.

..,;.,..J. .~."",," ,,._-;:i1"._-',¡,,;~, ;.,":,~.,

(J)

, SHOC/(~

A Comparison of Hypertonic to Isotonic Fluid in the Resuscitationof Brain Injury and Hemorrhagic Shock1

JON C. W ALSH, M.D.,* JING ZHUANG, M.D., t AND STEVEN R. SHACKFORD, M.D., F .A.C.S. t

tDepartment of Surgery, College of Medicine, Universityof Vermont, Burlington, Vermont 05405, and *Department of Surge/)'College of Medicine, Universit." of California, San Diego, La Jolla, California 92122

Submitted for publication April12, 1990,.~--G".L..:;. ..L:.,

-

We studied the early and late effects of hypertonicresuscitation (HR) on the injured brain using a porcinemodel of hemorrhagic shock and focal cryogenic braininjury. After shock, swine were randomly assigned toreceive a bolus (4 cc/kg) of either Ringers lactate (RL)or 7.5% hypertonic saline in 6% Dextran 70, followedby either RL or hypertonic sodium lactate to restoremean arterial pressure to baseline. AII animals werestudied for 24 hr after the start of resuscitation. BolusHR improved cerebral blood ftow (CBF) with a lowerintracranial pressure (ICP) than RL. Continued hyper-tonic resuscitation prolonged the period of improvedCBF and low ICP. At 24 hr CBF had deteriorated in theregio n of injury in all study groups and in the uninjuredhemisphere in swine receiving RL. These data suggestthat rapid resuscitation without increasing ICP for upto 6 hr as seen with hypertonic ftuid could conceivablyallow adequate time for surgical evacuation of mass le-sions and effectively prevent secondary brain injury.This work underscores the importance of prolonged pe-riods of study when evaluating brain resuscitationfrom traumatic shock. @ 1991 Academie Preso,lDeo

water content compared toRinger's lactate [6-9]. Thesestudies were of short duration «6 hr), however, and theearly benefit of reduced ICP and water content may notpersist since maximal brain swelling does not often oc-cur unti112-18 hr after injury [10].

Since there have been no previous studies which evalu-ated the use of hy-pertonic fluid for prolonged periodsafter the initial resuscitation, we compared three differ-ent fluid regimens in a 24-hr porcine model of hemor-rhagic shock and focal brain injury. \Ve postulated thathypertonic fluid would restore systemic and cerebralperfusion more rapidly than isotonic fluids. \Ve furtherhypothesized that the initial use of hypertonic fluid fol-lowed by additional resuscitation with isotonic fluidwould eliminate any early benefits seen from the use ofhypertonic fluid alone.

METHODS

Animal Preparation

Twenty-three matuTe swine averaging 40.3 :!: 8.6 kgwere fasted overnight but allowed free access to water.On the morning of the experiment, they were premedi-cated with ketamine (20 ml/kg) and atropine (0.5 mg)given intramuscularly along with cephalozin sodium (1g) and gentamicin (80 mg). Anesthesia was induced with2% halothane and oxygen by mask. The animals wereendotracheally intubated and connected to a Harvardventilator (Harvard Co., Chicago, IL), delivering a tidalvolume of 10-14 mI/kg. Anesthesia was maintained with0.5-1.0% halothane and a 2% succinylcholine drip at1.5-2.0 mg/kg/hr for the duration of the experimento Aheating blanket was used to keep the temperature be-tween 37 and 39°C. The femoral artery and vein werecannulated bilaterally under sterile conditions with poly-ethylene catheters (PE 200) for blood sampling andpressure monitoring. A quadruple lumen pulmonary ar-tery catheter (VIP No. 93 A-831, 7.5 F, American Ed-wards Labs, Irvine, CA) was inserted through the femo-Tal ve in for determination of cardiac filling pressuresand cardiac output by thermodilution, using a cardiac

INTRODUCTION

Resuscitation of the head-injured patient in shock isoften complicated by the need to treat hypotension dueto hemorrhage without exacerbating brain swelling. Ide-ally, one would prefer to administer a fluid that wouldrapidly restore perfusion and improve cerebral bloodflow without increasing the intracranial pressure. Hy-pertonic fluids are capable of rapid volume expansionand improved perfusion following resuscitation fromhemorrhagic shock in a variety of animal models [1-4]and in initial clinical trials [5]. In various types of headinjury models, hypertonic resuscitation resulted in alower intracranial pressure (ICP) and a lower cerebral

1 Supported by grants froro NINCDS (1-R01-NS 28637-01) and

NIGMS (GMO8247).

2840022.4804/91 $1.50Copyright 1&, 1991 by Academic Press, Inc.All rights of reproduction in any form reserved.

WALSH, ZHUANG, ANO SHACKFORO: RESUSCITATION OF BRAIN INJURY ANO SHOCK 285

ANTERIOR

..Blood Flow Probeo

o IntracraníalPressure Monitor

.Cryogeníc lesionSite

X Sagittal SinusAccess Site

POSTERIOR

FIG.1. An illu¡;tration ofthe top surface ofthe skull demonstrat-ing the locations of t he flow probes and pressure monitors relative tothe lesion site. The vertical line in the center of the skull surfacerepresents the ¡;agittal suture and the two oblique liDes intersectingthe sagittal suture just posterior to the sagittal sinus access site repre-sent the coronal suture.

