Mechanisms of Ischemic Injury Are Different in theSteatotic and Normal Rat Liver

MARKUS SELZNER,1,3 HANNES A. RUDIGER,1,3 DAVID SINDRAM,1 JOHN MADDEN,2 AND PIERRE-ALAIN CLAVIEN1,3

Hepatic steatosis is associated with significant morbidityand mortality after liver resection and transplantation. Al-though apoptosis is a key mechanism of reperfusion injuryin the normal liver, the pathway leading to cell death insteatotic hepatocytes is unknown. A model of hepatic is-chemia and reperfusion injury in fatty and lean Zucker ratswas used. Fatty animals had increased aspartate amino-transferase (AST) release and decreased survival after 60minutes of ischemia compared with lean animals. Apoptosiswas the predominant form of cell death in the lean rats(82%), whereas necrosis was minimal. In contrast, fatty an-imals developed only moderate amounts of apoptosis butshowed massive necrosis (73%) after 24 hours of reperfu-sion. Intracellular mediators of apoptosis, such as caspase 8,caspase 3, and cytochrome c, were significantly lower in thesteatotic than in the lean liver indicating dysfunction inactivation of the apoptotic pathway. The high percentage ofnecrosis in the steatotic rats was associated with renal acutetubular necrosis after 24 hours of reperfusion in the fatty,but not in lean rats. Caspase inhibition significantly de-creased reperfusion injury in lean animals, but was ineffec-tive in fatty animals. The results indicate that the increasedsusceptibility of fatty livers to reperfusion injury is associ-ated with a change from an apoptotic form of cell death tonecrosis. We conclude that new therapeutic strategies arenecessary in the fatty liver. (HEPATOLOGY 2000;32:1280-1288.)

Hepatic steatosis is a common problem in industrializedcountries, found in 6% to 11% of autopsies.1,2 It can be causedby obesity, ethanol toxicity, or a variety of metabolic disor-ders.3,4 Hepatic steatosis is a major risk factor after liver sur-gery. For example, steatosis is associated with an operativemortality exceeding 14% following major liver resection.5This figure is less than 2% in the normal liver.6 In addition, theuse of steatotic livers for transplantation is associated with anincreased risk for primary nonfunction or dysfunction after

surgery, and most centers will not consider transplantation ifthe donor organ contains more than 50% steatosis.7

Apoptosis has recently been identified as a key mechanismof reperfusion injury in the liver.8-10 The degree of hepatocel-lular apoptosis after normothermic ischemia correlates withmarkers of hepatic injury, including aspartate aminotransfer-ase (AST) release and animal survival. The apoptotic pathwayafter ischemia and reperfusion is mediated by activation of theintracellular cascade of caspases and the release of cyto-chrome c.11,12 Caspase inhibition protects against reperfusioninjury with a decrease of AST release and improved animalsurvival.10,12 Similarly, up-regulation of antiapoptotic media-tors, like Bcl-2, completely inhibits apoptosis and increasesanimal survival after ischemia and reperfusion.13,14

Although accumulating evidence suggests that the fattyliver tolerates an ischemic insult poorly,5,15-17 the mecha-nisms of injury in steatotic livers remains unknown. For ex-ample, it is unclear whether apoptosis is also a feature ofischemic injury in the fatty liver. The identification of specificmechanisms of ischemic injury in the fatty liver may open newtherapeutic avenues with important clinical application.

Zucker rats are a well-characterized model of nutritionallyinduced obesity, which closely simulates the most commoncause of steatosis in the western hemisphere.18 HomozygousZucker rats lack the cerebral leptin receptor and develop obe-sity at the age of 8 weeks because of markedly increased foodintake and decreased energy expenditure. In contrast, het-erozygous Zucker rats have leptin receptors and maintain alean phenotype throughout life. In addition, steatosis inZucker rats is not associated with inflammation, which is animportant shortcoming in other models of steatosis using eth-anol ingestion or a choline-deficient diet.16,19

The aim of this study was to evaluate the mechanisms ofreperfusion injury in the steatotic liver. Experiments weredesigned to compare injury in lean and obese rats, in partic-ular with respect to the mechanism of cell death, and to de-termine whether mediators of apoptosis involved in the is-chemic normal liver are active in fatty livers. Furthermore, weevaluated whether protective strategies effective in normallivers are also effective in the presence of steatosis, and ifdistant organ injury occurs after reperfusion injury in fattyanimals.

MATERIALS AND METHODS

Animals. All experiments were performed in male Zucker rats(Harlan, Indianapolis, IN). Homozygous and heterozygous Zuckerrats aged 14 weeks were used for the experiments as previouslydescribed.20 Animals were fed a laboratory diet containing 12% fat,28% protein, and 60% carbohydrates (5001 rodent diet, PMI Inc,Brentwood, MO) with water and food ad libitum until use and werekept under constant environmental conditions with a 12-hour light-

Abbreviations: AST, aspartate aminotransferase; TUNEL, TdT-mediated dUTP-digoxigenin nick-end labeling; HE, hematoxylin-eosin; PMN, polymorphonuclear leu-kocytes; ANOVA, analysis of variance.

