes. Sham, sham-operated diabeticrats; NG, normoglycemic rats; and HG, diabetic hyperglycemicanimals with and without seizures. Each symbol represents oneanimal. Histopathological examination was performed after 1 weekof recovery. Damage is presented as number of dead neurons/mm2

in the structures. ns denotes no significant difference. Asterisksdenote extent of differing damage; ***P , 0.001 (two-factor ANOVAfollowed by Scheffe’s test).

FIG. 3. Damage score in the thalamus nuclei and substantia nigrapars reticulata (SNPR) following 10 min of ischemia in eachexperimental group at 1 week after the induction of diabetes.Histopathological evaluation was performed after 7 days of recov-ery. Damage is presented as damage scores. For explanation ofabbreviations see legend to Fig. 2. Asterisks denote damage scoresdifferent from each other; **P , 0.01, and ***P , 0.001, respectively(Mann–Whitney U test).

124 Li et al.

Copyright r 1998 by Academic PressAll rights of reproduction in any form reserved.

of brain damage in diabetic hyperglycemic animalswere largely identical to those observed in glucose-infused animals. We will discuss the results underthree headings.

Seizure Development

One of the cardinal features of hyperglycemia-related brain damage is the development of postisch-emic seizures (Myers, 1979; Siemkowicz, 1985; Warneret al., 1987). The incidence of seizures in hyperglycemicanimals depends on the severity of the ischemic insult,on blood glucose concentration, and on the period ofrecovery. In glucose-infused hyperglycemic animalssubjected to transient brain ischemia, the incidence ofpostischemic seizures has been somewhat variable. Forexample, Lundgren et al. (1990) reported that among 26hyperglycemic animals (plasma glucose 20.7–24.0 mM)with 6 and 18 h of recirculation following 15 min ofischemia, 15 (58%) showed seizure activity. The rela-tively low incidence of seizures in that study couldhave resulted from the brief recovery times. Thus, in

another experiment these authors found that all 6(100%) hyperglycemic animals subjected to 10 min ofischemia developed seizures after 1–4 and 18 h ofrecovery (Lundgren et al., 1991). The latter result is inaccordance with other data, demonstrating that hyper-glycemic animals almost invariably develop postisch-emic seizures (Siemhowicz & Hansen, 1978; Warner etal., 1987). It has been shown that the preischemicplasma glucose level has a marked effect on thedevelopment of postischemic seizures. In glucose-infused hyperglycemic rats, Li and colleagues reportedthat seizures were observed only at plasma glucoseconcentrations exceeding 11–12 mM and that no sei-zure-free animals were seen with a glucose concentra-tion above 15.5 mM (Li et al., 1994, 1995). It should beemphasized, though, that other authors, using a differ-ent strain of rats, have reported a lower incidence ofpostischemic seizures. Most importantly, Warner et al.(1992) reported that the incidence of postischemicseizures in glucose-infused hyperglycemic rats, sub-jected to 10 min of forebrain ischemia, was 67%.

The present results on animals with streptozotocin-induced diabetes differ from those of Warner et al.

FIG. 4. Neuronal damage in the parietal and cingulate cortexfollowing 10 min of ischemia in each experimental group at 4 weeksafter the induction of diabetes. Histopathological examination wasperformed after 1 week. For explanation of abbreviations andsymbols see the legend to Fig. 2.

FIG. 5. Damage score in the thalamic nuclei and SNPR following10 min of ischemia in each experimental group at 4 weeks after theinduction of diabetes. Histopathological evaluation was performedafter 7 days of recovery. Damage is expressed as damage scores. Forexplanation of abbreviations and symbols see legend to Fig. 3.

Diabetic Hyperglycemia and Ischemic Brain Damage 125

Copyright r 1998 by Academic PressAll rights of reproduction in any form reserved.