LEFT

(MAP), heart rate, cardiac output (CO), central venouspressure (CVP), ICP, and CBF. Serum sodium was mea-sured on a standard serum chemistry analyzer (EASYST Analyzer, EM Diagnostics, Gibbstown, NJ) andserum osmolarity was determined by vapor pressurechanges (Advanced Wide-Range osmometer Model3W2, Advanced Instruments, Inc., Needham Heights,MA). Fluid intake and urine output were measured andcumulative fluid balance was calculated. The animalswere then injected with 0.5 cc/kg of a 2% Evan's bluesolution to stain afeas of blood-brain barrier disruptionand to allow measurement of lesion size at the time ofpostmortem examination.

The animals were then randomized to one of fourgroups. Control animal s (Group 1, N = 5) were instru-mented and studied under anesthesia for 24 hr to deter-mine the effects of anesthesia on the study variables andto evaluate maintenance fluid requirements. Study ani-mals all had a focal cryogenic brain les ion and a period ofhemorrhagic shock prior to resuscitation with a specificfluid regimen. Following the BL measurements, thebrain lesion was created by applying liquid nitrogen tothe previously exposed inner table of the skull (Fig. 1).The liquid nitrogen was applied for 2 min vía a polyethyl-ene funnel with a 3-cm mouth attached by adhesivecaulk to the 3-cm exposed afea. Residualliquid nitrogenwas aspirated at the end ofthe 2-min periodo Study vari-ables were measured 5 min after the completion of thebrain lesion (T + 5 studyperiod). The animals were hem-orrhaged rapidly to a MAP of 50 mm Hg. The blood wascollected in ACPD bags for preparation as packed cellsto be returned to the animals later in the protocolo Thislevel of hemorrhage was maintained for 45 min by allow-ing the animals to bleed by gravity into a collection bagwhenever their blood pressure rase above 55 mm Hg.

Following measurement of the study variables (H + 45study period), the animals were resuscitated with one ofthree regimens: Group 2 animals (RL/RL, N = 6) re-ceived Ringer's lactate (RL) as an initial bolus (4 ml/kg)followed by constant infusion of RL to return MAP andCVP to BL values. Group 3 animals (HSD/RL, N = 6)received 7.5% hypertonic saline in 6% Dextran 70 as aninitial bolus (4 ml/kg) followed by RL to restore MAPand CVP to BL values. Groups 4 animals (HSD/HSL, N= 6) received 7.5% hypertonic saline in 6% Dextran 70as an initial bolus followed by hypertonic sodium lactateinfligían (HSL) to restore MAP and CVP to BL values.The HSL was prepared by adding sodium chloride andsodium lactate to Ringer's lactate, resulting in a solutionof 250 mEq of sodium per liter (calculated osmolarity500 mOsm/liter). Variables were measured after the ini-tial bolus (RES), then at 30 and 60 min after the initialbolus (R + 30, R + 60, respectively). Shed blood wasreturned as packed red cells after the R + 60 study pe-fiad. The animals were then studied 3, 6, 12, 18, and 24hr after the initial bolus (R + 3H, 6H, 12H, 18H, and

output computer (American Edwards, Model 9520A).Through a low midline laparotomy, the ureters were ex-posed and cannulated with silastic catheters which werebrought out through a separate low midline stab incisionand connected to an ultrasonic urine collection chamber(Vitalmetrics, San Diego, CA). AII incisions were closedin layers and the animal was then carefully turned to theprone position. A midline cranial incision was made toexpose the coronal and sagittal sutures (Fig. 1). An afeaon the left sirle adjacent to the confluence of the sutureswas prepared for a focal cryogenic injury by removingthe outer table of a small afea ofbone (3 cm in diameter)with an engraving drill bit. The inner table of bone wasleft intacto Fiberoptic brain tissue pressure transducers(Camino 420, Camino Labs, San Diego, CA) were in-serted into each hemisphere through 2-mm twist drillholes [11]. Platinum electrodes for measuring regionalcerebral blood flow (CBF) by hydrogen clearance \\'ereplaced into each hemisphere vía 2-mm twist drill holes[12]. On the lesioned sirle (left), the flow probe was in-serted approximately 1 cm away from the afea preparedfor the lesiono The contralateral flow probe was insertedin a stereotactically similar po sitio n (Fig. 1). The hydro-gen flow probes were connected to an amplifier and,along with the transduced hemodynamic pressures andICP's, were recorded on a Gould strip chart recorder(Gould Instruments, Cerritos, CA). Following surgicalpreparation, the animals were left undisturbed on halo-thane and succinylcholine for 60-90 mino Arterial PO2was maintained at >90 mm Hg and PCO2 was kept be-tween 35 and 45 mm Hg during all phases of the experi-ment by ventilator adjustments based on arterial bloodgas analysis.

Study Variables

Following stabilization, the following baseline (BL)measurements were performed: mean arterial pressure

~

~

286 JOURNAL OF SURGICAL RESEARCH: VOL. 50, NO. 3, MARCH 1991

~

sue by the platinum ftow probes. Raw values were, there-fore, converted to percentages of BL.

Differences within groups were evaluated using Stu-dent's t test with Bonferroni correction. Differences be-tween groups were evaluated using ANOV A with correc-tion for multiple comparisons. Significance was attrib-uted to a P value of <0.05.

