JOURNAL OF SURGICAL RESEARCH 66,81-89 (1996)ARTICLE NO. 0376

An Optimized Model for Rat Liver Perfusion Studies

KEE CHEUNG, PH.D., PETER E. HICKMAN, MBBS, PH.D., JULIA M. POTTER, MBBS, PH.D.,NEAL l. W ALKER, MBBS, MEGAN JERICHO, B.Sc., Ross HASLAM, B.Sc., AND MICHAEL S. ROBERTS, PH.D.

Division of Chemical Pathology and University of Queensland Department of Medicine, Princess Alexandra Hospital,Woolloongabba, Queensland 4102, Australia; and University Department of Pathology, University of Queensland

Medical School, Herston, Queensland 4029, Australia

Submitted for publication March 27, 1996

Conditions which influence the viability, integrity,and extraction efficiency ol the isolated perfused ratliver were examined to establish optimal conditionslor subsequent work in reperfusion injury studies in-cluding the choice ol buffer, use ol oncotic agents, he-matocrit, perfusion flow rate, and pressure. Hat liverswere perfused with MOPS-buffered Ringer solutionwith or without erythrocytes. Perfusates were col-lected and analyzed lor blood gases, electrolytes, en-zymes, radioactivity in Mill studies, and lignocaine inextraction studies. Liver tissue was sampled lor histo-logical examinations, and wet:dry weight ol the liverwas also determined. MOPS-buffered Ringer solutionwas lound to be superior to Krebs bicarbonate buffer,in terms ol pH control and buffering capacity, espe-cialIy during any prolonged period ol liver perfusion.A pH ol 7.2 is chosen lor perfusion since this is thephysiological pH ol the portal blood. The presence olalbumin was important as an oncotic agent, particu-larly when erythrocytes were used in the perfusate.Perfusion pressure, resistance, and vascular volumeare flow-dependent and the inclusion ol erythrocytesin the perfusate substantially altered the flow charac-teristics lor perfusion pressure and resistance but notvascular volume. Lignocaine extraction was relativelyflow-independent. Perfusion injury as defined by en-zyme release and tissue fine structure was closely re-lated to the supply ol O2, The optimal conditions lorliver perfusion depend upon an adequate supply ol ox-ygen. This can be achieved by using either erythro-cyte-free perfusate at a flow rate greater than 6 mI/min/g liver or a 20% erythrocyte-containing perfusateat 2 mI/min/g. ~ 1996 Academic Presa, Inc.

INTRODUCTION

The isolated perfused rat liver is a widely used modelfor simulating in vivo conditions, with the advantage ofbeing able to precisely control experimental conditions.However, the optimal conditions for perfusion researchremain ill defined in terms of such considerations asthe inclusion of erythrocytes in the perfusing solutions[1], appropriate flow rates, buffer composition, and the

preferred concentration of albumin [2, 3]. While moststudies report O2 consumption and hile fiow, other mea-sures ofliver function are less well defined and in manycases the viability of the preparations used is not wellestablished.

In this study, we have examined the conditions whichinfiuence the viability, integrity, and extraction effi-ciency ofthe isolated perfused rat liver. The key condi-tions examined were the nature of the buffer, the pres-ence of erythrocytes and albumin, and perfusate fiowrateo Tissue function and viability have been assessedby many separate procedures including histological ex-amination of tissues, O2 consumption and hile fiow,wet:dry tissue weight ratios, enzyme release, perfusionpressure, multiple indicator dilution (MID) studies,and lignocaine extraction.

MA~ AND METHODS

0022-4804/96 $18.00Copyright @ 1996 by Academic Press, Inc.

AIl rights oí reproduction in any íorm reserved.

81

Animals and surgery. Female Sprague-Dawley rats weighingbetween 200 and 275 g were obtained from the Medical School,University of Queensland. AlI animal studies received ethical ap-proval. Animals were fasted for 24 hr before experimentation. Anes-thesia consisted of a single intraperitoneal injection of a mixtureof 2.5 mg xylazine, 12.5 mg ketamine, and 0.5 mg diazepam in atotal volume of 350 ,ul. Five hundred units of heparin was injectedinto the inferior vena cava to prevent blood clotting. The portal veinwas cannulated with a 16-gauge iv catheter and the liver perfusedwith MOPS-buffered Ringer solution at a flow rate of 10 rol/minoThe abdominal inferior vena cava was cut quickly and the flowrate increased to 0.05-0.15 mI/min/g body weight. The chest oftheanimal was opened and the inferior vena cava was cannulated viathe right atrium of the heart using a lO-cm length of polyethylenetubing (Intramedic, Clay Adams, Parsippany, NJ; PE-240, i.d. 1.67mm). The abdominal inferior vena cava was ligated and the perfu-Bate outflow was drained to a waste bottle, thereby flushing theblood from the liver. The hile duct was alBO cannulated using a 5-cm length of polyethylene tubing (PE-10, i.d. 0.28 mm). Bile wascollected into Eppendorf tubes. Liver wet weight ranged from 4.6to 6.8 g and averaged 5.7 :t 0.6 g (mean :t SEM). Erythrocytes werecollected from domestic dogs. The dogs were being destroyed underguidelines approved by the University of Queensland Animal Ex-perimentation Ethics Committee, and the collection procedure wasapproved by the same committee. Blood was collected directly intocollection bags containing citrate (citric acid 3.27 g/liter; sodiumcitrate 26.30 g/liter; Tuta Laboratories, Lane Cave, NSW, Australia)by cardiac puncture. The dog blood was washed three times in phos-phate-buffered saline, with the buffy coat removed after each wash.The washed erythrocytes were then further washed with three