From the Laboratory of Liver Transplantation and Hepatobiliary Surgery, Departmentof 1Surgery and 2Pathology, Duke University Medical Center, Durham, NC; 3Departmentof Visceral and Transplant Surgery, Zurich University Hospital, Zurich, Switzerland.

Received March 28, 2000; accepted September 22, 2000.Supported by NIH-Grant DK54048-01A1.Address reprint requests to: Pierre-Alain Clavien, M.D., Ph.D., F.A.C.S., Professor of

Surgery, Department of Visceral and Transplant Surgery, University of Zurich, Ramistr.100, 8091 Zurich, Switzerland. E-mail: clavien@chir.unizh.ch.

Copyright © 2000 by the American Association for the Study of Liver Diseases.0270-9139/00/3206-0014$3.00/0doi:10.1053/jhep.2000.20528

1280

dark cycle (light 7:00 AM to 7:00 PM). All surgical procedures wereperformed under aseptic conditions between 7:00 AM and 11 AM toavoid circadian variations. Animals received humane care in compli-ance with the Duke University Institutional Animal Care & UseCommittee.

Reagents. The caspase inhibitor IDN 1965 was provided by IDUNPharmaceutical Incorporation (La Jolla, CA).

Hepatic Ischemia. A model of segmental (70%) hepatic ischemiawas used. Rats were anesthetized with inhalation of isoflurane (Pitt-man-Moore, Chicago, IL). After a midline laparotomy, all structuresin the portal triad (hepatic artery, portal vein, and bile duct) to theleft and median liver lobes were occluded for the period of ischemiaunder study. This method of partial hepatic ischemia prevented mes-enteric venous congestion by permitting portal decompressionthrough the right and caudate lobes. Reperfusion was initiated byremoval of the clamp. Animals were again anesthetized at differentperiods after reperfusion and liver biopsies were performed for fur-ther evaluation. Experiments to assess animal survival were per-formed by resecting the nonischemic lobes (30%) at the time ofreperfusion.

Measurement of the Portal Pressure. The abdominal cavity wasopened by a midline laparotomy. A 24-gauge cannula was insertedinto the portal vein and connected to a pressure device (Horizon2000; Mennen Medical Inc, Clarence, NY). The intraportal pressurewas determined before ischemia and after 1 hour of ischemia and 1hour of reperfusion.

TdT-Mediated dUTP-Digoxigenin Nick-End Labeling Staining. Afterischemia/reperfusion, livers were perfused with freshly prepared 4%paraformaldehyde in phosphate-buffered saline (pH 7.2) under aconstant pressure of 10 cm H2O for 5 minutes through the portalvein. The liver was cut in 3- to 5-mm sections and was stored in 70%alcohol after additional overnight fixation in 4% paraformaldehyde.Tissues were than incubated in 30% sucrose in phosphate-bufferedsaline, embedded in 7.5% gelatin and finally frozen in isopentanesubmerged in dry ice and 95% alcohol slush. Sections of 5 mm wereplaced on silanized slides, and were treated with terminal deoxynu-cleotidyl transferase from calf thymus in the presence of fluorescein-dUTP and d-NTP, according to the supplier’s recommended protocol(Boehringer Mannheim, Indianapolis, IN; catalog #1767305). Thiswas followed by poststaining using horseradish peroxidase–conju-gated anti-fluorescein antibody and development using diamino-benzidine/H2O2. Positive and negative controls were done using testsections pretreated with DNAse I and staining without deoxynucle-otide substrate, respectively. Thirty random sections were investi-gated per slide to determine the percentage of TdT-mediated dUTP-digoxigenin nick-end labeling (TUNEL)-positive cells.

Caspase Assays. Caspase 3– and caspase 8–like activity was deter-mined by measuring proteolytic cleavage of the respective speci-fic substrates N-acetyl-Asp-Glu-Val-Asp-7-amino-4-trifluoromethylcoumarin (Ac-DEVD-AFC; Biomol, Plymouth Meeting, PA) or N-acetyl-Ile-Glu-Thr-Asp-7-amino-4-trifluoromethyl coumarin (Ac-IETD-AFC)in the presence or absence of the specific caspase 3 or 8 aldehyde inhibitorbased on the same amino acid sequence (Ac-DEVD-CHO and Ac-IETD-CHO, respectively; Biomol). Liver tissue was quickly excised and soni-cated in assay buffer (1 mmol/L ethylenediaminetetraacetic acid, 145mmol/L NaCl, 100 mmol/L TRIS, 0.1 mmol/L dithiothreitol, 0.1%CHAPS, 10% glycerol). Protein content was determined using the Brad-ford protein assay. The samples were diluted and incubated at room tem-perature with Ac-DEVD-AFC substrate in the presence or absence of theinhibitor Ac-DEVD-CHO. AFC release was assayed over 2 hours in afluorometer, using 400 nm excitation and measurement of 505 nm emis-sion. AFC release is expressed as arbitrary fluorescence units per milli-gram of liver tissue after subtracting the reading in the inhibited samplefrom the noninhibited sample.