RESULTS

Hemodynamic VariablEs

I Group 2:~ ~ Group 3: Bolu8 "SD I Group 4: Bolus HSD /lI -.-i;;f~si~;;.Ri..-' Infuslon AL Infuslon HSL I

FIG. 2. A diagram of the experimental protocolo Control animalswere instrurnented only and studied for the 24-hr periodo After braininjury and shock animals were randomized to the experimental groupswhich differed not only in the bolus injection, but algo in the t)'Pe offluid infused to maintain hemodynamic stability after the initial bo-lus.

There was no significant difference between thegroups in CVP (data not shown) or in MAP at BL orfollowing the brain lesion in Groups 2, 3, or 4 (Fig. 3).The MAP was significantly lower after hemorrhage inGroups 2, 3, and 4 compared to BL values and comparedto the control animals (Group 1). There was no signifi-cant difference in the volume of hemorrhage betweengroups (34.7 ml/kg in Group 2, 37.0 ml/kg in Group 3,and 34.8 ml/kg in Group 4). Following the initial HSDbolus, MAP was significantly increased in Groups 3 and4 (P < 0.05). The bolus of RL increased the MAP inGroup 2, but the increase was not significant. After 30min of additional fluid infusion the MAP in all the studygroups had returned to BL and was not significantly dif-ferent from that in the control group. The MAP wasmaintained at or near BL in all groups for the following24 hr and there were no significant differences betweengroups after the initial resuscitation period (Fig. 3).Changes in CVP (data not shown) paralleled changes inMAP. With hemorrhage, CVP decreased significantlyfrom BL but was resto red to BL with resuscitation.

24H, respectively). The study protocol is summarized inFig.2.

At the end of the experiment, the animal was eutha-nized with pentobarbital. The calvarium was promptlyremoved with a casi saw and the brain carefully re-moved. The brain was then weighed and small biopsies(2 mm) were taken of the cortex from the lesioned afeastained by Evan's blue. Mirror image biopsies weretaken from uninjured cortex of the right hemisphere.Similar biopsies were taken from the underlying whitematter of each hemisphere. Relative cerebral water con-tent was determined on each of these samples by mea-suring the specific gravity, using a kerosenejbromoben-zene gradient column [13]. Accurate changes in tissuewater can be determined from tissue-specific gravity andhave been shown to correlate well with drying to con-stant weight [13a].

AII animals were cared for in a University of Califor-nia approved vivarium in a humane fashion by personnelskilled in the care of laboratory animals. The use of ani-mals in ibis study was approved by the University ofCalifornia at San Diego Committee on InvestigationsInvolving Animal Subjects, using guidelines establishedby the National Institutes of Healih.

.--°t ~ G1:CONTROL

o ~ .G2: RURL.G3: HSO/RL

.= ANOVA, t = '1' teat .G4: HSO/tiSL

140

120

100

80

60

40

20

OH eL HE'" RES RJO RJH R6H R12H R24H

mm 9Study Period

FIG. 3. Mean arterial pressure (mm Hg, :tSEM) was signifi-cantly reduced by hemorrhage in the experimental animals (t P < 0.05vs BL, t test; * P < 0.05 vs Control, ANOV A). Bolus injection of HSD

restored the MAP to near baseline values in Groups 3 and 4. Bolusinjection of RL increased MAP in Group 2, but it remained signifi-cantly lower than BL (tP < 0.05, t test) and lower than the othf:r 3groups (* P < 0.05, ANOV Aj. (Study periods: BL, baseline; HEM, im-mediately after hemorrhagf:; RES, immediately after the 4 ml/kg bo-lus; R30, 30 min after RES; R3H, R6H, R12H, and R24H, 3, 6,12,24hr after HEM.)

Data Management and Analysis

Data were recorded on a comprehensive data sheetand entered into a computer spreadsheet for analysis(Microsoft Excel, Microsoft Corp., Redmond, W A).Comparisons were performed using Statview II statisti-cal analysis software (Abacus Concepts, Inc., Berkeley,CA). Results are displayed as the mean :t standard errorof the mean.

Cerebral blood ftow measurements demonstrated vari-ability between animals in all groups due to slightly dif-ferent positions and depths of penetration of brain tis-

WALSH, ZHUANG, AND SHACKFORD: RESUSCITATION OF BRAIN INJURY AND SHOCK 287

FIG. 4. Cardiac output Oiters/min. mean :!: SEM) \Vas signifi-cantly reduced by hemorrhage in the experimental animals (t P < 0.05\s BL. t test: * P < 0.05 \s Control, ANOV A). Bolus injection oí HSD

restored CO to BL in Groups 3 and 4, while bolus injection oí RLresulted in only. a slight increase in Group 2 such that the CO re-mained slgnificantly reduced (tP < 0.05 vs BL, t test; *P < 0.05 vsGroups 1, 3, 4. ANOV A). Continued iníusion oí HSL in Group 4 re-sulted ina CO "hich "as significantly greater than BL (tp < 0.05, ttest) and significantly greater than that in the other groups (* P < 0.05,ANOV A).. (Study. periodo: BL, baseline; HEM, immediately aíter hem-orrhage; RES, immediately aíter the 4 ml/kg bolus; R30, 30 min aíterRES; R3H,R6H, R12H, and R24H, 3, 6, 12, 24, hr aíter HEM.)

groups and significantly greater than BL (P < 0.05). TheICP in Groups 3 and 4 remained significantly lower thanthat in both control and Group 2 animals during thistime. After 60 min of infusion of RL Group 3 animals(HSD/RL) demonstrated a progressive increase in ICPto greater than BL, but the difference was not statisti-cally significant. The ICP in Group 4 (HSD/HSL) algorase but did not exceed BL until6 hr after resuscitation,and the in crease \vas not significant. By 24 hr postre-suscitation, the ICP was higher in Groups 2 and 3 thanin Groups 1 and 4, however, this difference did not reachstatistical significance (P = 0.07). There was moderatevariability in the ICP at 24 hr in the groups that receivedRL (Group 2: 25.8:t 8.8; Group 3: 21.7:t 8.2), but not inthe controls (Group 1: 13.2 :t 2.8) or in the group receiv-ing HSL (Group 4: 13.7 :t 1.5).