82 JOURNAL OF SURGICAL RESEARCH: VOL. 66, NO. 1, NOVEMBER 1996

PeristalticPump

~GasFlow I

PeñU5ateFlow

FIG. 1. Semidiagrammatic view of the perfusion system. Direction of flow is indicated by arrows.

changes of MOPS buffer (3-[N-morpholino]propanesulfonic acid; The membrane oxygenator generated a partial pressure of O2 ofSigma, Sto Louis, MO) containing 2% bovine serum albumin (BSA). 576:t 9 mm Hg (n = 16). We found that the p02 of the infiowThe pH of fue MOPS buffer was carefully adjusted with NaOH so samples was greatly affected by fue mode of specimen collection. Ifas to achieve the desired pH of 7.2 after the 20% (v/v) erythrocytes the sample was collected by a plastic syringe inserted into the injec-were added to the buffer. The hematocrit of the final suspension tion port, the p02 obtained would be about 50-100 mm Hg lowerwas determined to a value of 15.4 :t 1.6%. than that collected by just allowing the perfusate to drip from the

Buffers. Perfusion media investigated were Krebs-Henseleit port into a collection vial which was then sealed before p02 mea-buffer and MOPS-buffered Ringer solution. Both the Krebs-Hensel- surement. This discrepancy is probably a result of gas exchangeeit buffer [4] and the MOPS-buffered Ringer solution [5] contained turbulence occurring when perfusate was drawn into the syringe.11.1 mM glucose, with or without 2% dialyzed BSA (Sigma; bovine Laboratory workers unaware of this discrepancy could have re-albumin fraction V, Cato No. A- 7906). Dialysis tubing was treated ported underestimated p02S and consequently updervalued O2 sup-with sodium bicarbonate and EDTA before use. Forty grams of BSA ply and O2 consumption (if the infiow sample is taken by syringewas dissolved in 400 mI ofMOPS buffer (25 mM) anddialyzed for 48 and outflow sample by dripping).hr with two changes of 4 liters of MOPS buffer. Preliminary studies The outflow perfusate was recycled to the central reservoir forshowed that dialyzed BSA effectively reduced hemolysis of dog eryth- recirculation, except when MID or lignocaine extraction studiesrocytes during liver perfusion studies. The pH of rat portal venous were performed. In these studies a single-pass perfusion design wasblood in vivo was measured as 7.209 :t 0.005 (n = 4) and all studies employed in which a bolus of an indicator or lignocaine solutionwere therefore performed at pH 7.2. The pH of the Krebs- Henseleit was injected into the portal vein and the outflow collected for analy-buffer was adjusted by the amount ofNaHC03 added. The pH ofthe siso Flow rate was regulated by a variable speed pump drive (Mas-MOPS-buffered Ringer solution was adjusted with 5 N NaOH, before terflex) and maintained constant throughout an experimento Theaddition of erythrocytes as described above. Prior to experiments actual rate was determined manually using measuring cylinder andcommencing, all buffer solutions (with Qr without BSA) were filtered timer.through a 5-Jl.m membrane filter (Millipore, Maidstone, England) to Histological studies. At the conclusion of each experiment, fueremove any potential microemboli [6]. The osmolality of the MOPS liver was completely excised and freed of any extraneous tissue. Mul-buffer used was 288 :t 8 mOsm/kg. tiple small samples were taken from each liver (mean 6.3 mm, range

Perfusion system. The buffer was drawn from its reservoir by a 2-9 mm) and placed in 10% neutral buffered formalin (SEA Trading,peristaltic pump (Masterflex; Cole-Parmer Instrument Co, Chicago, Brisbane, Australia) for histological examination. Sections were ex-IL) and passed in sequence through a bubble trap with in-line filter, amined blind and the extents of apoptosis and necrosis graded sepa-a membrane oxygenator comprising 4 m of coiled silastic tubing (Dow rately (O to 3+). The remaining liver was carefully blotted and theCorning Co, Midland, Michigan; i.d. 1.5 mm), a Y-tube connected to wet weight determined. The liver was then placed in an oyen for 24a manometer, and eventually into the portal vein of the animal hr at 80°C, after which time fue dry weight was determined, and fuethrough a Teflon cannula (Fig. 1). The inclusion of an injection port wet:dry weight ratio was calculated.positioned immediately prior to fue portal vein cannula allowed col- Biochemical studies. pH, p02' pC02, and hematocrit of the col-lection of portal inflow samples. Outflow samples were collected via lected perfusate were determined with an ABL 520 analyzer (Radi-the inferior vena cava cannula. The perfusate was stored in a glass ometer, Copenhagen, Denmark) and total hemoglobin and O2 contentreservoir standing on a magnetic stirrer. The perfusion system was (vol%) were measured with an OSM3 analyzer (Radiometer) op-housed in a perfusion cabinet with temperature and humidity main- erating in fue dog mode. Hepatic oxygen consumption (Jl.mol/min/gtained by a water bath undemeath. liver) was determined by the formula