FIG. 6. Indicating the neuronal damage in the parietal and cingulate cortex after 10 min of ischemia in 1-wk (A) and 4-wk (B) diabetichyperglycemic animals with and without seizures. Histopathological evaluation was performed after 7 days of recovery. Damage is presented asnumber of dead neurons/mm2. Asterisks denote significant difference in the extent of damage (*P , 0.05 and ***P , 0.001) between the animalswith and without seizures (unpaired t test).

126

(1992). Thus, whereas these authors failed to recordpostischemic seizures in their diabetic rats, more thanone-third of the present animals showed postischemicseizures, whether the duration of diabetes was 1 or 4weeks. However, in one respect our results agree withthose of Warner et al. (1992). Thus, both studies showthat diabetic animals have a lower incidence of postisch-emic seizures, even when acutely hyperglycemic anddiabetic rats have similar blood glucose concentra-tions. The reason for this is unknown. However, thepresent data demonstrate that extending the durationof diabetes from 1 to 4 weeks does not additionallydecrease the incidence of seizures. In the present studywe observed that no animal with a glucose concentra-tion of ,12 mM developed postischemic seizures,demonstrating that there seems to exist a glucosethreshold of about 12 mM for seizure induction indiabetic hyperglycemic rats, above which seizuresdevelop. The results are thus similar to those reportedby Li et al. (1994, 1995), who showed that there was aglucose threshold of 10–13 mM for seizure develop-ment in glucose-infused hyperglycemic animals. Asdemonstrated in Table 2, however, not all diabeticanimals with plasma glucose levels exceeding 12 mMshowed postischemic seizures. In fact, plasma glucoseconcentrations, duration of ischemia, and period ofrecirculation were similar between the animals withand without seizures. The findings suggest that thefactors involved in seizure development remain to beelucidated.

It has been suggested that damage to the SNPR isessential for seizure development after transient brainischemia (Smith et al., 1984). The results of this studyconfirmed that the lesions in SNPR were observed inall 1- and 4-wk diabetic hyperglycemic animals withseizures; however, some animals showed somewhatless severe SNPR damage. A spongy appearance ofSNPR was seen in most of the diabetic animals withseizures.

The present results demonstrated that 7/11 animalswithout seizures in each of the 1- and 4-wk diabetichyperglycemic groups showed unilateral or bilateralSNPR damage, with sponginess or sponginess com-bined with infarcts. The results indicate that the SNPRdamage was seen not only in the diabetic animals withseizures but also in animals without seizures, suggest-ing that the SNPR damage is not unique for seizureanimals. Similar findings were reported by Warner etal. (1992), who found that, in 10 experimental diabeticrats, postischemic seizures were absent, but SNPRdamage was found in all animals.

The pathogenesis of postischemic seizures has not

been elucidated. There is evidence indicating thatinhibitory afferents to the SNPR are largely derivedfrom caudoputamen (Gale, 1985; Saji & Volpe, 1993).One could assume that excessive damage to the caudo-putamen would reduce suppression of hyperexcitationof SNPR, leading to SNPR necrosis and seizure devel-opment. Since hyperglycemia enhances damage to thecaudoputamen, the possibility of seizure inductionwould increase in hyperglycemic animals. Data fromthe present study are not fully consistent with thishypothesis. Thus, we noted in the present experimentthat some diabetic animals showed only mild or nodamage in the caudoputamen, but they had moresevere damage in the SNPR and developed seizures,and that some diabetic hyperglycemic animals withoutseizures showed SNPR damage while no or milddamage was seen in the caudoputamen. Based onthese findings, we suggest that a significant correlationbetween damage to the caudoputamen and damagedSNPR could not be established, nor was there acorrelation between seizures or either caudoputamenor SNPR damage.