CBF in all animals ranged from 25 to 43 ml/IOO g/min(average 32.3 ml/IOO g/min). Because there was variabil-ity between animals due to probe location the valueswere converted to percentages of baseline for analysis(Table 1). The CBF increased initially in the controlanimals (Group 1) because of vasodilation associatedwith halothane [14]. The cryogenic lesion resulted in areduction in CBF to both the lesioned and the nonle-sioned hemispheres in the study animals (Groups 2, 3,and 4). The reduction in CBF was significant on the sideofthe lesion (P < 0.05). The CBF was further reduced byhemorrhage in Groups 2, 3, and 4to approximately 50%of baseline values (Table 1). Bolus fluid infusion in-creased CBF in all animals. The CBF, however, re-mained significantly lower than BL in the RL group

There was no significant difference between the groupsin CVP at any time during the study.

Cardiac out-put was not significantly different be-tween groups prior to hemorrhage (Fig. 4). Groups 2, 3,and 4 demonstrat-ed a significant drop in cardiac outputby the end of hemorrhage, falling to approximately 40%of prehemorrhage levels. The4 mljkg bolus of HSD in-creased the CO significantly in Groups 3 and 4, return-ing it BL within 5 min.The HSDjHSL group (Group 4)had a significantly higher cardiac output than the othergroups for the first 60 min of resuscitation (P < 0.05).After t-he 4 mljkg RL bolus, the CO in Group 2 raseslightly, but t-he increase was not significant. After thisstudy period, there was no difference in CO between thegroups for the balance of the study period (Fig. 4).

C~rebrdl VariablesII ICP was elevated slightly following the cryogenic le-

sibn in the three lesioned groups (Groups 2, 3, and 4);however, hemorrhage was begun immediately followingthe brain lesion so any swelling or edema was accommo-dated by reductions in cerebral blood volume due to hem-orrhage (Fig. 5). The ICP increased in Group 1 duringthis period as a result of cerebral vasodilatation and anassociated increase in cerebral blood volume associatedwith the use of halothane [14]. As a result, the ICP wassignificantly higher in the control group immediatelyafter hemorrhage and the bolos injection. Following re-suscitation, the RL animals (Group 2) demonstrated asignificant increase in ICP such that the lCP in Group 2was significantly greater than that in any of the other

Study Period

FIG.5. lntracranial pressure increased slightly in the controlgroup (Group 1) after BL due to the effects of anesthesia (see text),while ICP fell in the hemorrhaged animals as a result of a decrease incerebral blood volume (tP < 0.05 vs BL, t test). The ICP in Group 2was significantly greater than BL (tp < 0.05, t test) and significantlygreater than that in the other groups after 30 min of resuscitation (* P< 0.05, ANOV A). The ICP Tose steadily in Group 3 after 30 min, whilethe ICP in Group 4 did not rise until after 3 hr. At 24 hr after re suscita-tion the ICP in Groups 2 and 3 was higher than that in controls andGroup 4 but the difference was not statistically significant (P = 0.07,ANOV A). (Study periods: BL, baseline; HEM, immediately after hem-orrhage; RES, immediately after the 4 ml/kg bolus; R30, 30 min afterRES; R3H, R6H, R12H, and R24H, 3, 6, 12,24 hr after HEM.)

288 JOURNAL OF SURGICAL RESEARCH: VOL ;0, NO. MARCH 1991

-Vari

Left (

RightLeft (

RightLeft (

RightLeft C

Right

Group

(Control)

Baseline

T ABLE 1

Cerebral Blood Flow

End of lesion End of hemorrhage Resus + ~lnin-

100.0100.0100.0100.0100.0100.0100.0100.0

113.9 ::!: 20.2.107.8::!: 15.457.4:!: 7.1t56.2:!: 10.0t45.4:!: 5.3t50.6::!: 3.3t51.1::!: 8.4t61.8 ::!: 12.5t

~ + 6H

98.1 :t 13.888.7 :t 10.0

108.9 :t 13.9113.2:t 22.39O.0:t 8.799.0 :t 12.6

128.0:t 12.9145.1 :t 26.5

Resus + 24H

(RL/RL

HSD¡

:HSD, [SI

Resus f-60mint14.1

t 7.6t 26.3!:34..1t 10.5t 13,2!:

11.9:: 12.5t

118.2 j

111.4 j

63.2 j

87.6 :!

62.8 :!

80.0 :!

