Ex vivo effects of high-density lipoprotein exposure on the lipopolysaccharide-induced inflammatory response in patients with severe cirrhosis

LIVER FAILURE/CIRRHOSIS/PORTAL HYPERTENSION

Ex Vivo Effects of High-Density Lipoprotein Exposureon the Lipopolysaccharide-Induced Inflammatory

Response in Patients with Severe CirrhosisArnaud Galbois,1,2,3 Dominique Thabut,1,4 Khalid A. Tazi,1,2,3 Marika Rudler,1,2,3 Morvarid Shir Mohammadi,1,2,3

Dominique Bonnefont-Rousselot,5,6 Hind Bennani,7 Annie Bezeaud,7 Zera Tellier,8 Cecile Guichard,1,2,3 Nicolas Coant,9

Eric Ogier-Denis,9 Richard Moreau,1,2,3* and Didier Lebrec1,2,3*

High-density lipoproteins (HDL) are known to neutralize lipopolysaccharide (LPS). Becausepatients with cirrhosis have lower HDL levels, this may contribute to LPS-induced ex vivomonocyte overproduction of proinflammatory cytokines. However, the effects of HDL on cyto-kine production by monocytes from patients with cirrhosis have never been studied. The aim ofthis study was to determine the effects of HDL on LPS-induced proinflammatory cytokineproduction in whole blood and isolated monocytes from patients with severe cirrhosis andcontrols. Plasma levels of HDL and cytokines were determined. The effects of reconstituted HDL(rHDL) on LPS-induced cytokine production in whole blood were assessed by cytokine arrayand on tumor necrosis factor alpha (TNF-�) and interleukin-10 (IL-10) production in isolatedmonocytes. Plasma HDL levels were significantly lower in patients with cirrhosis than in controls.Plasma levels of TNF-� and IL-6 were significantly higher in patients with cirrhosis than in controls.Incubation of rHDL with whole blood prevented LPS-induced TNF-� and IL-6 overproductionin patients with cirrhosis. LPS-induced TNF-� production and CD14 expression were signifi-cantly more marked in cirrhotic monocytes than in control monocytes, and both decreased signifi-cantly after rHDL incubation. LPS-induced down-regulation of scavenger receptor, class B, type I(SR-BI) expression was abolished in cirrhotic monocytes. Conclusions: This study shows thatrHDL abolishes the LPS-induced overproduction of proinflammatory cytokines in whole bloodfrom patients with severe cirrhosis. These results were confirmed in isolated monocytes from thesepatients. This suggests that administration of rHDL might be a useful strategy for the treatmentof cirrhosis to limit LPS-induced cytokine overproduction. (HEPATOLOGY 2009;49:175-184.)

In patients with cirrhosis, gram-negative bacterial in-fections induce an overproduction of the proinflam-matory cytokine tumor necrosis factor alpha (TNF-

�), which is involved in both systemic vasodilatation1,2

and progression of liver injury.3 Moreover mortality fromseptic shock reaches 80% in patients with cirrhosis4 com-pared with 30% in patients without.5 In case of infection,increased TNF-� production is much higher in patientswith than in those without cirrhosis.6 It has also beendemonstrated that administration of or exposure to lipo-polysaccharide (LPS) induces marked TNF-� productionin vivo in cirrhotic rats7,8 and ex vivo in monocytes frompatients with cirrhosis.9 Recently, it was shown thatmonocytes from patients with severe cirrhosis were spon-taneously activated to produce proinflammatory cyto-kines and exhibited a hyperresponsiveness to LPS.10-12

High-density lipoprotein (HDL) particles are multi-functional lipoprotein complexes that transport lipids andhave several anti-inflammatory properties.13 These in-clude lipid transport and antioxidant activities that re-

Abbreviations: apo, apolipoprotein; EDTA, ethylenediaminetetra-acetic acid;HDL, high-density lipoproteins; I�B-�, Inhibitor kappaB-alpha; IL-6, interleu-kin-6; IL-10, interleukin-10; LBP, lipopolysaccharide binding protein; LDL, low-density lipoprotein; LPS, lipopolysaccharide; NF-�B, nuclear factor kappaB;rHDL, reconstituted high-density lipoproteins; SDS, sodium dodecyl sulfate; SR-BI,scavenger receptor, class B, type I; TNF-�, tumor necrosis factor alpha.

From 1INSERM, U773, Centre de Recherche Biomedicale Bichat-BeaujonCRB3, Hopital Beaujon, Clichy, France ; 2Universite Denis Diderot-Paris 7, siteBichat, Paris, France; 3Service d’Hepatologie, Hopital Beaujon, Clichy, France;4Service d’Hepato-Gastroenterologie, Hopital de la Pitie-Salpetriere, Paris, France;the 5Laboratoire des Lipides, Hopital de la Pitie-Salpetriere, Paris, France6; Labo-ratoire de Biochimie Metabolique (EA 3617), Faculte de Pharmacie, Paris, France;7Service d’Immunologie et d’Hematologie, Hopital Beaujon, Clichy, France; 8Labo-ratoire Francais du Fractionnement et des Biotechnologies, Courtaboeuf, France;9Universite Denis Diderot-Paris 7, site Bichat, Paris, France.

*These authors contributed equally to this work.Received July 14, 2008; accepted August 7, 2008.Address reprint requests to: D Thabut, INSERM, Hopital Beaujon, 92118

CLICHY, France. E-mail: [email protected]; fax: 01 42161427.Copyright © 2008 by the American Association for the Study of Liver Diseases.Published online in Wiley InterScience (www.interscience.wiley.com).DOI 10.1002/hep.22582Potential conflict of interest: Nothing to report.

175

move or inactivate inflammatory lipids as well as theability to inhibit various inflammatory responses in endo-thelial cells. HDL also can bind and neutralize the bioac-tivity of LPS.13 Declining HDL levels, a well-knowncomplication in cirrhosis,14 is thought to impair the host’sability to neutralize LPS.15 Perfusion of reconstitutedHDL (rHDL) decreases cytokine production, organ dys-function, and mortality in animal models of septicshock.16-18 In healthy volunteers, a double-blind random-ized placebo-controlled study showed that rHDL infu-sion before LPS administration reduced flu-likesymptoms during endotoxinemia.19 In cirrhotic rats withascites, two recent studies have shown that rHDL admin-istration reduced the effects of endotoxin on TNF-� pro-duction and systemic hemodynamics.20,21 These studiessuggest that the excessive proinflammatory response toLPS in cirrhosis may be attributable, at least in part, toreduced LPS neutralization by HDL. Thus, it may bebeneficial to increase plasma HDL concentrations in pa-tients with cirrhosis to increase neutralization of circulat-ing LPS, prevent septic shock, and decrease mortality.However, ex vivo, the effects of HDL on LPS-inducedproinflammatory cytokine overproduction have neverbeen studied in humans with cirrhosis. Thus, the aim ofthe current study was to determine the effects of rHDL onLPS-induced TNF-� production in whole blood and inisolated monocytes from patients with advanced cirrhosisand controls. In addition, because HDL causes down-regulation of CD14 expression in monocytes,19 and be-cause CD14 modulates the proinflammatory response toLPS,22,23 the effects of HDL on CD14 expression weredetermined in whole blood. Because HDL receptor scav-enger receptor, class B, type I (SR-BI) expression may beregulated by cholesterolemia, LPS, and TNF-�,24,25

SR-BI expression was also determined in monocytes afterLPS challenge.