Western Blot to Determine Mitochondrial Cytochrome c Release. Thetissue was homogenized in 10 volumes/weight buffer A (20 mmol/LHEPES-KOH [pH 7.5], 10 mmol/L KCl, 1.5 mmol/L MgCl2, 1mmol/L ethylenediaminetetraacetic acid, 1 mmol/L ethylene glycol-bis(b-aminoethyl ether)-N,N-tetraacetic acid, 1 mmol/L dithiothrei-

tol, 0.1 mmol/L phenylmethyl sulfonyl fluoride, 250 m mol/L su-crose) using a Dounce grinder. The homogenate was centrifuged at400g for 5 minutes at 4°C to remove unbroken cells and nuclei. Thesupernatant was recentrifuged at 12,000g to pellet the mitochondria.The supernatant was boiled with an equal volume of 23 sodiumdodecyl sulfate sample buffer for 90 seconds and subjected to 4% to20% sodium dodecyl sulfate-polyacrylamide gel electrophoresis. Theproteins were transferred onto a nitrocellulose filter. Blots wereprobed with a rabbit polyclonal anti-cytochrome c antibody (SantaCruz Biotechnology, Santa Cruz, CA) followed by a secondary anti-body conjugated to horseradish peroxidase and detected with en-hanced chemiluminescence Western blotting detection reagents(Amersham, Little Chalfont, UK).

Determination of Necrosis. Hepatocellular necrosis was determinedin hematoxylin-eosin (HE)-stained tissue using a semiquantitativescale by a point counting method as previously described.21 Thirtyrandom sections were investigated per slide to determine the per-centage of necrotic cells. In this study, only grade 3 injury withdisintegration of hepatic cords was counted as necrosis.

Determination of Polymorphonuclear Leukocyte Adhesion. Formalin-fixed slides were paraffin embedded. Five-micrometer sections werecut and stained with HE. Sinusoidal polymorphonuclear leukocytes(PMN) adhesion was determined by morphology.

Parameter of Hepatocyte Injury. The degree of hepatic injury wasassessed by serum levels of AST, an established marker of hepatocel-lular injury in the rodent liver.21,22 AST serum levels were analyzedusing a serum multiple analyzer (Johnson & Johnson, EktachemDTSC II multianalyzer, Rochester, NY).

Determination of Lung and Kidney Injury. At the end of liver reperfu-sion kidney and lung tissue were excised and fixed in 10% formalin.The tissue was HE stained and evaluated by a blinded investigator(J.M.). Light microscopic evidence of acute tubular injury in thekidney was graded on an ordinal scale, as follows. Thirty randomsections were investigated per slide to determine the grade of tubularnecrosis. Grade 0 was applied to specimens that showed normal,fully intact tubular cytoarchitecture; grade 1 to specimens thatshowed tubular vacuolization, but substantially intact tubular archi-tecture; grade 2 to specimens that showed sloughing of tubular cells;and grade 3 to specimens that showed wholesale loss of tubularepithelial integrity accompanied by intratubular hemorrhage and in-flammatory infiltration.

Statistical Analysis. The data were analyzed with the SPSS (Statis-tical Product and Service Solutions) statistics package. A Student’s ttest and a two-way analysis of variance (ANOVA) were used for thecomparison between fatty and normal livers as indicated. The resultsare presented as mean 6 SEM and are considered significant when Pis less than .05.

RESULTS

Do Ischemia and Reperfusion in Steatotic Livers Result in In-creased Injury Compared With the Lean Animal? AST release is awell-established marker of hepatocellular injury after is-chemia and reperfusion.23 We measured serum AST levelsafter 60 minutes of ischemia followed by 3, 6, 12, 24, or 48hours of reperfusion. Lean rats had only a moderate increasein AST with maximum AST levels at 6 hours of reperfusion. Incontrast, AST levels were significantly higher in fatty rats ateach time point (Fig. 1) (Student’s t test, P , .01; two-wayANOVA for factor steatosis is P , .001). In addition, ASTlevels remained elevated in steatotic animals over 48 hours ofreperfusion.

As a second endpoint of liver injury we determined animalsurvival in a model of total hepatic ischemia, as previouslydescribed.11 Survival was considered permanent if the ratswere alive 30 days after surgery. All lean animals survivedpermanently after 60 minutes of hepatic ischemia, whereas all

HEPATOLOGY Vol. 32, No. 6, 2000 SELZNER ET AL. 1281

fatty animals died within 3 days under the same experimentalcondition (n 5 7 in each group).