61.5 i

83.2 :t

Resus H

Re"""

roup

tltrol)

Variable

Left CBFRight CBFLeft CBFRight CBFLeft CBFRight CBFLeft CBFRight CBF

99.9 :

90.6 :

131.5 :126.8 :

82.8 :

97.0 j

127.3 j

140.1 j

80.8:t 8.478.1:t 3.968.4:t 8.0t73.9:t 13.972.2:t 6.2t86.2:t 8.389.3 :t 14.822.6 :t 20.3

Resuscitation

106.7 :t 18.8,101.4 :t 12.368.5:t 8.2t68.1 :t 10.4t83.7:t 11.196.1:t 11.3

126.7 :t 16.7*136.4 :t 28.8

Resus + 12H-

78.4 :t 11.876.7 :t 14.574.6:t 5.8t87.8:t 14.978.2:t 15.4181.8:t 7.0t71.9:t 6.4t106.2

:t 27.3

97.2 :t 16.486.6 :t 14.865.6 :t 15.553.8:t 13.1t36.0:t 9.3t49.8 :t 10.9t65.5:t 9.3t91.2 :t 18.9

92.0:t 11.186.3:t 5.169.4:t 8.9t75.7 :t 13.978.7:t 8.986.7:t 7.276.4 ~ 5.2t

102.7 :t 20.3

2

(RL/Rl

(HSD/RI

4 (HSD/HSI

Note. CBF, percentage of baseline, mean :t SE* P < 0.05, ANOVA.

t P < 0.05 vs BL, t test.

(Group 2). On the other hand, bolus injection of HSDsignificantly increased the CBF in Groups 3 and 4 (P< 0.05), returning it to near BL values (Table 1). CBFwas greatest in Group 4 during this study interval (P< 0.05). Group 2 animal s required additional fiuid to re-turn CBF to BL. Following an initial period of hyper-emia, which occurred in all groups immediately after re-suscitation, CBF decreased on the sirle ofthe lesion in allgroups and was significantly less than BL at 24 hr. Thiswas most pronounced in Group 3. The CBF of the urlin-jured (right) hemisphere was alBo significantly reducedfrom BL at 24 hr in Groups 2 and 3, but not in Group 4(HSD/HSL).

pronounced in the Group 3 animals since they had re-ceived RL after the bolus injection of HSD. Serum os-molarity was highest in Group 4 animals at this time.Continued resuscitation with HSL in the Group 4 ani-mals resulted in a serum osmolarity which was signifi-cantly greater than BL and significantly greater thanthe other groups at 24 hr (P < 0.05, Table 2). The ani-mals receiving RL for the balance of the resuscitation(Groups 2 and 3) had serum osmolarities which werelower than BL but the decrease was not significant.

Changes in serum sodium paralleled those of serumosmolarity. The peak serum sodium in Group 3 was153.8 :t 1.6 mEqJliter. occurring immediately after thebolus injection of HSD. The peak serum sodium inGroup 4 was 165 :t 3.8 mEqJliter, occurring at 24 hr.

At the conclusion of the experiment all animal s werehemodynamically stable with urine outputs which ex-ceeded 1 rill/kg/min. The netfiuid balance in Group 4 atthe end of the study was significantly lower than the netfiuid balance in the other experimental groups (P < 0.05,Fig. 6). In addition, the net fiuid balance in Group 4 wasnot significantly different from that in the controlgroup.

Fluid and Electrolyte Data

Serum osmolarity increased in Groups 3 and 4 within30 min of resuscitation (Table 2). This increase was less

Cerebral Water Canten

After sacrifice and removal of the calvarium the brainwas carefully removed and the afea of Evan's blue stain-ing was measured with calipers on the surface and oncoronal and sagittal sections. In all animals the lesionwas well circumscribed and easily measured (Fig. 7). The

-able

~BFCBF;BFCBF;BFCBF:BF

CBF

!:

21.6*!: 15.5: 9.7t: 9.8: 5.0t: 9.0:

11.3t8.7

""AI,SH, ZHllAN( ANO SHACKFOHO: RESUSCITATION OF BRAIN INJURY AND SHOCK 289

12000 Cortical cerebral water content (CWC) was signifi-cantly higher (lower specific gravity) in the left (le-sioned) hemisphere of the study animals (Groups 2, 3,and 4) than in the controls (P < 0.05). There was nosignificant difference in cortical water content in thenonlesioned hemisphere between the groups (Fig. 8).The water content of the unlesioned white matter con-tained significantly less water than lesioned white mat-ter (data not shown) in Groups 3 and 4 (P < 0.05), whilein Group 2 (RL) there was no significant difference be-tween the lesioned and unlesioneci white matter. Theunlesioned white matter in Group 4 animals containedthe least water, although this difference was not statisti-cally significant.

10000

8000

6000

4000

2000

ml.$:OControl AL/Al HSD/Al HSD/HSl

FIG.6. The net Huid balance (Huid infu;;ed + blood infused)-(urine output + bJood Joss) at t he end of the experiment (24 hr afterresuscitation), The net fluid balance in (-;roup 4 (HSD/HSL) ,,'as sig-nificantly lo,,'er than that in the other experimentaj groups (*P< 0,05, ANOV A) 3nd not signilicantly ditlerent than control (NS. t

test). (Control: Group 1; RL/RL: (-;roup 2; HSD¡RL: Group 3; HSD/HSL: Group 4,)

DISCUSSION

Despite significant improvements in modern traumacare, the combination oí asevere head injury with hemor-rhagic shock continues to have a high mortality rate[15]. While blood pressure can be restored with intrave-nous fluid, the effects of rapid fluid iníusion on an evolv-ing brain injury are poorly understood. The volume oíasanguinous fluid necessary to restore blood pressure

average lesion dimensions were: width 22 :t 4 mm; length32 :t 5 mm; depth 5 :t 2 mm. There ,vas no significantdifference in les ion size between groups.