Patients and Methods

Patients. Included patients were hospitalized in theLiver Unit. Inclusion criteria were age over 18 years, di-agnosis of advanced cirrhosis (B or C in the Child-Pughclassification). Healthy subjects (controls) were hospitalemployee volunteers. Noninclusion criteria were: evi-dence of recent gastrointestinal bleeding, bacterial infec-tion in the past 15 days, treatment with antibiotics,corticosteroids, other immunosuppressive drugs, inter-feron, lipid-lowering therapy, or pentoxifylline in the past30 days, presence of hepatocellular carcinoma or othercancer, malignant hemopathy, inflammatory disease,acute alcoholic hepatitis, or human immunodeficiencyvirus infection. The presence of a bacterial infection was

excluded by clinical history, physical examination, whiteblood cell count, blood culture, urine analysis and cul-ture, white blood cell count in ascitic fluid and its culture.Written informed consent was obtained from all patientsand controls.

Plasma Cholesterol, HDL Cholesterol, Low-DensityLipoprotein Cholesterol, Very-Low-Density Lipopro-tein Cholesterol, and Triglyceride Levels. Plasma cho-lesterol concentrations were determined enzymaticallywith a Konelab kit (Konelab, Thermoclinical Labsystmes,Cergy Pontoise, France) on a Konelab 30i analyzer. Be-cause altered composition of lipoproteins in cirrhosiscompromises homogeneous methods,26,27 the ultracen-trifugation technique was used. Plasma HDL-cholesterolconcentration was determined using a phosphotungsticacid/MgCl2 reagent (Boehringer, Mannheim, Germany)to precipitate the apoB-containing lipoproteins, and cho-lesterol was measured in the supernatant, as describedearlier in the serum. Low-density lipoprotein (LDL)-cho-lesterol and very-low-density lipoprotein cholesterol lev-els were assessed using the Friedewald formula whenserum triglyceride was less than 3.40 g/L28 or was mea-sured enzymatically with a Konelab kit. Serum triglycer-ide levels were automatically determined enzymatically(Biomerieux, Marcy l’Etoile, France).

Plasma Lipopolysaccharide-Binding Protein Con-centration. Venous blood samples from patients withcirrhosis and controls were collected in endotoxin-free,ethylenediaminetetra-acetic acid (EDTA) anticoagulatedVacutainer tubes (Beckton Dickinson, San Jose, CA),centrifuged at 20°C at 5,000 rpm � 4,000 g for 15 min-utes, and plasma samples were stored at �80°C untilanalysis. Plasma lipopolysaccharide-binding protein(LBP) level was measured by enzyme-linked immunosor-bent assay according to the protocol supplied by the man-ufacturer (HyCult biotechnology b.v., Cell Sciences,Uden, Netherlands). The intensity of the color was mea-sured in duplicate in a micro plate reader PR-2100 (Bio-Rad, Montlucon, France).

Proinflammatory Cytokines and Chemokine Levelsin Whole Blood. Venous blood samples from patientswith cirrhosis and controls were collected in endotoxin-free EDTA anticoagulated Vacutainer tubes (BecktonDickinson, San Jose, CA). LPS (100 ng/mL) or rHDL (2mg apoA1/mL) or LPS and rHDL were added. TherHDL was prepared from human plasma by cholate dial-ysis by ZLB Behring (Bern, Switzerland) and containedapolipoprotein A1 and phosphatidylcholine (supplied byDr Peter G. Lerch). The fourth tube was used as a control.The samples were incubated at 37°C in 5% CO2/95% airatmosphere for 2 hours. Plasma was then collected aftercentrifugation and frozen immediately at �80°C until

176 GALBOIS ET AL. HEPATOLOGY, January 2009

analysis. Cytokine and chemokine concentrations inplasma samples were studied by human cytokine antibodyarray. The human cytokine array was compared using aHuman Cytokine Antibody Array I kit (Ray Biotech,Norcross, GA) according to manufacturer’s instructionsand with the agent provided by the kit. Briefly, mem-branes were incubated for 30 minutes in a blockingbuffer. Membranes were then incubated for 2 hours withplasma samples (1 mL) followed by three 5-minutewashes with a wash buffer. The membranes were incu-bated for 1 hour with biotin-conjugated cytokines. Afterincubation, membranes were washed three times and in-cubated for 30 minutes with horseradish peroxidase–con-jugated streptavidin. Unbound reagents were washed,and the membranes were developed with the enhancedchemiluminescence system. After exposure of the mem-brane to x-ray film, the signal intensities and relative ex-pression levels of cytokines were quantified by videodensitometry. Values are given in arbitrary units. Resultsare expressed as percentages of control values at baselineconditions (100%).

Human Monocyte Isolation and Culture. Venousblood samples from patients with cirrhosis and controlswere collected in endotoxin-free EDTA anticoagulatedVacutainer tubes (Beckton Dickinson, San Jose, CA).Blood was diluted 1:1 in Roswell Park Memorial Institute1640 (Sigma Aldrich, Steinheim, Germany). Peripheralblood mononuclear cells were isolated by density-gradi-ent centrifugation over endotoxin-free Ficoll-Hypaque(GE Healthcare, Uppsala, Sweden). The cell layer con-taining peripheral blood mononuclear cells was collectedand washed three times in phosphate-buffered saline(Sigma Aldrich, Mannheim, Germany)/0.1% bovine se-rum albumin (Amersham Biosciences, Buckinghamshire,England). Purified monocytes were isolated by negativeselection by depletion of T and B cells using magneticpolystyrene M-450 Dynabeads (Dynal Biotech, Oslo,Norway) coated with antibodies specific for CD2 (T cells)and CD19 (B cells). Monocytes isolated by this procedurewere more than 95% pure. Monocytes (2.5 � 105 cells/well) were cultured at 37°C in 5% CO2/95% air atmo-sphere in Roswell Park Memorial Institute 1640, 100U/mL penicillin, 100 �/mL streptomycin, 2 mmol/L glu-tamine, and 10% heat-inactivated fetal bovine serum(Gibco, Les Ullis, France). After 1 hour, nonadherentcells were removed by washing, and fresh medium wasadded. Then, adherent cells were cultured in fresh me-dium for various times (0-24 hours) with or without 100ng/mL LPS (Escherichia coli serotype 0111:B4) and withor without rHDL (0.5 or 1 or 2 mg of apoA1/mL). Su-pernatant from cultured monocytes was collected and fro-zen immediately at �80°C until analysis of TNF-� and

interleukin-10 (IL-10) concentrations. Monocytes werelysed as described later and frozen immediately at �80°Cuntil analysis by western blot. Viability of monocytes wastested in each condition by trypan blue exclusion assayand was found to be greater than 90% in both control andcirrhotic monocyte cultures.