Is Apoptosis a Feature of Ischemia and Reperfusion Injury in Stea-totic Livers? Apoptosis of hepatocytes is a key feature of is-chemia and reperfusion injury in normal livers.9 In this studywe investigated whether the same pattern of injury is presentin the steatotic liver. Steatotic and lean livers were exposed to15, 30, 45, or 60 minutes of ischemia, and then reperfused insitu for 3 hours. Apoptosis was determined with the TUNELassay. Both groups showed an increased number of apoptotichepatocytes with prolonged ischemic times (Fig. 2). However,the amount of apoptosis was significantly less in fatty whencompared with lean rats at each respective time (two-wayANOVA for factor steatosis P , .001) (Fig. 3A and B). Forexample, after 60 minutes of ischemia and 3 hours of reper-fusion 39% 6 9% of the hepatocytes in fatty rats were apopto-tic compared with 85% 6 12% in lean animals (Student’s ttest, P 5 .01). In the next set of experiments, each rat wassubjected to 60 minutes of ischemia, and then reperfused for1, 3, 6, 12, 24, or 48 hours. Minimal hepatocyte apoptosis(,5%) was present after 1 hour of reperfusion in both groups.Subsequently, lean rats developed massive hepatocytes apo-ptosis peaking at 3 hours of reperfusion (85%). In contrast,fatty rats had less extensive apoptosis with a maximum degreeof apoptosis 6 hours after reperfusion (55%) (Student’s t test,P , .05; two-way ANOVA for factor steatosis is P , .001).

We further investigated the apoptotic cascade to determineif the decreased amount of apoptosis is related to a failure ofinduction of apoptosis or to a specific block in the apoptoticcascade. Several cellular mediators of apoptosis were investi-gated. Caspase 8 is an upstream mediator of apoptosis. It be-comes activated early in the apoptotic cascade and was there-fore determined shortly (1 hour) after reperfusion. Activationof caspase 8 results in cytochrome c release from the mito-chondria. The release of cytochrome c activates caspase 3,which leads to DNA fragmentation and cell death. Cyto-chrome c release, caspase 3 activation, and DNA fragmenta-tion are downstream events in the apoptotic cascade, and were

evaluated at 3 hours of reperfusion.24 After 60 minutes ofischemia and 1 hour of reperfusion, fatty rats had significantreduction in caspase 8 activity when compared with lean an-imals (Fig. 4). Similarly, at 3 hours of reperfusion cytochromec release (data not shown) and caspase 3 activity (Fig. 5) weresignificantly reduced in steatotic animals when comparedwith lean rats (Student’s t test, P , .05).

Finally, we evaluated neutrophil adhesion in fatty and nor-mal livers before and after 60 minutes of ischemia followed by3, 6, 12, 24, or 48 hours of reperfusion. Minimal sinusoidalPMN adhesion was present at any time point studied, andtherefore no difference between fatty and normal livers wasdocumented (data not shown).

Is Steatosis Associated With Necrotic Cell Death After Ischemiaand Reperfusion? Because apoptosis is a less important featureof reperfusion injury in the steatotic liver, we investigatedwhether necrosis is a major feature of reperfusion injury infatty animals. Necrosis was determined by HE staining of liverbiopsy specimens taken after 60 minutes of ischemia and 3, 6,12, 24, and 48 hours of reperfusion. The percentage of ne-crotic liver tissue was determined by a blinded investigator(M.S.) using a point counting method as previously de-scribed.21 Ischemia and reperfusion in steatotic livers wereassociated with a high percentage of necrosis peaking at 24hours of reperfusion (73% necrosis). In contrast, livers fromlean animals showed minimal necrosis with a maximum of18% after 48 hours of reperfusion (Fig. 6A and B and Fig. 7)(Student’s t test, P , .01; two-way ANOVA for factor steatosisis P , .01).

Does Steatosis Result in Increased Portal Pressure After Ischemiaand Reperfusion? Disturbance in the hepatic microcirculationresulting in portal hypertension has been described by othersas an important contributing factor of injury in fatty livers.25

Therefore, we measured intraportal pressure in lean and fattyrats before ischemia and after 1 hour of total hepatic ischemiafollowed by 60 minutes of reperfusion. Steatotic and lean an-imals had similar levels of portal pressure before ischemia(Student’s t test, P 5 .5). In addition, no significant change ofthe portal pressure was observed after ischemia and 1 hour ofreperfusion between both groups (Student’s t test, P 5 .35)(data not shown).

FIG. 2. TUNEL staining after various times of ischemia followed by 3hours of reperfusion. Prolonged ischemia times resulted in increased apopto-tic hepatocytes in fatty (h) and lean (■) rats. However, lean animals showedsignificantly more apoptosis after 30, 45, and 60 minutes of ischemia com-pared with the fatty animals (Student’s t test, *P , .05; two-way ANOVA forfactor steatosis is P , .001; n 5 5).

FIG. 1. Serum AST levels in lean (h) and fatty (■) Zucker rats after 60minutes of ischemia and various times of reperfusion. Lean rats showed onlymoderate AST elevation with a peak AST level 6 hours after reperfusion. Incontrast, steatotic rats had higher AST levels compared with the lean, and thelevels remained elevated during the entire reperfusion period. (*P , .05,Student’s t test; two-way ANOVA for factor steatosis is P , .001; n 5 5.)