.~~

290 JOl.¡RNAL OF SURGICAL RESEARCH: VOL. 50, NO. 3, MARCH 1991

o Left CortexANOVA. t = 'l'lest .Right Cortexr~

1.050

1.040 t

1.030 i

~~

AL/AL HSD/RL HSD/HSL

Control

1.020

.s.Q ~--

FIG.8. Water content in ¡he injured (left) and uninjured (right)cerebral corte x determined b,,' measurement of gpecific gravity. Notethat biopsies from the afea of injur;.' contained significantly morewater (lower specific gravit,,') than biopsies from the uninjured hemi-sphere (t P < 0.05. t tegt) and biopsies from the left cerebral cortex ofthe control group (*P < 0.05. ANOVA). (Control: Group 1; RL/RL:Group 2; HSD/RL:Group 3: HSD/HSL: Group 4.)

may lead to ftuid accumulation in the afea of a disruptedblood-brain barrier, with resultant increased ICP, re-duced cerebral perfusion pressure, and a reduction incerebral blood flow leading to secondary brain injury[15). Methods to reduce ICP (e.g., fluid restriction andforced duresis) may have untoward effects on bloodpressure and CBF and may not be possible in patientswith intravascular volume depletion.

Hypertonic fluids have been shown to rapidly expandplasma volume and restore both blood pressure and car-diac output [1-5), but little is known about their effectson ICP and CBF. Shackford et al. [5], in a porcine modelof hemorrhagic shock \vithout brain injury, found thatresuscitation with hypertonic sodium lactate solutionwas associated with a lower ICP than isotonic resuscita-tion with Ringer's lactate. Prough and co-workers [16],using a canine model of hemorrhagic shock, algo found asignificantly lower ICP in animals resuscitated with hy-pertonic saline than in animals resuscitated \vithRinger's lactate. Gunar et al. found that 3% hypertonicsaline resuscitation of hemorrhagic shock and a simu-lated head injury was associated \vith a lower ICP andless brain edema than resuscitation with isotonic ftuid orcolloid [6, 17]. Cerebral blood ftO\V and brain water con-tent were not measured in these studies. Wisner et al.[7], in an avine model of traumatic shock and cryogenicbrain injury, found no difference in either ICP or cere-bral water content bet\\'een animals resuscitated withcolloid and those resuscitated with RL. Zornow and col-leagues [8], in a rabbit model of isovolemic hemodilutionand cryogenic brain injury without shock, found that theICP in animals receiving RL was significantly greaterthan that in those receiving h~'Pertonic sodium lactate.While the cerebral water content in the afea ofthe injurydid not differ between the groups, Zornow et al. found

that the cerebral water content in the uninjured hemi-sphere was significantly lower in the group receiving hy-pertonic sodium lactate, suggesting that dehydration ofnormal brain might be a possible explanation for thelower ICP in the hypertonic group. Wisner et al. [9). in arat model of shock and cryogenic brain injury, algo notedsignificantly reduced cerebral water content in the unin-jured hemisphere of animals resuscitated \,'it h h:--per-tonic saline compared to those resuscitated \\"ith RL.

All of the studies examining resuscitation from shockand brain injury have been of relatively short duration(less than 6 hr). Such short study periods focus only onthe early effects of resuscitation when hemodynamicfunction is optimized, but do not allow an assessment ofcerebral variables when brain swelling and ICP are max-imal at 12-18 hr after injury [10). OUT study invol,'ed a24-hr period of study following brain injury and hemor-rhagic shock and included the measurement of cerebralblood ftow and cerebral water contento

We used a focal cryogenic lesion to compare resuscita-tion regimens. While the cryogenic lesion lacks the rota-tion and shear force s of impact injury, it is very reproduc-ible and results in a focal and uniform disruption of theblood-brain barrier which is similar in nature to thatseen with cerebral contusion [18~. The model is similarin other respects to the clinical situation in which head-injured patients in shock are given asanguinous salt so-lution followed by blood to resto re blood pressure andare often intubated, hyperventilated, sedated, and para-lyzed to control ICP [19). In arder to exaggerate anydetrimental effects that the study solutions might haveon ICP, we chose not to hyperventilate the animals andwe selected an anesthetic agent which produces cerebralvasodilation.

The control group (Group 1) had no brain injury,shock, or resuscitation but was instrumented and moni-tored to evaluate the stability of Qur model over the 24 hrof the experimento The control group demonstratedstable hemodynamics, ICP, and CBF ayer the course ofthe experiment and had no evidence of brain edema orswelling at the completion ofthe study. Group 2 receivedisotonic ftuid (RL) throughout the study period, both forbolus injection and for further resuscitation and mainte-nance of hemodynamic stability and urine output.Group 3 received a hypertonic b01us (HSD) follo,,"ed byan isotonic ftuid to complete the resuscitation and tomaintain blood pressure and urine output. Group 4 re-ceived only hypertonic ftuids, HSD for the bolus injec-tion and HSL to complete the resuscitation.