CHEUNG ET AL.: RAT LIVER PERFUSION MODEL 83

The mean transit time (M1vr) for each solute was calculated fromthe ratio ofthe area under the first moment curve, i.e., outflow con-centration time-time profile (AUMC) to the area under the curve(AUC), i.e., MTr = AUMC/AUC. The volume of distribution for eachmarker was then estimated from the product of MTr and flow rate[8]. The resistance was estimated as the observed perfusion pressurecorrected for catheter effects divided by the observed flow rateo

Lignocaine extraction. Lignocaine extraction across the liver wasused as a viability measure for the efficiency of hepatic metabolism[9]. Lignocaine was given as a bolus injection (approximately 20 p.gin 100 p.l) and the outflow from the liver collected for 5 mino A 10-m! sample was taken from this pooled collection and lignocaine ex-tracted on 1-m! phenyl bond-elute columna (Varian, Harbor City,CA). The lignocaine was eluted with 17.5% (v/v) acetonitrile in phos-phate buffer. Lignocaine was measured by injection onto a reverse-phase HPLC system (Waters, Milford, MA) with detection at 210nro. The lignocaine extraction was estimated as the total amount oflignocaine recovered in the perfusate divided by the dose injected.

Statistical analysis. AlI values are expressed as means :t SEM.Experiments were carried out in triplicate unless stated otherwise.Results were considered significant if P was <0.05 when comparedby ANOVA or Student's t test.

0.0 0.5 1.0 1.5 2.0 2.5

Perfusion Time (hr)

FIG. 2. Changes in perfusate pH dUring prolonged perfusion ofrat livers with MOPS-buffered Ringer solution or Krebs-Henseleitbicarbonate buffer. Isolated rat liver was perfused with MOPS-buf-fered Ringer solution (O) or Krebs-Henseleit bicarbonate buffer (8)for a period of 2 hr. The Krebs-Henseleit buffer was equilibratedwith 5% CO2 and 95% O2 and the MOPS buffer was equilibratedwith 100% O2, The perfusate was not recirculated and pH of theoutflow from the liver was determined at regular intervals. Datashown were obtained from single typical experimento

RESULTS

and oxygen supply (JI,mol/min/g liver) was determined by the formula

inflow O2 contentf1l.. h x ow rateo

lver wet welg t

MOPS buffer was more effective than Krebs-Hensel-eit buffer at controlling perfusate pH as shown in Fig.2. A comparison of various indices of liver function be-tween the two buffer systems is shown in Table 1. Theonly significant difference was in the wet:dry weightratios, which were significantly lower when MOPSbuffer was used.

ln the absence of albumin, wet:dry weight ratios oflivers after 3 hr of perfusion were significantly higherthan in the presence of undialyzed albumin (3.92 :t:0.04 versus 3.23 :t: 0.04; P < 0.001), indicating thedevelopment of significant tissue edema in the former.Hemolytic indices of the perfusate during a 2-hr perfu-sion were algO much higher in the absence of albumin.Dialysis of the albumin before use resulted in a furtherlowering of the hemolytic index as shown in Fig. 3. Itshould be noted that these differences in hemolytic in-dex are associated with significant LD release (>50 U/liter) although no hemolysis is apparent on naked eJeexamination.

Sodium, potassium, aspartate transaminase (AST), and lactate de-hydrogenase (LD) were determined with a Hitachi 747 analyzer (Hi.tachi, Tokyo, Japan). Hemolytic index, which indicates the extentof hemolysis, is automatically calculated on the Hitachi 747 fromdifferential wavelength readings at 408, 505, 570, and 600 nro, andwas routinely determined on all samples containing dog erythrocytes.(A hemolytic index of 100 equates to approximately 1 g of free hemo-gIobin per liter of solution.) Because LD may be released from eithererythrocytes or hepatocytes, the hemolytic index was an integral partof determining the source of any increases in LD activity. Wherenecessary, fue source of LD was followed up by isoenzyme fraction-ation on agarose gel [7].

Multiple indicator dilution studies. MID technique used in thisstudy essentially foIlows the methodology utilized in previous studies[3]. The injectate used to define the vascular, extracellular, and totalwater volumes consisted of [SH]H2O (1 jJ.Ci), ['4C]sucrose (0.2 jJ.Ci),and 125I-BSA (2 jJ.Ci) in 50 jJ.l ofperfusate. The injectates were deliv-ered as rapid boluses via the inflow catheter, and their effiuent wascollected at 1- to 2.5-sec intervals ayer fue initial 70 seco AlI sampleswere stored at 4°C prior to analysis. All outflow samples and appro-priate injectate dilutions were centrifuged and 100 jJ.I of supernatantwas transferred to an Eppendorf tube. Two hundred microliters ofacetonitrile was added and all samples were centrifuged again, priorto samples being taken for counting of gamma radiation (PackardCobra 111 gamma counter; Downers Grove, IL) and beta radiation(1600TR series liquid scintillation counter; Packard). The outflowconcentration for each solute at each collection time was expressedas a fraction ofthe injected clase per milliliter in the perfusate elutingat fue midpoint of fue collection intervalo The total recovery of allsolutes was assessed from fue area under the outflow concentration-time profile (AUC), the measured perfusate flow rate, and fueamount of marker injected (Dose), using the trapezoidal rule forintegration:

TABLE 1

Comparative Indices oí Liver Function in LiversPerfused with either MOPS or Krebs-Henseleit Bu:ffers

Krebs-Henseleit

MOPSbufferParameter

Bile flow (¡;,Vrnin/g)O2 consumption(¡;,moVminlg)LD release (U/g/120 min)AST release (U/g/120 min)Wet:dry weight ratio

0.62 :!: 0.031.9 :!: 0.1

47.5 :!: 7.35.9 :!: 1.63.6 :!: 0.2*

0.57:!: 0.061.8 :!: 0.03

72.4 :!: 18.410.3 :!: 2.64.0 :!: 0.1*

AUCDose

% recovery = x flow rateo

Note. Rat livers were perfused at 3.5-4.5 mI/min/g with the eryth-rocyte-free buffers which contained 2% BSA. The livers were removedfor weight determination at the end of fue perfusion. The values areexpressed as means ::!: SEM (n = 4 for MOPS and n = 3 for Krebs-

Henseleit).*p < 0.05, unpaired t test.