Distribution and Extent of Brain Damage

A major objective of the present study was to explorewhether STZ-induced ‘‘chronic’’ hyperglycemia hasthe same effect as glucose-infused acute hyperglyce-mia on ischemic neuronal damage. The results demon-strate that the both the anatomical distribution and theseverity of brain damage in diabetic hyperglycemicanimals are similar to those observed in acutely glucose-infused hyperglycemic animals (Smith et al., 1988; Li etal., 1994, 1995). Thus, the damage is exaggerated indiabetic animals compared with normoglycemic sub-jects; furthermore additional structures such as thecingulate cortex, thalamus nuclei, SNPR, the hippocam-pal CA3 sector, and medial geniculate nuclei wererecruited in the damage process, and infarcts or micro-infarcts were frequently observed in the neocortex,thalamus, and SNPR.

The percentage of neuronal damage in the hippocam-pal CA1 sector in normoglycemic animals was about70–80%, which is lower than that reported previously(Smith et al., 1984). This might be related to the residualblood flow during the 10-min ischemic period. Dia-betic hyperglycemic animals showed a tendency to-ward exacerbation of hippocampal CA1 and subiculardamage.

There is experimental evidence demonstrating thatCA4 pyramidal cells are as susceptible to ischemia as

Diabetic Hyperglycemia and Ischemic Brain Damage 127

Copyright r 1998 by Academic PressAll rights of reproduction in any form reserved.

are the CA1 and subicular neurons (Smith et al., 1984).Damage to the CA4 neurons of the hippocampus wasobserved in normoglycemic and glucose-infused hyper-glycemic animals, and in the latter the CA4 cells wereseverely affected (Smith et al., 1988). However, in thepresent experiments, the neuronal damage in the CA4sector of the hippocampus was detected in only a fewdiabetic hyperglycemic animals without seizures andin half of the animals with seizures, while suchneuronal damage was not found in normoglycemicanimals subjected to 10 min of brain ischemia.

In summary, the present results have shown thatone-third to one-half of diabetic animals (1 or 4 weeksfollowing streptozotocin injection) subjected to 10 minof forebrain ischemia develop postischemic seizuresand that they all show postischemic neuronal damagealmost identical to that observed in animals madeacutely hyperglycemic by glucose infusion. Thus, al-though streptozotocin-injected animals show a lowerincidence of postischemic seizures, the pathology ob-served in these animals is almost identical to thatobserved in acutely glucose-infused animals. Thus, amajor determinant of the final damage incurred as aresult of transient ischemia is the plasma glucoseconcentration.

ACKNOWLEDGMENTS

This work was supported by the Swedish Medical ResearchCouncil (14X-263), the United States Public Health Service via theNational Institutes of Health (5 R01 NS07838), and the JuvenileDiabetes Foundation International.

REFERENCES

Abbott, R. D., Donahue, R. P., MacMahon, S. W., Reed, D. M., & Yano,K. (1987) Diabetes and the risk of stroke. J. Am. Med. Assoc. 257,949–952.

Barrett-Connor, E., & Khaw, K.-T. (1988) Diabetes mellitus: Anindependent risk factor for stroke? Am. J. Epidemiol. 128, 116–123.

Bell, D. S. H. (1994) Stroke in the diabetic patient. Diabetes Care 17,213–219.

Gale, K. (1985) Mechanisms of seizure control mediated byg-aminobutyric acid: Role of the substantia nigra. Fed. Proc. 44,2414–2424.

Li, P.-A., Shamloo, M., Smith, M.-L., Katsura, K., & Siesjo, B. K. (1994)The influence of plasma glucose concentrations on ischemic braindamageis a threshold function. Neurosci. Lett. 177, 63–65.

Li, P.-A., Shamloo, M., Katsura, K., Smith, M.-L., & Siesjo, B. K. (1995)Critical values for plasma glucose in aggravating ischemic braindamage: Correlation to extracellular pH. Neurobiol. Dis. 2, 97–108.

Lundgren, J., Cardell, M., Wieloch, T., & Siesjo, B. K. (1990) Preisch-emic hyperglycemia and postischemic alteration of rat brainpyruvate dehydrogenase activity. J. Cereb. Blood Flow Metab. 10,536–541.