Hemorrhage to 50 torr resulted in approximately a50% blood volume deficit and a significant decrease inMAP and CO in all treatment groups. The HSD bolus inGroups 3 and 4 significantly improved MAP and CO,while the isotonic bolus in Group 2 did not significantlyimprove either parameter. Continued infusion of RL,however, returned both MAP and CO to BL values after30 min of resuscitation. Continued hypertonic infusion

WALSH, ZHUANG, AND SHACKFORD: RESUSCITATION OF BRAIN INJURY AND SHOCK 291

receiving RL (Groups 2 and 3) flow algo decreased to theuninjured hemisphere. We hypothesize that this reduc-tion in flow was due to edema and swelling which com-pressed the microcirculation in the afea of the flowprobe. Edema was definitely present since the corticalwater content in the afea of the injury was significantlygreater than that in the uninjured cortex in all experi-mental animals. There was, however, no difference inwater content of uninjured cortex between the groups,suggesting that a mechanism other than edema reduc-tion may be responsible for the improved flow to thisafea in the animals resuscitated with HSDjHSL. Thesechanges in flow were most marked in Group 3 and arenot explained by our data. While the reduced flow mayhave been due to endothelial swelling induced by a rapidreduction in serum osmolarity which occurred with infu-sion ofthe RL, we have no data to support this and it willrequire further investigation.

These data suggest that hypertonic resuscitation morerapidly resto res CBF at a lower ICP than isotonic resus-citation in a model of focal cryogenic brain injury andshock. Moreover, improved flow and lower ICP persistfor up to 3 hr in h)rpertonic resuscitated animals. A delayin the onset of raised intracranial pressure would con-ceivably allow adequate time for surgical evacuation ofmass lesions and suggests to us that hypertonic resuscita-tion might prevent or reduce secondary brain injury insuch cases. This study algo demonstrates that signifi-cant changes in both CBF and ICP occur after 4-6 hr ofstudy and suggests that prolonged periods of study arenecessary when evaluating the effects of fluid resuscita-tion on the injured brain.

ACKNOWLEDGMENTS

The authors gratefully acknowledge the assistance of Ms. DaleRathe, Ms. Beni Twitchell, and Ms. Ann Rowan in the preparation ofthe manuscript and tables, and Mr. Anthony Quinn in the preparationof the figures.

REFERENCES

in Group 4 produced a transient rise in CO which wassignificantly greater than BL, while MAP was essen-tially unchanged. This enhancement of CO by hyper-tonic fluid has been observed by others [1-3] and isthought to be due to a combination of rapid volume ex-pansion from the intracellular space, improved cardiaccontractility, and systemic vasodilatation [1-3,20].

Resuscitation and maintenance of blood pressure andorille output requil"ed significantly less fluid in Group 4than in the other experimental groups and resulted in asignificantly lower net fluid balance in the animals re-ceiving only hypertonic fluido Hypertonic resuscitation,however, resulted in significant increases in both serumsodium and serum osmolarity. These increases weregreatest in Group 4 at the completion of the study pe-fiad. Increases of similar magnitude in sodium and os-molarity have been observed in both animals andhumans after hemorrhage and resuscitation with hy-pertonic fluid [4, 21, 22]. Hypernatremia and hyperos-molarity resolved without sequelae in those studies byrenal mechanisms of increased osmolar clearance asso-ciated with a negative free water clearance over the 48 hrfollowing resuscitation.

Cerebral blood flow was significantly reduced by hem-orrhage in all experimental animals. A bolos injection ofHSD rapidly resto red CBF to prehemorrhage valueswithout elevating the ICP. Since cerebral blood volumehad to increase with increased flow, the ICP would beexpected to increase unless the volume ofbrain tissue orthe volume of cerebrospinal fluid was reduced. On thebasis of our previous work (8) and that of others (9), weh~'Pothesize that the hypertonic fluid reduced the brain\.olume by extracting water from uninjured cells acrossan intact blood-brain barrier. This cerebral dehydrationallowed for accommodation of the increased volume dueto reperfusion and edema without increasing the ICP.This is a very important finding since hypotension andreduced CBF associated with head injury and shock canlead to hypoxic damage and cytotoxic edema which canincrease ICP and further reduce cerebral perfusion pres-sure, leading to a true secondary injury ofthe brain [15].The brain is particularly sensitive to hypoxic injurysince it has a high energy requirement and no capabilityof storing substrate. Hence, interruption of oxygen deliv-ery for even short periods (4-6 min) can result in irre-versible ischemic change [23]. Animals resuscitated withRL did not have CBF restored to baseline until 30 minafter resuscitation commenced, at which time ICP rasesignificantly and remained significantly above baselinefor the duration of the study. Animals resuscitated withonly hypertonic fluid had CBF restored rapidly and hadthe ICP maintained at baseline for up to 6 hr, subse-quent to which therewas a gradual rige. At 24 hr, how-ever, ICP in this group (Group 4) was not significantlygreater than baseline.

After a transient period of hyperemia, flow to the in-jured afea deteriorated in all animals. In the animals

1. Velasco, l. T., Pontierri, V., Rocha, E., Silva, M., and Lopes, O.Hyperosmotic NaCl and severe hemorrhagic shock. Amer. J.Physiol. 239: H664-73, 1980.

2. Nakayama, S., Sibley, L., Gunther, R. A., Holcroft, J. W., andKramer, G. C. Small volume resuscitation with hypertonic saline(2400 mOsmfliter during hemorrhagic shock. Circo Shock 13:149-159, 1984.

3. Kramer, G. C., Perron, P. R., Lindsey, C., Ho, H. S., Gunther,R. A., Boyle, \V. A., and Holcroft, J. W. Small-volume resuscita-tion with hypertonic saline dextran solution. Surgery 100: 239-247,1986.

4. Shackford, S. R., Norton, C. H., and Todd, M. M. Renal, cerebraland pulmonary effects of hypertonic resuscitation in a porcinemodel of hemorrhagic shock. Surgery 104: 553-560.