TNF-� Production in Isolated Monocytes. Mono-cytes from patients with cirrhosis and controls were iso-lated and cultivated as described previously. TNF-�concentrations in the monocyte culture supernatant weremeasured just after fresh medium was added (H0), after 2hours (H2) and after 24 hours (H24) by a commerciallyavailable enzyme-linked immunosorbent assay kit (Bio-source Europe, Nivelles, Belgium). The protocol suppliedby the manufacturer was respected. The color intensitywas measured in duplicate in a microplate reader PR-2100 (Bio-Rad, Montlucon, France).

IL-10 Production in Isolated Monocytes. Mono-cytes from patients with cirrhosis and controls were iso-lated and cultivated as described previously. IL-10concentrations in the monocyte culture supernatant weremeasured just after fresh medium was added (H0) andafter 24 hours (H24) by a commercially available enzyme-linked immunosorbent assay kit (Biosource Europe,Nivelles, Belgium). The protocol supplied by the manu-facturer was respected. The color intensity was measuredin duplicate in a microplate reader PR-2100 (Bio-Rad,Montlucon, France).

Effects of rHDL Incubation on LPS-Induced Nu-clear Factor KappaB Pathway Activation in IsolatedMonocytes. Nuclear factor-kappaB (NF-�B) is main-tained in an inactive state bound to an inhibitory subunitInhibitor kappaB-alpha (I�B-�) in the cytoplasm. In re-sponse to LPS, activation of NF-�B occurs when I�B-� isphosphorylated, ubiquitinated, and degraded by theubiquitin–proteasome pathway. This releases the activeNF-�B heterodimer, which then translocates to the nu-cleus, where it binds to the promoter region of a variety oftarget genes.29 In the current study, activation of NF-�Bwas assessed by measuring I�B-� degradation by westernblotting. Monocytes from patients with cirrhosis and con-trols were isolated and cultivated as described previously.After supernatant recovery, adherent cells were collectedthanks to the lysis buffer [50 mM 4-(2-hydroxyethyl)-1-piperazine ethanesulfonic acid (pH 7.5), 150 mM NaCl,10% glycerol, 1% Triton X-100, 1.5 mM MgCl2, 100mM NaF, 100 mM sodium orthovanadate, 1 mM ethyl-ene glycol tetra-acetic acid (pH 7.7)] and immediatelyfrozen at �80°C until analysis by western blot. Even if thenumber of cells was the same in each well (2.5 � 105), theprotein concentration was determined using the Bio-Radprotein determination assay (Bio-Rad, Montlucon,

HEPATOLOGY, Vol. 49, No. 1, 2009 GALBOIS ET AL. 177

France), and an aliquot of homogenate was suspended insodium dodecyl sulfate (SDS) sample buffer (2% SDS,2% �-mercaptoethanol, and 10% glycerol in 300 mMTris-Cl, pH 6.8). Equal amounts of total protein werethen subject to SDS polyacrylamide gel electrophoresis.Gels were transferred to nitrocellulose (Amersham Bio-sciences, Buckinghamshire, England). Nitrocellulose fil-ters were blocked by incubation in 5% nonfat dry milk,1% bovine serum albumin, 0.1 mol/L NaCl, 0.01 mol/LTris-HCl (pH 7.5), and 0.1% Tween-20 for 2 hours. Themembranes were probed with an anti- I�B-� (Santa CruzBiotechnology, Inc., Santa Cruz, CA) antibody overnightat 4°C. Blots were then incubated in a peroxidase-conju-gated secondary anti-rabbit antibody at room tempera-ture for 1 hour. All immunoblots were visualized byenhanced chemiluminescence (Amersham Biosciences,Buckinghamshire, England). Blots were again washed in0.1 mol/L NaCl, 0.01 mol/L Tris-HCl (pH 7.5), and0.1% Tween 20, and immunoreactivity was assessed us-ing an enhanced western blot detection system. The im-munoreactive bands were quantified by videodensitometry. Values are given in arbitrary units. Resultsare expressed as percentages of control values (100%).

Monocyte CD14 Expression Determination. Ve-nous blood samples from patients with cirrhosis andcontrols were collected in endotoxin-free, EDTA antico-agulated Vacutainer tubes (Beckton Dickinson, San Jose,CA). They were incubated at 37°C (95% air/5% CO2) for1 hour with or without LPS (100 ng/mL) and with orwithout rHDL (2 mg apoA1/mL). One hundred micro-liters whole blood was incubated with 10 �L anti-humanCD14 monoclonal antibody labeled with fluorescein iso-thiocyanate and 10 �L anti-human CD4 monoclonal an-tibody labeled with phycoerythrin R (BD BiosciencesPharMingen, San Diego, CA) for 20 minutes in the darkat room temperature. Murine immunoglobulin G 2a-flu-orescein isothiocyanate and immunoglobulin G 2b-phy-coerythrin (BD Biosciences PharMingen, San Diego,CA) were used as isotype controls. Cell lysis and fixationof the preparation were then performed by the Q-prepanalyser (Beckman Coulter, Marseille, France). Flow cy-tometry was performed on an EPICS XL flow cytometerwith EXPO 32 software (Beckman Coulter, Marseille,France). The region corresponding to monocytes was de-fined using forward light scatter versus side-angle lightscatter. Results are expressed as percentages of controlvalues at baseline conditions (100%).