1282 SELZNER ET AL. HEPATOLOGY December 2000

Do Ischemia and Reperfusion of the Liver Result in SecondaryInjury of the Lung and the Kidney? Necrosis is associated withthe release of a variety of proinflammatory mediators. Thesemediators may result in additional injury in distant organs.Therefore, we evaluated tissue injury in kidneys and lungs 60minutes after hepatic ischemia in fatty and lean rats. The tis-sue was stained with HE and examined by a blinded investi-gator (J.M.). No pathologic findings were detected in lungtissue in fatty and lean rats. In contrast, the kidney of steatoticrats showed a significant amount of acute tubular necrosis(grade 4) after 24 hours of reperfusion. Only minimal tubularnecrosis was detected in the kidneys of lean animals (Fig. 8

and Fig. 9A and B) (Student’s t test, P , .01; two-way ANOVAfor factor steatosis is P , .001).

Are Protective Strategies Against Ischemic Injury in Lean LiversAlso Effective in the Fatty Animals? Inhibition of apoptosis byblocking the caspase cascade is a highly effective strategy toreduce ischemic injury of the normal liver. We10 and others12

have shown the protective effects of a pan caspase inhibitorafter normothermic12 and cold10 ischemic injury in differentanimal models. It has been suggested that under certain cir-cumstances the apoptotic process can be abrogated, and ne-crosis may occur. If necrosis after ischemia and reperfusion insteatotic livers is the result of incomplete apoptosis, then

FIG 3. TUNEL staining of fatty(A) and lean (B) livers after 60 min-utes of ischemia and 3 hours ofreperfusion. At this time point, mosthepatocytes in lean livers wereTUNEL positive, whereas only min-imal TUNEL staining cells werefound in steatotic livers.

HEPATOLOGY Vol. 32, No. 6, 2000 SELZNER ET AL. 1283

blocking the apoptotic pathway may also be protective in fattylivers. Therefore, we tested the effect of a pan-caspase inhibi-tor (IDN 1965) on liver injury in fatty and lean rats. We usedAST release, hepatocyte apoptosis, and necrosis as markers ofhepatocellular injury. In lean animals caspase inhibition re-sulted in a complete protection of hepatocytes against apopto-sis (8% with caspase inhibitor vs. 85% in controls). AST re-lease was also significantly reduced in lean rats by caspaseinhibition (664U/L vs. 1538U/L; P , .05, Student’s t test).Lean animals subjected to caspase inhibition had a moderateamount of necrosis after 24 hours of reperfusion (20% withcaspase inhibition vs. 13% in controls; P 5 .1, Student’s t test).

Similarly, caspase inhibition significantly reduced apoptosis infatty rats (19% with caspase inhibition vs. 39% in controls). How-ever, fatty rats showed no change in the amount of necrosis (66%with caspase inhibition vs. 73% in controls) and AST release(3506U/L with caspase inhibition vs. 3648U/L in controls) (Stu-dent’s t test, P 5 .34) (Fig. 10).

DISCUSSION

In this study we show different mechanisms of ischemiaand reperfusion injury in the normal and the steatotic liver.Rapid apoptosis was the predominant form of hepatocytedeath in the ischemic normal liver, whereas the amount ofapoptosis was significantly reduced and the onset of apoptosiswas delayed in obese rats. In contrast, the fatty livers devel-oped massive necrosis after an ischemic insult. Mediators ofthe apoptotic cascade were significantly decreased in steatoticanimals when compared with the lean. Caspase inhibition, ahighly protective strategy in the normal liver, had no effect onhepatocyte injury in the fatty liver. Finally, fatty but not leananimals developed severe acute tubular necrosis in the kidneyas a result of hepatic ischemia.

Our results confirm that steatotic livers tolerate ischemicinjury poorly. All fatty animals died after 60 minutes of is-chemia, whereas all lean animals survived the same ischemicchallenge. Of particular interest was the serum AST levelsduring reperfusion. The AST levels increased more rapidlyand to higher values in fatty rats when compared with the

lean. This might be related to the early onset of necrosis insteatotic livers after ischemia. One of the hallmarks of necrosisis the loss of membrane integrity resulting in the release ofcytoplasmic contents including AST. In contrast, during apo-ptosis cells are fragmented into smaller apoptotic bodies withintact membranes, which may reduce the release of AST. Also,lean animals had an early AST peak at 3 hours with decliningserum AST levels thereafter. Correspondingly, it has been pre-viously shown that the percentage of apoptotic cells decreasesafter 3 hours of reperfusion by Kupffer cell phagocytosis.8,9

This may indicate that the injury is limited to the initial periodof reperfusion in the normal liver. In steatotic rats, by con-trast, AST levels remained elevated during the reperfusionperiod, possibly because of ongoing loss of hepatocyte viabil-ity. This is supported by the finding that necrosis increasedwith prolonged reperfusion times. Another contributing fac-tor might be decreased clearance of AST because of Kupffercell dysfunction. Kupffer cells are important for the removalof AST and severe steatosis has been associated with a mal-function of Kupffer cells,26 which may prevent the clearanceof elevated AST levels.27