5. Holcroft, J. \V., Vasser, M. J., Turner, J. E., Derlet, R. W., andKramer, G. C. 3% NaCl and 7.5% NaClfDextran 70 in the resus-

~

-,"'-'"~-"'~'i',

292 JOURNAL OF SURGICAL RESEARCH: VOL. 50, NO. 3, MARCH 1991

In H. M. Shapiro (Ed.), Anesthesia, 2nd ed. New York: Chur-chill-Livingston, 1985. Pp 1249-1288.

15. Adams, J. H., Graham, D. l., Scott, G., Parker, L. S., and Doyle,D. Brain damage in fatal non-missile head injury. J. Clin. Pathol.33: 1132-1145, 1980.

16. Prough, D. S., Johnson, J. C., Stump, D. A., Stullken, E. H.,Poole, G. V., and Howard, G. Effects of hypertonic saline versuslactated Ringer's solution on cerebral oxygen transport duringresuscitation from hemorrhage shock. J. Neurosurg. 64: 627-632, 1986.

17. Gunnar, W. P., Jonasson, O., Merlotti, G. J., Stone, J., andBarrett, J. Head injury and hemorrhagic shock: Studies of theblood-brain barrier and intracranial pressure after resuscitation\\.ith normal saline solution, 3% saline solution andDextran-40.Surgery 103: 398-407, 1988.

18. Klatzo, l. Neuropathologic aspects ofbrain edema. J. Neuropath.E.-rp. Neurol. 26: 1-14, 1967.

19. Shackford, S. R., Baxt, W. G., Hoyi;, D. B., Eastman, A. B., Ha-milI, F. N., Knotts, F. B., Virgilio, R. W., and McCardle, M.lmpact of a trauma system on severely injured patients. Arch.Surg. i22: 523-527,1987.

20- Mazzoni, M. C., Borgstrom, P., Arfors, K. E., and lntaglietta, M.Dynamic ftuid redistribution in hyperosmotic resuscitation ofh~"po\'olemic hemorrhage. Amer. J. Physiol. 255: H629-H637,1988.

21. Shackford, S. R., Sise, M. J., Fridlund, P. H., Ro,,'ley, W. R.,Peters, R. M., Virgilio, R. W., and Brimm, J- E. Hypertonic so-dium lactate versus lactated Ringer's solution for intravenousftuid therapy in operations on the abdominal aorta. Surgery 94:41-51, 1983.

22. Shackford, S. R., Fortlage, D. A., Peters, R. M., Fridlund, P. F.,and Sise, M. J. Serum osmolar and electrol~'te changes asso-ciated with large infusions of hypertonic sodium lactate for aor-tic reconstruction. Surg. Gynecol. Obstet. 164: 127-136, 1987.

23. Brierly, J. B., Brown, A. W., Excell, B. J., and Meldrum, B. S.Brain damage in the Rhesus monkey resulting from profoundarterial hypotension. lts nature, distribution, and general physio-logic correlates. Brain Res. 13: 68-100, 1969.

citation oí severely injured patients. Ann. Surg. 206: 279-287,1987.

6. Gunnar, W. P., Merlotti, G. J., Jonasson, O., and Barrett, J. Re-suscitation from hemorrhagic shock: Alterations ofthe intracra-nial pressure after normal saline, 3% saline and Dextran 40. Sur-gery 103: 398-407,1988.

7. Wisner, D., Busche, F., Sturm,J., Gaab, M., and Meyer, H. Trau-matic shock and head injury: Effects of fluid resuscitation on t hebrain. J. Surg. Res. 46: 49-59, 1989.

8. Zornow, M. R., Scheller, M. S., and Shackford, S. R. Effect of ahypertonic !actated Ringer's solution on intracranial pressureand cerebral wat~r content in a model oftraumatic brain injury.J. Trouma 29: 484-488, 1989.

9. Wisner, D. H., Schuster, L., and Quinn, C. Hypertonic resuscita-tion of head injury: Effects on cerebral water contento J. Trauma30: 75-78, 1990.

10. Pitts, L. H., Kaktis, J. V., Juster, R., and Heilbron, D. ICP andoutcome in patients with severe head injury. In K. Shulman, A.Marmarou, J. D. Miller,D. P. Becker, G. M. Hochwald, and M.Brock (Eds.), Intracranial Pressure IV. Berlin: Springer-Verlag,1980. Pp 5-90.

11. Ostrup, R. C., Leurssen, T. G., Marshall, L. F., and Zornow,M. H. Continuous monitoring of intracranial pressure with aminiaturized fiberoptic device. J. Neurosurg. 67: 206-209, 1987.

12. Pasztor, E., Symon, L., Dorsch, N.W., and Branston, N. M. Thehydrogen clearance method in assessment of blood flow in cor-tex, white matter and deep nuclei of baboons. Stroke 4: 556-567,1973.

13. Nelson, S. R., Mantz, M. L., and Maxwell, J. A. Use of specificgravity in the measurement of cerebral edema. J. Appl. Physiol.30: 268-271, 1971.

13a. Shigeno, T., Brock, M., Shigeno, S., Fritschka, E., and Cervos-Nevarro, J. The determination of brain water content: Microgra-vimetry versus drying-weighing method. J. Neurosurg. 57: 99-107, 1982.

14. Shapiro, H. M. Anesthesia effects upon cerebral blood flo\v, cere-bral metabolism, electroencephalogram, and evoked potentials.