Monocytes SR-BI Protein Expression. Monocytesfrom patients with cirrhosis and controls were isolatedand cultivated as described previously. After supernatantrecovery, adherent cells were collected thanks to the lysisbuffer [50 mM 4-(2-hydroxyethyl)-1-piperazine ethane-

sulfonic acid (pH 7.5), 150 mM NaCl, 10% glycerol, 1%Triton X-100, 1.5 mM MgCl2, 100 mM NaF, 100 mMsodium orthovanadate, 1 mM ethylene glycol tetra-aceticacid (pH 7.7)], and immediately frozen at �80°C untilanalysis by western blot. Even if the number of cells wasthe same in each well (2.5 � 105), the protein concentra-tion was determined using the Bio-Rad protein determi-nation assay (Bio-Rad, Montlucon, France), and analiquot of homogenate was suspended in SDS samplebuffer (2% SDS, 2% �-mercaptoethanol, and 10% glyc-erol in 300 mM Tris-Cl, pH 6.8). Equal amounts of totalprotein were then subjected to SDS polyacrylamide gelelectrophoresis. Gels were transferred to nitrocellulose(Amersham Biosciences, Buckinghamshire, England).Nitrocellulose filters were blocked by incubation in 5%nonfat dry milk, 1% bovine serum albumin, 0.1 mol/LNaCl, 0.01 mol/L Tris-HCl (pH 7.5), and 0.1%Tween-20 for 2 hours. The membranes were probed withan anti–SR-BI (Interchim, Montlucon, France) antibodyovernight at 4°C. Blots were then incubated in a peroxi-dase-conjugated secondary anti-rabbit antibody at roomtemperature for 1 hour. All immunoblots were visualizedby enhanced chemiluminescence (Amersham Bio-sciences, Buckinghamshire, England). Blots were againwashed in 0.1 mol/L NaCl, 0.01 mol/L Tris-HCl (pH7.5), and 0.1% Tween 20, and immunoreactivity wasassessed using an enhanced western blot detection system.The immunoreactive bands were quantified by video den-sitometry. Values are given in arbitrary units. Results areexpressed as percentages of control values (100%).

Statistical Analysis. The values are shown asmeans � standard error of the mean. A two-way analysisof variance with repeated measures and Fisher’s t test forpaired and unpaired data were used when appropriate. AP value of �0.05 was considered significant.

Results

Patients. Twenty-four patients with cirrhosis were in-cluded. Eighteen patients had Child-Pugh C, and six pa-tients had Child-Pugh B cirrhosis. Patient characteristicsare shown in Table 1. Thirteen healthy volunteers (fourmen, seven women) were included, with a mean age of38.6 � 3.5 years.

Plasma Cholesterol, HDL Cholesterol, LDL Choles-terol, Very-Low-Density Lipoprotein Cholesterol, andTriglyceride Levels. Plasma cholesterol levels were sig-nificantly lower in patients with cirrhosis (1.08 � 0.08g/L) than in controls (1.87 � 0.12 g/L) (P � 0.0001) andsignificantly lower in patients with Child-Pugh C cirrho-sis than in those with Child-Pugh B (P � 0.05). Thedifference in plasma cholesterol in patients with Child-

178 GALBOIS ET AL. HEPATOLOGY, January 2009

Pugh B cirrhosis and controls remained significantly dif-ferent (P � 0.0016) (Fig. 1).

Apolipoprotein (Apo)-A1 plasma concentrations werealso decreased in patients with cirrhosis (0.38 � 0.12 g/Lin patients with Child-Pugh C, 1.04 � 0.17 g/L in pa-tients with Child-Pugh B cirrhosis versus 2.04 � 0.18 g/Lin controls, P � 0.0001 and P � 0.0027, respectively)with no decrease in the Apo-A1/HDL-cholesterol ratio(2.20 � 0.39 in patients with Child-Pugh C cirrhosis,2.34 � 0.34 g/L in patients with Child-Pugh B cirrhosisversus 2.52 � 0.10 g/L in controls).

Plasma LDL cholesterol levels were significantly lowerin patients with Child-Pugh C as well as in patients withChild-Pugh B cirrhosis (0.62 � 0.14 and 0.63 � 0.15g/L, respectively) than in controls (0.89 � 0.22 g/L) (P �0.03).

Plasma very-low-density lipoprotein cholesterol levelswere not different in patient with Child-Pugh C andChild-Pugh B cirrhosis and in controls (0.14 � 0.05,0.16 � 0.08, and 0.15 � 0.09 g/L, respectively).

Plasma triglyceride levels were not different in patientswith Child-Pugh C and Child-Pugh B cirrhosis and in

controls (0.69 � 0.23, 0.78 � 0.38, and 0.74 � 0.43 g/L,respectively).

Even if plasma LDL cholesterol levels were decreasedin patients with cirrhosis, the decrease in their plasmacholesterol levels was mainly attributable to a significantdecrease in HDL cholesterol concentrations (Fig. 1), asshown by a significantly increased serum cholesterol/HDL cholesterol level ratio in patients with cirrhosis (P �0.05) (Fig. 2).

Plasma LBP Concentrations. Plasma LBP concen-trations were significantly higher in patients with cirrhosisthan in controls (P � 0.0124). Plasma LBP concentra-tions were significantly higher in patients with Child-Pugh C than in those with Child-Pugh B cirrhosis (P �0.0292), which were not different from controls (Fig. 3).

Ex Vivo Effects of rHDL Incubation on Proinflam-matory Cytokines Levels and Chemokines in WholeBlood. Basal plasma levels of TNF-� and interleukin-6(IL-6) were significantly higher in patients with cirrhosisthan in controls, respectively 154% � 12% versus 100%

Fig. 1. Plasma concentrations of cholesteroland HDL in controls and patients with severecirrhosis. N � 13 for controls, n � 6 forpatients with cirrhosis classed Child-Pugh B,n � 18 for patients with cirrhosis classedChild-Pugh C. *Significantly different from con-trols. #Significantly different from patients withcirrhosis classed Child-Pugh B.

Fig. 2. Plasma cholesterol/HDL ratio in controls and patients withsevere cirrhosis. N � 13 for controls, n � 6 for patients with cirrhosisclassed Child-Pugh B, n � 18 for patients with cirrhosis classed Child-Pugh C. *Significantly different from controls. #Significantly different frompatients with cirrhosis classed Child-Pugh B.

Table 1. Characteristics of Patients with Severe Cirrhosis

Child-Pugh B(n � 6)

Child-Pugh C(n � 18)

Age (years) 53.7 � 7.5 56.2 � 2.7Sex (male/female) 4/2 11/7Etiology of cirrhosis (n)Alcohol 2 9Viral hepatitis C 2 5Viral hepatitis B 1 2Primary sclerosing cholangitis 1 0Nonalcoholic fatty liver disease 0 2Serum creatinine (�mol/L) 73 � 7 102 � 19Plasma C-reactive protein(mg/L) 7 � 2 23 � 7Blood white cells/mm3 4340 � 565 6813 � 1075Monocytes/mm3 257 � 84 742 � 63TP (% of control) 63 � 10 45 (�4)Albuminemia (g/L) 28 � 7 24 � 2Bilirubinemia (�mol/L) 67 � 44 120 � 29Child-Pugh score 8.0 � 0.6 11.6 � 0.3

Values are means � SEM or numbers of patients.