84 JOURNAL OF SURGICAL RESEARCH: VOL. 66, NO. 1, NOVEMBER 1996

250

200

150

100

50

~ 100~

~ Os 80a"O 80-¡¡CIl1'='" 40"-9~ ZO

?. o

iH.,'t1E()

:.:;':>,'OS.,«1

:I:

..-V'---"7

~

v--V

~: ,..-

"-

..,... .---0--0-0-0--0-0

I I , , , ,

10 20 30 40 50 60

Time (min)

FIG. 3. Effect of albumin and dialysis upon the hemolytic index.MOPS-buffered Ringer solution containing noBSA ('\7), 2% undia-lyzed BSA (8), or 2% dialyzed BSA (O) wás pumped through fueperfusion system without the rat liver for a period of 60 mino Thebuffer was sampled at regular intervals for the determination ofhemolytic index. Data shown were obtained from single typical ex-periment.

-1.' 2.0 2.' J.O

Oxygen Consumption (~mol/min/g liver)

FIG. 5. Dependence of enzyme release on O. consumption. (A)Release of LD and (B) release of AST from perfused rat livers as afunction of O. consumption. lsolated rat livers were perfused withMOPS-buffered Ringer solution with (8) or without (O) dog erythro-cytes for 2 hr. Data are plotted as means :t SEM &om a group ofthree animals at each value of O. consumption indicated.

The effect of O2 supply and O2 consumption on vari-ous parameters of liver function is shown in Figs. 4and 5. From Fig. 4 it is apparent that hile flow is verydependent upon O2 supply until a critical value of ap-proximately 3.5 ,umol/min/g of liver is reached, for per-fusion in the presence or absence of erythrocytes. Atthis oxygen supply, the hile flow was 0.93 :t: 0.06 ,ul/min/g liver measured at 1 hr after onset of perfusionand 0.72 :t: 0.06 ,ul/min/g at 2 hr. The average hile flowayer the 2-hr perfusion period was 0.82 :t: 0.10 ,ul/-min/g.

Similarly, wet:dry weight ratios are higher at O2 sup-ply values below 3.5 ,umol/min/g liver for perfusion inthe absence of erythrocytes, though the differences are

A

-+-

; 1.2~~0-~~ 0.8" ~--.~ SIQ", 0.4

.3

¡.,

~

l-.+-

-J.c-

15-'--204.4 r

as1':...~>...Q:,;"~

B

~

not statistically significant (at O2 supply of 1.7 Ji,moVmin/g the wet:dry weight ratio was 3.87 :!:: 0.25, and atO2 supply of 4.7 Ji,moVmin/g the ratio was 3.43 :!:: 0.13;P > 0.2). For perfusion in the presence of erythrocytes,the wet:dry weight ratio stayed low at approximately3.3-3.5. An O2 supply of 3.5 Ji,moVmin/g of liver corre-sponds to an O2 consumption of 2.3 Ji,moVmin/g in theabsence of erythrocytes and appears to be the criticalO2 requirement for healthy liver tissue from fasted ani-mals. This is confirmed in Fig. 5 which shows the en-zyme release at different O2 consumptions, in the pres-ence or absence of erythrocytes. Oxygen consumptionwas matched by adjusting flow rates. There was effec-tively no enzyme released when erythrocytes werepresent (minimum O2 consumption possible was 1.9Ji,moVmin/g, due to flow considerations), though en-zyme release was observed in the absence of erythro-cytes up to and including an O2 consumption of 2.3Ji,moVmin/g. When the O2 consumption was identical(2.3 :!:: 0.1 Ji,moVmin/g), there was significantly less LDenzyme released when erythrocytes were present (inthe presence of erythrocytes 19.5 :!:: 3.5 U/gi120 min;in the absence of erythrocytes 38.3 :!:: 4.3 U/gi120 min;n = 3, P < 0.03). Thus it appears that erythrocytesoffer Borne protection from damage that is separatefrom their ability to provide adequate O2.

Histological assessment was performed on alllivers.N ecrosis occurred to varying degrees in a1l animals inthe group with lowest oxygen consumption, but not inother groups. Compare Figs. 6a and 6b which showhistology of liver after a 2-hr infusion at an O2 con-sumption of 2.8 Ji,moVmin/g liver, in the presence ofalbumin but without erythrocytes, and Figs. 6c and 6dwhich show a section of liver 2 hr after an infusion atan O2 consumption of 1.3 Ji,moVmin/g liver, algo in thepresence of albumin but in the absence of erythrocytes.In the former experiment LD release was 18.2 U/gi120min, AST release was 2.4 U/gi120 min, and appearanceon histology was normal. In the latter, LD release was

-+- --++o~

5 10 15 20

Oxygen Supply (IJ.mol/min/g liver)

FIG. 4. Dependence of hile flow and wet;dry tissue weight ratioon oxygen supply. (A) Change in hile flow and (B) change in wet:dryweight ratio as a function of oxygen supply. Isolated rat livers wereperfused with MOPS-buffered Ringer solution with (.) or without(O) dog erythrocytes for 2 hr. The different levels of oxygen supplydetermined were a result of perfusing the livers of different animal sat different flow rates or different hematocrits. Data are plotted asmeans :t SEM from a group ofthree animals at each value of oxygensupply indicated.