Reperfusion in normal livers resulted in rapid activation ofthe apoptotic cascade evidenced by activation of caspase 8,mitochondrial cytochrome c release and activation of caspase3. In contrast, in fatty livers each tested mediator of apoptosiswas significantly decreased compared with the lean indicatinga dysfunction in the apoptotic pathway. The failure of apopto-sis in the fatty liver could be linked to decreased adenosinetriphosphate production in fatty hepatocytes. Steatosis is as-sociated with mitochondrial dysfunction and decreased intra-cellular energy levels.3,4,28 Apoptosis is an active, energy-re-quiring process, and low adenosine triphosphate levels couldresult in a failure to activate the apoptotic cascade. A secondpossibility is a dysfunction of regulators of apoptosis in fattyhepatocytes. In a recent study Rashid et al.29 described up-regulation of antiapoptotic and proapoptotic factors, likeBcl-2, Bcl-xL, and Bax, in steatotic mice. The imbalance inmediators controlling apoptosis might be linked to the reduc-

FIG. 5. Caspase 3 activity after 60 minutes of ischemia and 3 hours ofreperfusion in fatty (■) and lean (h) rats. Steatotic livers had a decreasedcaspase 3 activity when compared with the lean livers (Student’s t test, *P ,.05, n 5 5).

FIG. 4. Caspase 8 activity, an early marker of apoptosis, after 60 minutesof ischemia and 1 hour of reperfusion. Fatty animals (■) had reduced activityof caspase 8 when compared with lean animals (h) (Student’s t test, *P , .05,n 5 5).

1284 SELZNER ET AL. HEPATOLOGY December 2000

tion of apoptosis after ischemia and reperfusion. Finally,Kupffer cell dysfunction associated with ischemia and reper-fusion injury in fatty liver may result in decreased Kupffercell–dependent production of proapoptotic mediators, suchas tumor necrosis factor a or oxygen-free radicals. This mightresult in a failure to initiate the apoptotic cascade.26 Indeed,the failure of caspase 8 activation, an early step of apoptosis,supports the idea that the initiation of apoptosis is impaired infatty animals.

In this model major apoptosis (80%) occurring after 3hours of reperfusion in lean rats was still compatible with100% animal survival. In contrast late (24-hour) necrosis

(70% of the tissue) was consistently associated with death infatty rats. This emphasizes the fact that necrosis is a poorlytolerated form of cell death, whereas even extensive apoptosisis compatible with animal survival. Rapid apoptosis is thephysiologic reaction of an acute ischemic insult. The damagedcells undergo apoptosis within a few hours and are thenquickly removed by macrophages without inflammation. Thislimits the organ injury and protects the surviving hepatocytesas well as distant organs. However, in the case of prolongedischemic injury, subtotal hepatocyte apoptosis occurs result-ing in animal death.8,9 We have previously shown that anti-apoptotic strategies are highly protective if massive apoptosis

FIG 6. HE staining of liver tissueafter 60 minutes of ischemia and 24hours of reperfusion in lean (A) andfatty (B) rats. Massive necrosis waspresent in fatty livers, whereas onlyminimal injury was visible in leanlivers.

HEPATOLOGY Vol. 32, No. 6, 2000 SELZNER ET AL. 1285

(.90%) is induced by ischemia/reperfusion injury resultingin increased animal survival and decreased liver injury.9-11

Necrosis, as an uncontrolled form of cell death, is associatedwith the release of cytoplasm contents and significant inflam-mation. Additional injury can occur from these secondaryeffects. This is supported by our findings that, first, the per-centage of necrotic liver tissue increased over the reperfusiontime, and second, that liver necrosis resulted in severe acutetubular necrosis of the kidneys.

Our results do not exclude the possibility that inhibition ofapoptosis might be protective in cases of mild or moderate

steatosis. In fact, results from a recent clinical study in hu-mans suggest that inhibition of apoptosis through a techniqueof ischemic preconditioning protects against reperfusion in-jury in the presence of mild to moderate steatosis (20%-50%).30

Several investigators have described increased intrahepaticresistance of the hepatic microcirculation as a consequence of

FIG. 7. Percentage of necrotic liver tissue after 60 minutes of ischemia anddifferent times of reperfusion. Only a small percentage of liver tissue devel-oped necrosis in lean animals (h). In contrast, massive necrosis developed insteatotic livers (■) after prolonged reperfusion times. (Student’s t test, *P ,.01; two-way ANOVA for factor steatosis is P , .01; n 5 5.)

FIG. 8. Grade of acute tubular necrosis (ATN) in kidneys of lean (h) andfatty (■) rats after 60 minutes of hepatic ischemia and different times ofreperfusion. Only minimal acute tubular necrosis was present after hepatic is-chemia and reperfusion in lean animals. In contrast, fatty animals developedin severe ATN beyond 12 hours of hepatic reperfusion (Student’s t test, *P ,.01; two-way ANOVA for factor steatosis is P , .01; n 5 5).

FIG. 9. HE staining of kidneys from lean (A) and fatty (B) rats 24 hoursafter 60 minutes of hepatic ischemia. The kidneys of fatty rats show severeacute tubular necrosis, whereas only minimal injury was present in leananimals.

FIG. 10. Serum AST levels in lean and fatty Zucker rats after 60 minutes ofischemia and 6 hours of reperfusion with and without caspase inhibition(IDN 1965). Caspase inhibition significantly decreased AST levels in lean rats(h) but not in fatty animals (■) (Student’s t test, *P , .05, n 5 5).