HEPATOLOGY, Vol. 49, No. 1, 2009 GALBOIS ET AL. 179

� 31% and 300% � 10% versus 100% � 10%; P �0.05 (Fig. 4A, B). Plasma levels of the 21 other cytokinesand chemokines of patients with cirrhosis and controlswere not significantly different. Incubation with rHDL alone in whole blood did not modify any cytokine or

chemokine levels. After LPS incubation in whole blood,the levels of TNF-� and IL-6 increased significantly andwere still significantly higher in patients with cirrhosisthan in controls (respectively, 323% � 16% versus 221%� 27% and 732% � 18% versus 509% � 58%, P �0.05) (Fig. 4A, B). Recombinant HDL incubation pre-vented LPS-induced TNF-� and IL-6 overproduction inpatients with cirrhosis (185% � 28% versus 323% �16% and 350% � 13% versus 732% � 18%, P � 0.05).The levels of these cytokines were not different betweenpatients with cirrhosis and controls (respectively, 185%� 28% versus 161% � 17% and 350% � 13% versus391% � 16%, P � 0.05) (Fig. 4A, B).

Ex Vivo Effects of rHDL Incubation on TNF-�Production after LPS Stimulation in Isolated Mono-cytes. Basal TNF-� concentrations in the supernatantwere not different between controls and patients withcirrhosis (Table 2). Two hours after incubation with cul-ture medium alone, TNF-� concentrations were not sig-nificantly different between controls and patients withcirrhosis. Addition of rHDL (2 mg apoA1/mL) had noeffect on TNF-� production. After 24 hours, with orwithout HDL, there was no significant difference be-tween TNF-� concentrations in controls and patientswith cirrhosis (Table 2).

After 2 hours’ incubation with LPS, TNF-� produc-tion in the supernatant was significantly increased in allgroups (P � 0.0001) (Table 2). TNF-� concentrationswere not significantly different between controls and pa-tients with Child-Pugh B cirrhosis; however, TNF-� con-centrations were significantly higher in those with Child-

Fig. 3. Plasma LBP concentrations in whole blood from controls andpatients with severe cirrhosis. N � 6 for controls, n � 6 for patients withcirrhosis classed Child-Pugh B, n � 6 for patients with cirrhosis classedChild-Pugh C. *Significantly different from controls.

Fig. 4. Ex vivo effects of rHDL Incubation on TNF-� (A) and IL-6 (B)concentrations in whole blood from controls and patients with severecirrhosis. N � 4 for controls and n � 4 for patients with cirrhosis.*Significantly different from controls. $Significantly different from thesame group at basal conditions. §Significantly different from the samegroup after LPS incubation

Table 2. Effect of rHDL on LPS-Induced TNF-alphaProduction in Cirrhotic Monocytes

TNF-alpha Concentrationin the Supernatant (pg/mL)

Time of Incubation Controls Child-Pugh B Child-Pugh C

0 21 � 5 17 � 6 22 � 82 Hours

Basal 47 � 12 56 � 32 39 � 11rHDL 23 � 5 29 � 17 24 � 6LPS 1689 � 224† 1263 � 155† 3497 � 332*#LPS � rHDL 295 � 45†‡ 171 � 36†‡ 459 � 141†

24 HoursBasal 58 � 20 78 � 43 31 � 6rHDL 29 � 8 18 � 10 11 � 3LPS 2393 � 484† 2002 � 331† 3717 � 1321†LPS�rHDL 337 � 94†‡ 31 � 21‡ 359 � 119†‡

n � 9 for controls, n � 6 for patients with cirrhosis classed Child-Pugh B, n �6 for patients with cirrhosis classed Child-Pugh C.

*Significantly different from controls in the same conditions.†Means significantly different with the same group at basal conditions.‡Means significantly different with the same group after LPS incubation.

180 GALBOIS ET AL. HEPATOLOGY, January 2009

Pugh C cirrhosis (P � 0.0002). After 2 hours’ rHDL (2mg apoA1/mL) and LPS incubation, TNF-� productionwas significantly decreased (P � 0.0001) and was notsignificantly different between controls, patients withChild-Pugh B, and Child-Pugh C cirrhosis. After 24hours’ incubation with LPS, TNF-� production was alsoincreased (P � 0.0001) and was not different in the threegroups. Addition of rHDL significantly decreased TNF-�production in all groups (P � 0.0001) (Table 2).

Moreover, a dose titration of rHDL in isolated mono-cytes from patients with Child-Pugh C cirrhosis was per-formed. After 2 hours’ incubation with LPS and fourdifferent concentrations of rHDL (0, 0.5, 1, and 2 mgapoA1/mL), TNF-� productions were determined(3497 � 332, 1658 � 164, 593 � 181, and 459 � 141pg/mL, respectively). Other experiments were performedwith the dose with the optimal effect (2 mg apoA1/mL).

Ex Vivo Effects of rHDL Incubation on IL-10 Pro-duction After LPS Stimulation in Isolated Monocytes.Basal IL-10 concentrations in the supernatant were notdifferent between controls and patients with cirrhosis(Fig. 5). Twenty-four hours after incubation with culturemedium alone, IL-10 concentrations were not signifi-cantly different between controls and patients with cir-rhosis. Addition of rHDL had no effect on IL-10production (Fig. 5).

After 24 hours’ incubation with LPS, IL-10 produc-tion in the supernatant was significantly increased in allgroups (P � 0.0001) (Fig. 5). IL-10 concentrations weresignificantly lower in patients with Child-Pugh C cirrho-sis than in patients with Child-Pugh B cirrhosis (10.68 �

2.62 versus 23.00 � 4.21 pg/mL, respectively, P � 0.03)and lower in patients with Child-Pugh B cirrhosis than incontrols (23.00 � 4.21 versus 48.23 � 11.20 pg/mLrespectively, P � 0.01). After 24 hours’ rHDL and LPSincubation, IL-10 production was significantly decreased(P � 0.0001) and was not significantly different betweencontrols (5.96 � 0.89 pg/mL), patients with Child-PughB cirrhosis (3.58 � 1.05 pg/mL), and patients withChild-Pugh C cirrhosis (2.29 � 1.20 pg/mL) (Fig. 5).