FIG. 6. Histological comparison of tissue perfused in the absence of erythrocytes, above and below the critical O2 consumption of 2.3JLmol/min/g. (a) and (b) Histologically nonnalliver (H&E, original magnüication a x130, b x165). Oxygen consumption was 2.8 JLmol/min/g liver, LD release was 18.2 U/l/120 min, and AST release was 2.4 U/1/120 mino (c) and (d) Early atinar zone 2 and 3 necrosis manifestedby altered staining, altered size and shape, and fine cytoplasmic vacuolation ofhepatocytes (H&E, original magnmcation c x110, d x175).Oxygen consumption was 1.3 JLmol/min/g liver, LD release was 112.6 U/1/120 min, and AST release was 19.0 U/l/120 mino Arrowheads,portal tracts; arrows, tenninal hepatic venules.

85

86 JOURNAL OF SURGICAL RESEARCH: VOL. 66, NO. 1, NOVEMBER 1996

Q¡~~ 2e.,.,

:: ~ 20~~§ S le

.; e~ ---10~~Q¡ 5~

...

1.

change in pressure, resistance, or distribution volumesayer the 2 hr of perfusion. A slight increase in pressurewas observed in the third hour (not shown). There is adirect, almost linear relationship, between pressureand flow for both e¡ythrocyte-containing and erythro-cyte-free perfusate with a slightly higher pressure be-ing observed for perfusate containing erythrocytes (Fig.7 A). As a consequence, fue resistance is relatively unaf-fected by the flow rate used (Fig. 7B). Consistent withfue flow dependence of pressure, the vascular volumes,as defined by BSA, in Fig. 7C algO show an apparentlylinear increase with a changing flow. Figure 7C algOshows that the distribution of sucrose and water in theperfused liver are linearly related to flow rateo Thedistribution of these solutes relative to albumin wereindependent of flow rate suggesting that no edema waspresento Figure 7D shows that lignocaine extraction isenhanced by the presence of erythrocytes in the perfu-Bate at flow rates less than or equal to 4 mJ/min/g oftissue (greater than 90% extraction). There was no dif-ference between the two perfusates at a flow rate of 6mJ/min.

ft~

A

J

l:=---"~'-- ~~ I~ -;l;:t

~

GICJdos...01

";¡GI

~

B

8 9~o~:1.o"í::.....:a~oG)

8:1o>~o~()G1-...HG)

G)

~"~()o~~~

DISCUSSION

..1-

FIG. 7. Dependence of perfusion pressure, resistance, volume ofdistribution, and lignocaine extraction on fiow rateo (A) Perfusionpressure, (B) resistance, and (D) lignocaine extraction as a functionof fiow rateo Isolated rat livers were perfused with MOPS-bufferedRinger solution with (8) or without (O) dog erythrocytes for 2 hr.Perfusion pressure was determined &om fue height of perfusate re-corded on fue manometer attached to the perfusion system. It isexpressed in mm Hg assuming the speclftc gravity of perfusate isthe same as water and fue specific gravity of mercury is 13.534. (C)Volumes ofdistribution as a function offiow late. The indicators usedincluded [~]H2O in the presence (8) and absence (O) oferythrocytes,r.C]sucrose in fue presence (T) and absence ('1) of erythrocytes, and126I-BSA in the presence (O) and absence (.) of erythrocytes. Flowrate was regulated by a variable speed pump drive and maintainedconstant throughout an experimento Data are plotted as means :tSEM &om a group of three animals at each value of fiow rate indi-cated.

The isolated perfused liver has a limited viabilityeven when studies are performed under optimal condi-tions. Under usual conditions it may be maintained for,¡approximately 3-4 hr [2, 3]. Nevertheless, ifmeaning-fuI biological questions are to be posed and answered,as near optimal conditions as possible must be utilized.It must algo be remembered that damage to the livermay not manifest for Borne time (for example, enzymerelease from an anoxic liver may not eventuate fornearly an hour after perfusion has commenced) and tob~valid either studies must be performed for sufficienttime for damage to be able to manifest or initial studiesmust demonstrate the validity ofthe experimental con-ditions.

A variety of criteria have been used to define thefunctioning and viability ofisolated perfused rat livers.These criteria include O2 consumption, gluconeogene-gis, redox state, hile flow, gross morphological appear-ance of the liver, clearance of carbon particles, and drugextraction [2, 3]. While Borne studies haye examinedthe role of variables such as hematocrit [10], we arenot aware of any study which has attempted to interre-late liver integrity, physiology, O2 and drug extractionefficiency, erythrocyte presence, and perfusate flowrateo

It is generally agreed that the most important agentsin determining tissue viability are the perfusate compo-sitian and 02-carrying capacity [2]. Hardison and Wood[11] suggested that bicarbonate, present in Krebs-Henseleit buffer is required to maintain the hile salt-independent fraction of hile flow. Our study, however,shows that the MOPS buffer which contains no bicar-bonate is at least as effective as, if not superior to,Krebs-Henseleit buffer in supporting hile flow andother viability índices (Table 1). Gores et al. [2] algO

112.6 U/gi120 min, AST release was 19.0 U/gi120 min,and histology showed early acinar zone 2 and 3 necro-gis, reflecting more severe cell injury at low levels ofO2 consumption.