1286 SELZNER ET AL. HEPATOLOGY December 2000

steatosis.25,31,32 Ohara33 reported a reduction of the sinusoidalspace by 53% compared with lean controls and Wada et al.25

described portal hypertension in the presence of advancedsteatosis. To evaluate whether steatosis or necrosis in ourmodel was associated with impaired portal flow and mesen-teric congestion, we measured the portal pressure in steatoticand lean animals before ischemia and after ischemia andreperfusion. No significant difference was detected in portalpressure in lean and fatty animals. The fact that the portalpressure was not elevated in fatty rats by adequate hepaticreperfusion indicates that a lack of reperfusion was probablynot a major contributing factor in fatty livers. However, wecannot exclude that disturbance of microcirculation is in-volved in the mechanism of injury in steatotic livers.

Because massive necrosis is associated with the release ofintracellular contents, which may induce tissue damage indistant organs we investigated kidney and lung histology afterreperfusion of fatty and lean rats. No lung injury was detectedin fatty and lean animals following hepatic ischemia andreperfusion. In contrast, we found a high amount of acutetubular necrosis in the kidney of fatty livers while the kidneysof lean animals showed only minimal injury. Indeed, kidneyfailure is a common finding in patients with severe liver in-jury, which often appears 8 to 24 hours after major liver in-jury.

In conclusion, hepatic ischemia and reperfusion werepoorly tolerated in steatotic rats with an increased AST releaseand decreased animal survival. Although rapid apoptosis wasthe predominant form of cell death in lean animals, apoptosiswas delayed and reduced in steatotic livers. In addition, intra-cellular mediators of apoptosis were decreased in steatoticlivers indicating a failure to activate the apoptotic cascade. Incontrast, massive necrosis was the dominant mechanism ofinjury after ischemia and reperfusion in fatty rats. Further-more, hepatocellular necrosis was associated with secondaryorgan injury in the form of acute tubular necrosis of the kid-neys. Caspase inhibition as an antiapoptotic strategy was noteffective in fatty livers, whereas lean animals were signifi-cantly protected. Future studies should focus on the search ofspecific protective strategies for the steatotic livers.

Acknowledgment: The authors thank V. Rousson from theInstitute of Biostatistics of the University of Zurich for hisassistance during the preparation of this manuscript.

REFERENCES

1. Hilden M, Christoffersen P, Juhl E, Dalgaard J. Liver histology in a nor-mal population-examinations of 503 consecutive fatal traffic casualties.Scand J Gastroenterol 1977;12:593-597.

2. Hornboll P, Olsen T. Fatty changes in the liver: the relation to age,overweight and diabetes mellitus. Acta Pathol Microbiol Immunol Scand1982;90:199-205.

3. Berson A, De Beco V, Letteron P, Robin, MA, Moreau C, Kahwaaji JE,Verthier N, et al. Steatohepatitis-inducing drugs cause mitochondrialdysfunction and lipid peroxidation in rat hepatocytes. Gastroenterology1998;114:764-774.

4. Fromenty B, Pessayre D. Impaired mitochondrial function in microve-sicular steatosis. J Hepatol 1997;26:43-53.

5. Behrns KE, Tsiotos GG, DeSouza NF, Krishna, MK, Ludwig J, NagorneyDM. Hepatic steatosis as a potential risk factor for major hepatic resec-tion. J Gastrointest Surg 1998;2:292-298.

6. Selzner M, Clavien P-A. Resection of liver tumors: special emphasis onneoadjuvant and adjuvant therapy. In: Clavien P-A, ed. Malignant LiverTumors—Current and Emerging Therapies. Malden, MA: Blackwell Sci-ence 1999;137-149.

7. D’Alessandro A, Kalayoglu M, Sollinger H, Hoffmann RM, Reed A,Knechtle SJ, Pirsch JD, et al. The predictive value of donor liver biopsies

for the development of primary nonfunction after orthotopic liver trans-plantation. Transplantation 1991;51:157-163.

8. Gao W, Bentley RC, Madden JF, Clavien PA. Apoptosis of sinusoidalendothelial cells is a critical mechanism of prevention injury in rat livertransplantation. HEPATOLOGY 1998;27:1652-1660.

9. Kohli V, Selzner M, Madden JF, Bentley RC, Clavien PA. Endothelial celland hepatocyte deaths occur by apoptosis after ischemia-reperfusion in-jury in the rat liver. Transplantation 1999;67:1099-1105.

10. Natori S, Selzner M, Valentino K, Fritz LC, Srinivasan A, Clavien PA,Gores GJ. Apoptosis of sinusoidal endothelial cells occurs during liverpreservation injury by a caspase-dependent mechanism. Transplantation1999;68:89-96.

11. Yadav SS, Sindram D, Perry D, Clavien P-A. Ischemic preconditioningprotects the mouse liver by inhibition of apoptosis through a caspase-dependent pathway. HEPATOLOGY 1999;30:1223-1231.

12. Cursio R, Gugenheim J, Ricci JE, Crenesse D, Rostagno P, Maulon L,Saint-Paul MC, et al. Caspase inhibitor fully protects rats against lethalnormothermic liver ischemia by inhibition of liver apoptosis. FASEB J1999;13:253-261.