Effects of rHDL Incubation on LPS-InducedNF-�B Pathway Activation in Isolated Monocytes.Under baseline conditions, I�B-� concentration inmonocytes was significantly lower in patients with cirrho-sis than in controls (55 � 26 versus 100 � 85%, P �0.02), meaning an increased NF-�B activation (Fig. 6).Incubation with rHDL during 2 hours had no effect onI�B-� concentration in monocytes. After 2 hours’ LPSincubation, I�B-� concentration was significantly de-creased in monocytes from controls (58% � 9.6%, P �0.01), whereas it was already resynthesized in monocytesfrom patients with cirrhosis (99% � 33%) (Fig. 6). Ad-dition of rHDL prevents any LPS-induced modificationof I�B-�, which remained at basal concentration (Fig. 6).

Ex Vivo Effects of rHDL Incubation on CD14 Ex-pression in Monocytes. Under baseline conditions,CD14 expression in monocytes was significantly higher inpatients with cirrhosis than in controls (Fig. 7). Incuba-tion with rHDL did not affect CD14 expression in mono-cytes from patients with cirrhosis or controls. After LPSincubation, CD14 expression in monocytes was signifi-

Fig. 5. Effects of rHDL incubation on LPS-induced IL-10 production incirrhotic monocytes. N � 9 for controls, n � 6 for patients with cirrhosisclassed Child-Pugh B, n � 6 for patients with cirrhosis classed Child-Pugh C. *Significantly different from controls in the same conditions.#Significantly different from the same group at basal conditions. $Sig-nificantly different from the same group after LPS incubation.

Fig. 6. Effects of rHDL and LPS incubation on NF-�B activation bymeasuring I�B-� in monocytes from patients with cirrhosis and controlsby western blotting. I�B-� degradation means NF-�B activation. *Sig-nificantly different from controls in the same conditions. #Significantlydifferent from the same group at basal conditions. $Significantly differentfrom the same group after LPS incubation.

HEPATOLOGY, Vol. 49, No. 1, 2009 GALBOIS ET AL. 181

cantly increased in patients with cirrhosis and controls(P � 0.0001 and P � 0.0002, respectively) (Fig. 7). Inmonocytes from controls, the increase in CD14 expres-sion after incubation with rHDL and LPS was signifi-cantly lower than that observed after LPS incubationalone (P � 0.0188). In monocytes from patients withcirrhosis, incubation with rHDL and LPS did not in-crease CD14 expression, which remained at baseline lev-els. This value was significantly lower than that observedafter LPS incubation alone (P � 0.0001)(Fig. 7).

Ex Vivo SR-BI Expressions in Monocytes. SR-BIexpressions were not different in monocytes from patientswith advanced cirrhosis and controls at basal conditionsor at 24 hours without incubation with LPS (Fig. 8). After24 hours’ incubation with LPS, SR-BI expressions weresignificantly decreased in monocytes from controls(57.0% � 4.1%, P � 0.004) and were not modified inmonocytes from patients with cirrhosis.

DiscussionThis study: (1) confirms that in patients with advanced

cirrhosis, proinflammatory cytokines and chemokineswere spontaneously produced in whole blood in absenceof LPS stimulation, and overproduced after LPS chal-lenge; (2) shows that this overproduction of proinflam-matory cytokines and chemokines after LPS challenge wasabolished by incubation of whole blood with rHDL, andthat the enhanced responsiveness to LPS (CD14 expres-sion and TNF-� production) was also abolished by rHDLincubation in monocytes isolated from patients with ad-vanced cirrhosis; (3) shows that after LPS challenge, theclassical down-regulation of HDL receptor SR-BI wasabolished in patients with cirrhosis, suggesting an expla-nation for the decreased HDL neutralization of LPS in

those patients. These findings confirm in vivo studies incirrhotic rats in which rHDL administration limited liverproinflammatory responses.21

Decreased plasma HDL cholesterol concentrationswere the major cause of hypocholesterolemia in patientswith advanced cirrhosis. These results suggest that LPSneutralization by HDL is reduced in patients with ad-vanced cirrhosis. In case of bacterial translocation, thereduction of plasma HDL could play a role in the activa-tion of monocytes.12 The lack of synthesis of HDL in theliver can explain the decrease in HDL plasmatic levels inpatients with cirrhosis and the following decrease in LPSneutralization. Moreover, we found that the HDL recep-tor SR-BI was not down-regulated in patients with cirrho-sis despite their lower HDL cholesterol level. This couldcontribute to the decrease in HDL cholesterol plasmaticlevels in patients with cirrhosis. The SR-BI mediates se-lective uptake of HDL cholesteryl ester. In the liver,SR-BI is highly expressed, recognizes HDL, and allowsthe control of cholesterolemia by eliminating it in bile.When HDL binds LPS, the HDL–LPS complex is alsorecognized by SR-BI in the liver, and LPS is eliminated inbile.25 SR-BI is also known to be expressed in monocytes.SR-BI expression on human liver, monocytes, and mac-rophages is decreased by LPS and TNF incubation.24

Thus, it has been suggested that suppression of SR-BIexpression could help stabilize the lipoprotein concentra-tions in the blood compartment that decrease during sep-sis,30 and thus is important for host defense against LPS.In the current study, SR-BI expression was not down-regulated in monocytes from patients with severe cirrhosisafter LPS incubation. This abnormal lack of down-regu-lation in monocytes from patients with cirrhosis couldmake the lipidic profiles of these patients worse in a case ofgram-negative infection and thus contribute to the de-crease of LPS neutralization in those patients.

Fig. 8. Effect of LPS incubation on SR-BI expression in monocytesfrom controls and patients with severe cirrhosis. N � 6 for controls, n �6 for patients with cirrhosis. *Significantly different from controls at basalconditions.

Fig. 7. Effects of rHDL incubation on CD14 expression in monocytesfrom controls and patients with severe cirrhosis. N � 6 for controls, n �8 for patients with cirrhosis. *Significantly different from controls in thesame conditions. #Significantly different from the same group at basalconditions. $Significantly different from the same group after LPS incu-bation.

182 GALBOIS ET AL. HEPATOLOGY, January 2009

After LPS incubation, cytokine production in wholeblood was significantly more marked in patients with ad-vanced cirrhosis than in controls. This effect was similarto a previous study.9 Recombinant HDL incubation lim-ited the inflammatory response in both controls and cir-rhosis, and cytokine concentrations were similar in bothgroups. These results again suggest that increasing HDLplasma concentrations in patients with advanced cirrhosiscould help in gram-negative infection by limiting the ex-aggerated proinflammatory response in these patients.