Lundgren, J., Smith, M.-L., & Siesjo, B. K. (1991) Influence ofmoderate hypothermia on ischemic brain damage incurred underhyperglycemic conditions. Exp. Brain Res. 84, 91–101.

Mujsce, D. J., Towfighi, J., Heitjan, D. F., & Vannucci, R. C. (1994)Difference in intraischemic temperature influence neurologicaloutcome after deep hypothermic circulatory arrest in newborndogs. Stroke 25, 1433–1442.

Myers, R. E. (1979) Lactic acid accumulation as a cause of brainedema and cerebral necrosis resulting from oxygen deprivation.In: Advances in Prenatal Neurology (R. Korobkin & G. Guillemi-nault, Eds.), pp. 85–114. Spectrum, New York.

Pulsinelli, W. A., Levy, D. E., Sigsbee, P., & Plum, F. (1983) Increaseddamage after ischemic stroke in patients with hyperglycemia withor without established diabetes mellitus. Am. J. Med. 74, 540–544.

Saji, M., & Volpe, B. (1993) Delayed histologic damage and neurondeath in the substantia nigra reticulata following transient fore-brain ischemia depends on the extent of initial striatal injury.Neurosci. Lett. 155, 47–50.

Sieber, F. E., Martin, L. J., Brown, P. R., Palmon, S. C., & Traystman,R. J. (1996) Diabetic chronic hyperglycemia and neurologic out-come following global ischemia in dogs. J. Cereb. Blood Flow Metab.16, 1230–5.

Siemkowicz, E. (1985) The effect of glucose upon restitution aftercerebral ischemia: A summary. Acta Neurol. Scand. 71, 417–427.

Siemkowicz, E., & Hansen, A. J. (1978) Clinical restitution followingcerebral ischemia in hypo-, normo- and hyperglycemic rats. ActaNeurol. Scand. 58, 1–8.

Siesjo, B. K. (1988) Acidosis and ischemic brain damage. Neurochem.Pathol. 9, 31–88.

Siesjo, B. K., Katsura, K., Mellergard, P., Ekholm, A., Lundgren, J., &Smith, M.-L. (1993) Acidosis-related brain damage. In: Neurobiol-ogy of Ischemic Brain Damage (K. Kogure, K.-A. Hossmann, & B. K.Siesjo, Eds.), pp. 23–48. Elsevier, Amsterdam.

Siesjo, B. K., Katsura, K., & Kristian, T. (1996) Acidosis-relateddamage. In: Advances in Neurology: Cellular and Molecular Mecha-nisms of Ischemia Brain Damage (B. K. Siesjo & T. Wieloch, Eds.), pp.209–236. Lippincott–Raven, Philadelphia.

Smith, M.-L., Bendek, G., Dahlgren, N., Rosen, I., Wieloch, T., &Siesjo, B. K. (1984) Models for studying long-term recoveryfollowing forebrain ischemia in the rat. 2. A 2-vessel occlusionmodel. Acta Neurol. Scand. 69, 385–401.

Smith, M.-L., Kalimo, H., Warner, D. S., & Siesjo, B. K. (1998)Morphological lesions in the brain preceding the development ofpostischemic seizures. Acta Neuropathol. 76, 253–264.

Tuomilehto, J., Rastenyte, D., Jousilahti, P., Sarti, C., & Vartiainen, E.(1996) Diabetes mellitus as a risk factor for death from stroke:Prospective study of the middle-aged Finnish population. Stroke27, 210–215.

Warner, D., Smith, M.-L., & Siesjo, B. K. (1987) Ischemia in normo-and hyperglycemic rats: Effects on brain water and electrolytes.Stroke 18, 464–471.

Warner, D., Gionet, T. X., Todd, M. M., & McAllister, A. M. (1992)Insulin-induced normoglycemia improves ischemic outcome inhyperglycemic rats. Stroke 23, 1775–1781.

128 Li et al.

Copyright r 1998 by Academic PressAll rights of reproduction in any form reserved.