The effect of changes in portal flow rate on hepaticcirculation is shown in Fig. 7. There was no significant

e 4"QD

= J

"8QD 2

:x=

S 1S

"8

---~

2.0

1.5

1.0

0.5

100

80

6

4-

2

CHEUNG ET AL.: RAT LIVER PERFUSION MODEL 87

suggested that the Krebs- Henseleit buffer maintainsa stable pH when equilibrated with a 95% Oz :5% COzatmosphere. We have found that fue perfusate pH in-creased with time when using Krebs-Henseleit buffer(as illustrated in Fig. 2). Rat portal blood has a pH of7.2, thus the pKa of MOPS buffer of 7.2 makes it anideal buffer for liver perfusion studies. We found thatthere was less change in pH of the perfusate pool (Fig.2) and less tissue edema developed when MOPS wasused instead of Krebs- Henseleit buffer (Table 1). Mis-chinger et al. [12] had reported falling pH over a 3-hr perfusion period with the Krebs-Henseleit buffer,whereas Schmucker and Curtis [13] found the pH ofKrebs-Ringer bicarbonate perfusate to increase dur-ing aerobic perfusion. OuT result indicated an initialfall in pH followed by gradual rige. Control of perfusatepH is of critical importance as it has been shown whenstudying the development of reperfusion injury in iso-lated perfused liver that changes in pH are closely re-lated to fue degree of injury observed [14]. We havechosen a pH of 7.2 for OuT perfusion experiments sincethis is the physiological pH ofrat portal blood we deter-mined. We feel that such a pH is more appropriate thana pH of7.4 used by most workers in this field [15-17].

Isolated organs have a time-dependent tendency toabsorb water, as with a relatively protein-free medium,water gradually escapes from the vascular space andinterstitial edema develops [6]. In OUT recirculating sys-tem, we found that presence of dialyzed BSA was im-portant. In the absence of BSA, significant edema de-veloped as assessed by wet:dry weight ratios. In addi-tion, significant hemolysis occurred as assessed by thehemolytic index. This is consistentwith the findings ofHems et al. [15] who reported thatomission ofalbuminfrom the perfusion medium caused gross swelling ofthe liver and exudation of Huid from the surface, butvarying the albumin concentration between 1.5 and 6%made no difference. Rosini et al. [18] algO found thatin the absence of an oncotic agent, release of cytosolicenzymes into the perfusate increased as the oncoticpressure decreased. We had similar findings with BSAwhen LD and AST were measured (data not shown).OuT results are at variance with those of Mischingeret al. [12] who found increased tissue damage in thepresence of BSA. They apparently did not dialyze theiralbumin before use and we found that dialysis reducedthe extent of hemolysis further. There is certainlybatch to batch variation of albumin and failure to dia-lyze may have contributed to the discrepancy betweenresults.

The values of wet:dry weight ratio obtained in thepresent study (Table 1, Fig. 4) compare well with thoseof Exton and Park [19] who found that in fasted ratsthe ratio was 3.78 before perfu~ion and 3.85 after perfu-sion for 2 hr. The hile How values obtained are algocomparable to those reported in the literature [2, 3].The observed dependence ofbile How on oxygen supply(Fig. 4) conflrms that reported by Schmucker and Cur-tis [13].

The Oz consumption consistent with viable tissue

varies depending upon whether animals are red orfasted. For red animal s it has been suggested that O2consumption should be greater than 2.0 j.lmol/min/gliver weight [2]. For fasted animals O2 consumptionmay be as low as 1.1 j.lmol/min/g liver. Our data indi-cate that in the absence of erythrocytes, an O2 con-sumption of approximately 2.3 j.lmol/min/g liver is thecritical O2 requirement for healthy liver tissue fromfasting animals. However, in the presence of erythro-cytes, little enzyme release occurred at O2 consumptionas low as 1.9 j.lmol/min/g liver. This study indicatesthat erythrocytes are doing something in addition tosimply delivering O2 and that in some manner theyameliorate the damage observed at that O2 consump-tion rateo The exact nature of this beneficial effect isunknown but it has been postulated that when erythro-cytes are absent from the perfusate, endothelial cellsmay change their shape, grow larger within the lumen,and thus create an obstacle to the free perfusion ofsinusoids [20].

We have algO studied the relationship between hema-tocrit and hepatic oxygen consumption (data notshown). The results obtained were similar to those re-ported by Riedel et al. [10], i.e., increasing the hemato-crit of the perfusate increased the uptake of oxygen ata given flow rateo

llitimately, tissue viability is judged by the goldstandard of histological appearance [13, 21]. We havedemonstrated a relationship between histological ap-pearance of damage, hepatocyte enzyme release (LDand AST), and O2 consumption rates. Our findingsagree with those of Hanks et al. [22] who observedthat perfusion of rat livers with 10% erythrocytes wasassociated with enhanced O2 delivery and less cellularinjury compared to perfusion with cell-free buffer solu-tion. Schmucker and Curtis [13] algo reported that thefine structure ofhepatocytes was best preserved in liv-ers perfused with dilute blood in comparison to perfu-sion with hemoglobin-free medium.