Increased plasma LBP concentrations were observed inpatients with Child-Pugh C cirrhosis, which also has beenfound in previous studies.31,32 Albillos et al.31,32 have sug-gested that the elevated LBP plasma concentrations inpatients with cirrhosis and ascites could predict severebacterial infection and reflect bacterial translocation. Inthe current study, although all patients had ascites, in-creased LBP plasma concentrations were only observed inpatients with Child-Pugh C cirrhosis but not in thosewith Child-Pugh B or healthy volunteers. These findingssuggest that an increase in plasma concentrations of LBPis also associated with the severity of cirrhosis. However,the role of LBP is not clearly elucidated. It has been de-scribed to have a pro-inflammatory effect (facilitating LPSbinding on CD14),33 and an anti-inflammatory effect (fa-cilitating LPS neutralization by HDL).34 Therefore, be-cause LBP concentrations were higher in patients withcirrhosis than in controls, we also studied TNF-� produc-tion by isolated monocytes, to ensure that LPS and rHDLeffect on cytokine production was not influenced by LBPconcentrations.

The overproduction of proinflammatory cytokines inisolated monocytes from patients with advanced cirrhosishas already been reported after LPS incubation10; how-ever, this study shows that an overproduction of TNF-�was observed only in patients with Child-Pugh C and notin those with Child-Pugh B cirrhosis. After LPS incuba-tion, TNF-� production was not different in monocytesfrom patients with Child-Pugh B cirrhosis and controls.These monocytes do not seem to be activated, unlikethose in patients with Child-Pugh C cirrhosis. Neverthe-less, patients with Child-Pugh B cirrhosis had decreasedplasma HDL concentrations and thus a reduced capacityfor LPS neutralization. The normal monocyte response toLPS incubation could be attributable to decreased bacte-rial translocation in these patients, as suggested by theirnormal plasma LBP concentrations. The neutralization ofLPS by rHDL caused a decrease in TNF-� production inmonocytes from patients with advanced cirrhosis andcontrols. In these conditions, monocytes from patientswith Child-Pugh C cirrhosis did not produce moreTNF-� than those with Child-Pugh B cirrhosis or con-

trols. This new result again indicates that rHDL admin-istration might be beneficial in patients with Child-PughC cirrhosis and a gram-negative infection. CD14 expres-sion was increased under baseline conditions in mono-cytes from patients with advanced cirrhosis. This result issimilar to that of a previous study that showed that solubleCD14 was increased in the plasma of patients with cir-rhosis.31 The increase in CD14 could be attributable toincreased plasma TNF-� concentrations in these patientsbecause it is already known that TNF-� increases mono-cyte CD14 expression.35 Our results showed that rHDLincubation has no effect on basal CD14 expression inmonocytes. An increase in CD14 expression in mono-cytes after LPS incubation has already been observed.35

The current study also found that rHDL incubation pre-vents LPS-induced increased CD14 expression. CD14 isthe main LPS receptor on the membrane of monocytes,and transgenic mice overexpressing CD14 are known tobe hypersensitive to LPS,22, whereas CD14-deficientmice are resistant to LPS.23 Thus, this could be anotherprotective effect of increased HDL plasma concentrationsin patients with advanced cirrhosis in cases of gram-neg-ative infection.

Some results of this study help to understand howthese beneficial effects of rHDL work. First, rHDL incu-bation in isolated monocytes does not increase IL-10 pro-duction and thus has no anti-inflammatory effect.Recombinant HDL incubation prevents LPS-inducedIL-10 production in isolated monocytes. Thus, rHDLseems to prevent LPS-induced pro-inflammatory(TNF-�) as well as anti-inflammatory (IL-10) cytokineproduction by monocytes by neutralizing LPS before itcould stimulate monocytes. Secondly, this is confirmedby the lack of LPS-induced NF-�B activation in mono-cytes in case of rHDL incubation. At last, rHDL incuba-tion prevents LPS-induced TNF-� production evenwhen its receptor on monocytes (SR-BI) is inactivated byblocking antibodies (data not shown). This result con-firms that rHDL’s beneficial effects must be related to theLPS neutralization and not to a direct effect of rHDL onmonocytes.

In conclusion, the current study showed that proin-flammatory cytokines were spontaneously expressed inmonocytes from patients with advanced cirrhosis. Theoverproduction of the proinflammatory cytokines in re-sponse to LPS incubation was abolished by HDL incuba-tion. The enhanced responsiveness to LPS was alsoabolished by HDL in isolated monocytes from patientswith advanced cirrhosis. These results suggest that theadministration of rHDL might be considered a usefulstrategy for the treatment of cirrhosis to increase LPSneutralization and to prevent septic shock.

HEPATOLOGY, Vol. 49, No. 1, 2009 GALBOIS ET AL. 183

Acknowledgment: Recombinant HDL was suppliedby CSL Behring, Bern, Switzerland.

References1. Bone RC. The pathogenesis of sepsis. Ann Intern Med 1991;115:457-469.2. Borrelli E, Roux-Lombard P, Grau GE, Girardin E, Ricou B, Dayer J, et al.

Plasma concentrations of cytokines, their soluble receptors, and antioxi-dant vitamins can predict the development of multiple organ failure inpatients at risk. Crit Care Med 1996;24:392-397.

3. Giron-Gonzalez JA, Martinez-Sierra C, Rodriguez-Ramos C, Macias MA,Rendon P, Diaz F, et al. Implication of inflammation-related cytokines inthe natural history of liver cirrhosis. Liver Int 2004;24:437-445.

4. Moreau R, Hadengue A, Soupison T, Kirstetter P, Mamzer MF, Vanjak D,et al. Septic shock in patients with cirrhosis: hemodynamic and metaboliccharacteristics and intensive care unit outcome. Crit Care Med 1992;20:746-750.

5. Angus DC, Linde-Zwirble WT, Lidicker J, Clermont G, Carcillo J, PinskyMR. Epidemiology of severe sepsis in the United States: analysis of inci-dence, outcome, and associated costs of care. Crit Care Med 2001;29:1303-1310.

6. Byl B, Roucloux I, Crusiaux A, Dupont E, Deviere J. Tumor necrosisfactor alpha and interleukin 6 plasma levels in infected cirrhotic patients.Gastroenterology 1993;104:1492-1497.

7. Heller J, Sogni P, Barriere E, Tazi KA, Chauvelot-Moachon L, GuimontMC, et al. Effects of lipopolysaccharide on TNF-alpha production, hepaticNOS2 activity, and hepatic toxicity in rats with cirrhosis. J Hepatol 2000;33:376-381.

8. Tazi KA, Bieche I, Paradis V, Guichard C, Laurendeau I, Dargere D, et al.In vivo altered unfolded protein response and apoptosis in livers fromlipopolysaccharide-challenged cirrhotic rats. J Hepatol 2007;46:1075-1088.

9. Deviere J, Content J, Denys C, Vandenbussche P, Schandene L, Wybran J,et al. Excessive in vitro bacterial lipopolysaccharide-induced production ofmonokines in cirrhosis. HEPATOLOGY 1990;11:628-634.