13. Selzner M, Rudiger HA, Sindram D, Clavien PA. Bcl-2 transgenic mice areresistant to ischemia/reperfusion injury of the liver [Abstract]. HEPATOL-OGY 1999;30:103,386A.

14. Bilbao G, Contreras JL, Eckhoff DE, Mikheeva G, Krasnykh V, DouglasJT, Thomas FT, et al. Reduction of ischemia-reperfusion injury of theliver by in vivo adenovirus-mediated gene transfer of the antiapoptoticBcl-2 gene. Ann Surg 1999;230:185-193.

15. Taneja C, Prescott L, Koneru B. Critical preservation injury in rat fattyliver is to hepatocytes, not sinusoidal lining cells. Transplantation 1998;65:167-172.

16. Teramoto K, Bowers JL, Khettry U, Palombo JD, Clouse ME. A rat fattyliver transplant model. Transplantation 1993;55:737-741.

17. Caraceni P, Nardo B, Domenicali M, Turi P, Vici M, Simoncini M, DeMaria N, et al. Ischemia-reperfusion injury in the rat fatty liver: role ofnutritional status. HEPATOLOGY 1999;29:1139-1146.

18. White B, Martin R. Evidence for central mechanism of obesity in theZucker rat: role of neuropeptide Y and leptin. Proc Soc Exp Biol Med1997;217:222-232.

19. Koneru B, Reddy M, Torre A, Patel D, Ippolito T, Ferrante RJ. Studies ofhepatic warm ischemia in the obese Zucker rat. Transplantation 1995;59:942-946.

20. Selzner M, Clavien P-A. Failure of regeneration of the steatotic rat liver—disruption at two different levels of the regeneration pathway. HEPATOL-OGY 2000;31:35-42.

21. Camargo CA, Jr., Madden JF, Gao W, Selvan RS, Clavien PA. Interleu-kin-6 protects liver against warm ischemia/reperfusion injury and pro-motes hepatocyte proliferation in the rodent. HEPATOLOGY 1997;26:1513-1520.

22. Selzner M, Camargo C, Clavien P-A. Ischemia impairs liver regenerationfollowing major tissue loss in the rodent. HEPATOLOGY 1999;30:469-475.

23. Iu S, Harvey P, Makowka L, Petrunka CN, Ilson RG, Strasberg SM. Mark-ers of allograft viability in the rat. Relationship between transplantationviability and liver function in the isolated perfused rat liver. Transplan-tation 1987;45:562-569.

24. Scaffidi C, Fulda S, Srinivasan A, Friesen C, Li F, Tomaselli KJ, DabatinKM, et al. Two CD95 (Apo-1/Fas) signaling pathways. EMBO J 1998;17:1675-1687.

25. Wada K, Fujimoto K, Fujikawa Y, Shibayama Y, Mitsui H, Nakata K.Sinusoidal stenosis as the cause of portal hypertension in choline defi-cient diet induced fatty cirrhosis of the rat liver. Acta Path Jap 1974;24:207-217.

26. Thurman R, Gao W, Connor H, Adachi Y, Stachlewitz RF, Zhong Z,Knecht KT, et al. Role of Kupffer cells in failure of fatty livers followingliver transplantation and alcoholic injury. J Gastroenterol Hepatol1995;10 Suppl:S24-S30.

27. Clavien PA, Camargo CA, Jr., Cameron R, Washington MK, Phillips MJ,Greig PD, Levy GA. Kupffer cell erythrophagocytosis and graft-versus-host hemolysis in liver transplantation. Gastroenterology 1996;110:1891-1896.

28. Letteron P, Fromenty B, Terris B, Degott C, Pessayre D. Acute andchronic hepatic steatosis lead to in vivo lipid peroxidation in mice.J Hepatol 1996;24:200-208.

29. Rashid A, Wu T-C, Huang C-C, Chen C-H, Lin HZ, Yang SQ, Janet FY,Diehl AM. Mitochondrial proteins that regulate apoptosis and necrosisare induced in mouse fatty liver. HEPATOLOGY 1999;29:1131-1138.

HEPATOLOGY Vol. 32, No. 6, 2000 SELZNER ET AL. 1287

30. Clavien P, Yadav SS, Sindram D, Bentley CR. Protective effects of is-chemic preconditioning for liver resection performed under inflow oc-clusion in humans. Ann Surg 2000;232:155-162.

31. Hakamada K, Sasaki M, Takahashi K, Umehara Y, Konn M. Sinusoidalflow block after warm ischemia in rats with diet-induced fatty liver. J SurgRes 1997;70:12-20.

32. Teramoto K, Bowers J, Kruskal J, Clouse M. Hepatic microcirculatorychanges after reperfusion in fatty and normal liver transplantation in therat. Transplantation 1993;56:1076-1082.

33. Ohara K. [Study of microcirculatory changes in experimental dietaryfatty liver]. Hokkaido Igaku Zasshi - Hokkaido Journal of Medical Sci-ence 1989;64:177-185.

1288 SELZNER ET AL. HEPATOLOGY December 2000