In addition to these basic indices of tissue viability,we have studied several other useful technical proce-dures in this current work. The availability of the he-molytic index is an important technical advance instudies ofthe perfused liver when erythrocytes are con-tained in the perfusate. The potential for confusion asto the source of LD is removed by use of this index. Inaddition, subtle changes such as those indicated above,i.e., the effect of albumin and dialysis on erythrocytehemolysis, can only be identified by means ofthe hemo-lytic index.

The relationship between perfusion pressure and re-sistance with perfusate flow rate is similar to that re-ported by Le Couteur et al. [23] for erythrocyte-freeperfusate. The addition of erythrocytes increases boththe perfusion pressure and the resistance (Figs. 7 A and7B). The portal vein pressures observed in this workare less than the 8 to 10 mm Hg reported for the liver[2] and may result in an uneven distribution of perfu-sate in the liver. However, injection of Evans blue intothe liver using albumin-free perfusate demonstrates an

~-

88 JOURNAL OF SURGICAL RESEARCH: VOL. 66, NO. 1, NOVEMBER 1996

:REFERENCESBallet, F. A., and Thurman, R. G. Why fue perfused liver? InF. Ballet and R. Thurman (Eds.), Pe¡fused Liver: Clinical andBasic Applications. London: Libbey, 1991. Pp. 1-20.

-. d t fum . h -f 13 3. Wolkoff, A. w., Johansen, K L., and Goeser, T. The isolated rat !lver system lS reqwre O lS a pressure O

rfus d at l .

P t .

d l . t .

Anal B . h..pe e r lver: repara Ion an app lca Ion. .lOC em. cm H2O (10 mm Hg), the cntlcal pressure for main- 167: 1, 1987.

taining opening of the vasculature. A parabolic de- 4. Krebs, H. A., and Henseleit, K Untersuchungen über die Harn-

crease in volume with flow may be observed below the stoflbildung im Tierkorper. Hoppe-Seyler's Z. Physiol. Chem.

critical pressure [22]. The parallel total waterand vas- 210: 33, 1932.

cular volume-flow rate relationships (Fig. 7C) are con- 5. Blanchard, J. S. Buffers for enzymes. Methods Enzymol. 104:

sistent with the known linear relationships between 461, 1984.

the hepatic blood volume and intrahepatic pressure 6. Yeh, T., and Wechsler, A. S. The isolated organ in research. In

[25] and minimal edema. A divergence in the relation- H. Tr?idl, W. O. Spitzer, an~ B. McPee~ (Eds.), P;inc.iples and

.Practlce of Research: Strategres for Surglcal Investlgatlons. N ew

ships WIth flow rate would have been anticipated if a York: Springer-Verlag, 1991. pp. 354-365.

significant fluid shift into hepatocytes had occurred. 7. Preston, J. A., Briere, R. O., and Batsaker, J. A. Rapid electro-

The relatively constant wet:dry weight ratio with in- phoretic separation oflactate dehydrogenase isoenzymes on cel-

creasing flow rate is algO cons,istent with minimal lulose acetate. Am. J. Clin. Pathol. 43: 256, 1965.

edema formation. Figure 7D shows tliat the extraction 8. Roberts, M. S., Fraser, S., Wagner, A., and McLeod, L. Resi-

of lignocaine is dependent on erythrocyte presence at de~ce tim.e dist~ibutions of solutes. in ~h~ pe.rrused rat liver

b 1 4 mI/ .nI fl t Th 1 t t .f usmg a disperslon model of hepatlc elimmatlon. I.. Effect of or e ow mI g ow ra e. e ower ex rac Ion o. '.

.'. '. changes m perfusate flow and albumIn concentraban on sucrose

lIgnocame m the absence of erythrocytes lS conslstent and taurocholate. J. Pharmacokinet. Biopharm. 18: 209, 1990.

with an impaired metabolism of lignocailie in fue hyp- 9. Pang, K S., and Rowland, M. Hepatic clearance of drugs. n.

oxic conditions presento Pang and Rowland [9] have Experimental evidence for acceptance of the "well-stirred"

suggested that a high lignocaine extraction is support- model over the "parallel tube" model using lidocaine in the per-

ive of good liver viability. Keiding el al. [26] have algo fused rat liver in situ preparation. J. Pharmacokinet. Biopharm.

emphasized that flow rate and hematocrit must be kept 5: 655, 1977.

b rt. li .t. t b li t d' .thth -10. Riedel,G.L.,Schelle,J.L.,Shepherd,A.P.,andWard,W.F.

tia oye li ~e am mI s m me a O c s u les WI e per Effects of hematocrit on oxygenation of the isolated perfused

used ver. rat liver. Am. J. Physiol. 245: G769, 1983.

In conclusion, the optimal conditions for liver perfu- 11. Hardison, W. G. M., and Wood, C. A. Importance ofbicarbonate

sion depend upon an adequate flow rate and hematocrit in hile salt independent fraction of hile flow. Am. J. Physiol.

to exceed the critical perfusion pressure and to provide 235: E158, 1978.

sufficieilt oxygenation. Gross liver appearance, pres- 12. Mischinger, H. J., Walsh, T. R., Liu, T., Rao, P. N., Rubin, R.,

sure and volumeassessments hile flow rate detennina- Nakamura, K., Todd, S., and Starzl, T. E. An improved tech-

t '

1 t.d ' .ht t .