10. Riordan SM, Skinner N, Nagree A, McCallum H, McIver CJ, Kurtovic J,et al. Peripheral blood mononuclear cell expression of toll-like receptorsand relation to cytokine levels in cirrhosis. HEPATOLOGY 2003;37:1154-1164.

11. Tazi KA, Moreau R, Herve P, Dauvergne A, Cazals-Hatem D, Bert F, et al.Norfloxacin reduces aortic NO synthases and proinflammatory cytokineupregulation in cirrhotic rats: role of Akt signaling. Gastroenterology2005;129:303-314.

12. Tazi KA, Quioc JJ, Saada V, Bezeaud A, Lebrec D, Moreau R. Upregula-tion of TNF-alpha production signaling pathways in monocytes frompatients with advanced cirrhosis: possible role of Akt and IRAK-M.J Hepatol 2006;45:280-289.

13. Ulevitch RJ, Johnston AR, Weinstein DB. New function for high densitylipoproteins: isolation and characterization of a bacterial lipopolysaccha-ride-high density lipoprotein complex formed in rabbit plasma. J ClinInvest 1981;67:827-837.

14. Zambon S, Orlando R, Sartore G, Bassi A, Manzato E, Crepaldi G. Thelipoprotein composition of plasma and ascitic fluid in liver cirrhosis. EurJ Clin Invest 1995;25:143-148.

15. Fraunberger P, Pilz G, Cremer P, Werdan K, Walli AK. Association ofserum tumor necrosis factor levels with decrease of cholesterol duringseptic shock. Shock 1998;10:359-363.

16. Hubsch AP, Casas AT, Doran JE. Protective effects of reconstituted highdensity lipoprotein in rabbit gram-negative bacteremia models. J Lab ClinMed 1995;126:548-558.

17. Levine DM, Parker TS, Donnelly TM, Walsh A, Rubin AL. In vivo pro-tection against endotoxin by plasma high density lipoprotein. Proc NatlAcad Sci U S A 1993;90:12040-12044.

18. McDonald MC, Dhadly P, Cockerill GW, Cuzzocrea S, Mota-Filipe H,Hinds CJ, et al. Reconstituted high-density lipoprotein attenuates organinjury and adhesion molecule expression in a rodent model of endotoxicshock. Shock 2003;20:551-557.

19. Pajkrt D, Doran JE, Koster F, Lerch PG, Arnet B, van der Poll T, et al.Antiinflammatory effects of reconstituted high-density lipoprotein duringhuman endotoxemia. J Exp Med 1996;184:1601-1608.

20. Ramirez MJ, Ibanez A, Navasa M, Casals E, Morales-Ruiz M, Jimenez W,et al. High-density lipoproteins reduce the effect of endotoxin on cytokineproduction and systemic hemodynamics in cirrhotic rats with ascites.J Hepatol 2004;40:424-430.

21. Thabut D, Tazi KA, Bonnefont-Rousselot D, Aller M, Farges O, GuimontMC, et al. High-density lipoprotein administration attenuates liver proin-flammatory response, restores liver endothelial nitric oxide synthase activ-ity, and lowers portal pressure in cirrhotic rats. HEPATOLOGY 2007;46:1893-1906.

22. Ferrero E, Jiao D, Tsuberi BZ, Tesio L, Rong GW, Haziot A, et al.Transgenic mice expressing human CD14 are hypersensitive to lipopoly-saccharide. Proc Natl Acad Sci U S A 1993;90:2380-2384.

23. Haziot A, Ferrero E, Kontgen F, Hijiya N, Yamamoto S, Silver J, et al.Resistance to endotoxin shock and reduced dissemination of gram-nega-tive bacteria in CD14-deficient mice. Immunity 1996;4:407-414.

24. Buechler C, Ritter M, Quoc CD, Agildere A, Schmitz G. Lipopolysaccha-ride inhibits the expression of the scavenger receptor Cla-1 in humanmonocytes and macrophages. Biochem Biophys Res Commun 1999;262:251-254.

25. Vishnyakova TG, Bocharov AV, Baranova IN, Chen Z, Remaley AT,Csako G, et al. Binding and internalization of lipopolysaccharide by Cla-1,a human orthologue of rodent scavenger receptor B1. J Biol Chem 2003;278:22771-22780.

26. Camps J, Simo JM, Guaita S, Ferre N, Joven J. Altered composition oflipoproteins in liver cirrhosis compromises three homogeneous methodsfor DL cholesterol. Clin Chem 1999;45:685-688.

27. Gomez F, Camps J, Simo JM, Ferre N, Joven J. Agreement study ofmethods based on the elimination principle for the measurement of LDL-and HDL-cholesterol compared with ultracentrifugation in patients withliver cirrhosis. Clin Chem 2000;46:1188-1191.

28. Friedewald WT, Levy RI, Fredrickson DS. Estimation of the concentra-tion of low-density lipoprotein cholesterol in plasma, without use of thepreparative ultracentrifuge. Clin Chem 1972;18:499-502.

29. Baldwin AS, Jr. The NF-kappa B and I kappa B proteins: new discoveriesand insights. Annu Rev Immunol 1996;14:649-683.

30. Fraunberger P, Schaefer S, Werdan K, Walli AK, Seidel D. Reduction ofcirculating cholesterol and apolipoprotein levels during sepsis. Clin ChemLab Med 1999;37:357-362.

31. Albillos A, de la Hera A, Gonzalez M, Moya JL, Calleja JL, Monserrat J, etal. Increased lipopolysaccharide binding protein in cirrhotic patients withmarked immune and hemodynamic derangement. HEPATOLOGY 2003;37:208-217.

32. Albillos A, de-la-Hera A, Alvarez-Mon M. Serum lipopolysaccharide-bind-ing protein prediction of severe bacterial infection in cirrhotic patients withascites. Lancet 2004;363:1608-1610.

33. Kitchens RL, Thompson PA. Modulatory effects of sCD14 and LBP onLPShost cell interactions. J Endotoxin Res 2005;11:225-229.

34. Wurfel MM, Kunitake ST, Lichenstein H, Kane JP, Wright SD. Lipopoly-saccharide (LPS)-binding protein is carried on lipoproteins and acts as acofactor in the neutralization of LPS. J Exp Med 1994;180:1025-1035.

35. Fearns C, Kravchenko VV, Ulevitch RJ, Loskutoff DJ. Murine CD14 geneexpression in vivo: extramyeloid synthesis and regulation by lipopolysac-charide. J Exp Med 1995;181:857-866.

184 GALBOIS ET AL. HEPATOLOGY, January 2009