O xt Dique for isolated perfusion of rat livers and an evaluation of

~on, enzyme re ease, we .ry. welg r.a lOS, .2 e rac- perfusates. J. Burgo Res. 53: 158, 1992.

tlon,. and.lignocaine .ex~ra~t~on proVIde ~VIden~e of 13. Schmucker, D. L., and Curtis, J. C. A correlated study of the

phYSlOlOglcal metabolIc VIabllity. COnfirmatlon ofmad- fine structure and physiology of the perfused rat liver. Lab.

equate conditions is provided by histological assess- Invest. 30: 201, 1974.

mento This study has shown that MOPS is superior to 14. Currm,R. T., Gores, G. T., Thurman,R. G.,andLemasters,J. J.

Krebs- Henseleit as a perfusion buffer and that fue Protection by acidotic pH against anoxic cell killing in perfused

optimal condition for the use of this buffer is at an rat liver: Evidence for a pH paradox. FASEB J. 5: 207,1991.

O2 consumption exceeding 2.3 ¡.tmoI/minlg liver when 15. Hems?~., Ross, B. D., Berry,.M. N.,.and Krebs, H. A. Gluconeo-

livers from fasting animals are used. The O2 consump- genesls m the perfused rat liver. Blochem. J. 101: 284, 196~.

tion is achieved using either erythrocyte-free perfusate 16. Thurman,.R. G.,. ~pel, E.?, and ~m:m~ster.s, J. J. Prote~tlve

..effect of mtrendipme agalnst hYPOXIC mJury m perfused hvers

at a flow rate greater than 6 mI/mlnlg lIver or a 20% from ethanol-treated rats. J. Cardiovasc. Pharm. 12(Suppl. 4):

erythrocyte-containing perfusate at 2 mI/min/g liver. It S113,1988.

appears that ifthese conditions are met, al1 physiologi- 17. Jaeschke, H., Smith, C. V., and Mitchell, J. R. Reactive oxygen

cal, biochemical; and histological criteria for the assess- species during ischaemic-reflow injury in isolated perfused rat

ment of liver viability are fulfilled. liver. J. Clin. Invest. 81: 1240, 1988.

18. Rosini, S., Benetti, D., and Kvetina, J. The functional capacity

ACKNOWLEDGMENTS of the isolated perfused rat liver in relation to fue colloidal-

The authors acknowledge the support of the Princess Alexandra osmotic composition of the perfusion medium. Farm. Pract. 31:

Hospital Research and Development Foundation, fue Royal Chil- 625, 1976.

dren's Hospital. Liver Transplant :rrust gran~ and t~e National 19. Exton,J. H., andPark, C. R. Control ofgluconeogenesis in liver.

He.8It.h and Medlcal R.esearch Counc~ of AustralIa. .Studie~ on hemo- l. General features of gluconeogenesis in fue perfused livers of

lytlC mdex were carned.out by Chns Corcoran; hght mlcrogr~phs rats. J. Biol. Chem. 242: 2622, 1967.

were prepared by Clay Wmterford, Department ofPathology, Umver- P M Ka ..1 d S t M H d..

sity of Queensland' and lignocaine determinations were carried out 20. Huet, .., SSISSla,., an emre, .aemo ynamlcs m

b B tt C M Whi ~ D...f Ch ~~

a l P

a t

ho lo

gy Princess perfused rat liver. In F. Ballet and R. Thurman (Eds.), Perfused y re .c nney, IVlSlon o e~c,

Cl .. l d B . A l..L d L .

bbe 1991Al dr H . tal Liver: lnlca an aSlC pp lcanons. on on: 1 y, . exan a OSpl .pp. 33-42.

21. Lemasters, J. J., and Thurman, R. G. Hypoxic, ischaemic and

preservation-induced injury in fue liver. In F. Ballet and R.

Thurman (Eds.), Perfused Liver: Clinical and Basic Applica-

nona. London: Libbey, 1991. pp. 97-120.

CHEUNG ET AL.: RAT LIVER PERFUSION MODEL 8922. Hanks, J. B., Meyers, W. C., Wellman, C. L., Hill, R. C., and

Jones, R. S. The effect of cell-&ee and erythrocyte-containingperfusion in rat livers. J. Surg. Res. 29: 149, 1980.

23. Le Couteur, D. G., Rivory, L. P., Roberts, M. S., andPond, S. M.Aging and fue response of the isolated perfused rat liver tovasoactive drugs. Biochem. Pharmacol. 43: 913, 1992.

24. Pang, K S., Lee, W. F., Cherry, W. F., Yuen, V., Accaputo, J.,Fayz, S., Schwab, A. J., and Goresky, C. A. Effects ofperfusateflow rate on measured blood volume, disse space, intracellular

water space, and drug extraction in the perfused rat liver prepa-ration: Characterization by the multiple indicator dilution tech-nique. J. Pharmacokinet. Biopharm. 16: 595, 1988.

25. Greenway, C. V., and Lautt, W. W. Hepatic circulation. In S. G.Schultz CEd.), Handbook of Physiology: The GastrointestinalSystem. Bethesda, MD: Am. Physiol. Soc., 1989. Sect. 6, Vol. 2,Chap. 41, pp. 1519-1564.

26. Keiding, S., Vilstrup, H., and Hansen, L. Importance of flowand haematocrit for metabolic function of perfused rat liver.Scand. J. Clin. Lab. Inuest. 40: 355, 1980.