Research Article

High-volume plasma exchange in patients with acute liver failure:An open randomised controlled trial

Fin Stolze Larsen1,⇑, Lars Ebbe Schmidt1, Christine Bernsmeier2, Allan Rasmussen3,Helena Isoniemi4, Vishal C. Patel2, Evangelos Triantafyllou2, William Bernal2, Georg Auzinger2,

Debbie Shawcross2, Martin Eefsen1, Peter Nissen Bjerring1, Jens Otto Clemmesen1,Krister Hockerstedt4, Hans-Jørgen Frederiksen5, Bent Adel Hansen1,

Charalambos G. Antoniades2,6,y, Julia Wendon2,y

1Department of Hepatology, Rigshospitalet, Copenhagen, Denmark; 2Institute of Liver Studies, King’s College Hospital, London,United Kingdom; 3Department of Surgery and Liver Transplantation C, Rigshospitalet, Copenhagen, Denmark; 4Transplantation and

Liver Surgery Clinic, Helsinki University Hospital, Finland; 5Department of Anaestesia AN-2041, Rigshospitalet, Copenhagen, Denmark;6Section of Hepatology, St. Mary’s Hospital, Imperial College London, London, UK

See Editorial, pages 10–12

Background & Aims: Acute liver failure (ALF) often results in car-diovascular instability, renal failure, brain oedema and deatheither due to irreversible shock, cerebral herniation or develop-ment of multiple organ failure. High-volume plasma exchange(HVP), defined as exchange of 8–12 or 15% of ideal body weightwith fresh frozen plasma in case series improves systemic, cere-bral and splanchnic parameters.Methods: In this prospective, randomised, controlled, multicen-tre trial we randomly assigned 182 patients with ALF to receiveeither standard medical therapy (SMT; 90 patients) or SMT plusHVP for three days (92 patients). The baseline characteristics ofthe groups were similar. The primary endpoint was livertransplantation-free survival during hospital stay. Secondary-endpoints included survival after liver transplantation with orwithout HVP with intention-to-treat analysis. A proof-of-principle study evaluating the effect of HVP on the immune cellfunction was also undertaken.Results: For the entire patient population, overall hospital sur-vival was 58.7% for patients treated with HVP vs. 47.8% for thecontrol group (hazard ratio (HR), with stratification for liver

Journal of Hepatology 20

Keywords: Fulminant hepatic failure; Plasmapheresis; Artificial liver support;Cerebral oedema; Ammonia; Sepsis; Multiorgan failure; Liver transplantation;Hepatic encephalopathy.Received 25 December 2014; received in revised form 25 July 2015; accepted 11August 2015; available online 29 August 2015qDOI of original article: http://dx.doi.org/10.1016/j.jhep.2015.09.010.⇑ Corresponding author. Address: Department of Hepatology, A-2121 Rigshospi-talet, Univ. Copenhagen, Blegdamsvej 9, 2100 Copenhagen, Denmark.E-mail address: stolze@post3.tele.dk (F.S. Larsen).

y These authors share last authorship.Abbreviations: ALF, acute liver failure; ALT, alanine-aminotransferase; CI,confidence interval; CLIF, Chronic liver failure consortium; FiO2, fractionalinspired oxygen; HR, hazard ratio; ICP, intracranial pressure; ITT, intention totreat; NH3, ammonia; LPS, lipopolysaccharide; PE, plasma exchange; RR,respiratory rate; IQR, interquartile range; SD, standard deviation; SMT, standardmedical treatment; SIRS, systemic inflammatory response syndrome; SOFA,sequential organ failure score.

transplantation: 0.56; 95% confidence interval (CI), 0.36–0.86;p = 0.0083). HVP prior to transplantation did not improve survivalcompared with patients who received SMT alone (CI 0.37 to 3.98;p = 0.75). The incidence of severe adverse events was similar in thetwo groups. Systemic inflammatory response syndrome (SIRS)and sequential organ failure assessment (SOFA) scores fell in thetreated group compared to control group, over the study period(p <0.001).Conclusions: Treatment with HVP improves outcome in patientswith ALF by increasing liver transplant-free survival. This is attri-butable to attenuation of innate immune activation and amelio-ration of multi-organ dysfunction.� 2015 European Association for the Study of the Liver. Publishedby Elsevier B.V. All rights reserved.

Introduction

Acute liver injury is most often associated with discrete non-specific symptoms and mildly elevated liver function tests. Insevere cases progressing to acute liver failure (ALF), the consciouslevel decreases, and brain oedema, hypoglycaemia, and extrahepatic organ failure evolves [1].

The exact pathophysiology for development of multi-organ dys-function (MOF) in ALF remains elusive. There is an accumulation ofvarious metabolites and toxins varying in size, distribution volume,lipophilicity, and protein binding [2,3]. Studies in ALF patients alsoindicate that a decreased hepatic capacity for synthesis of coagula-tion factors, complement, and lipoproteins may be of importance inthe evolution of MOF. The importance of systemic inflammatoryresponses (SIRS) in the outcome in ALF has been established forover a decade, being associated with MOF, progression inencephalopathy and increased mortality [4–7]. Recent evidenceindicates that following overwhelming hepatocyte death, there isthe release of damage associated molecular patterns (DAMPS),

16 vol. 64 j 69–78

Research Article

e.g. DNA, histones, HMGB-1, trigger Toll-like receptor (TLR) depen-dent activation of innate immune cells, hepatic and systemicinflammatory responses [1,4].

Survival with medical management has increased over thelast three decades but a significant mortality remains [2,3]. Themanagement strategy of ALF patients is to restore and preservevital organ function and mitigate or limit the progression ofMOF until either spontaneous liver regeneration occurs or a suit-able donor liver becomes available for emergency liver transplan-tation in those identified to be likely non-survivors [3,8,9]. In thiscontext an effective liver assist procedure may improve survivalfor those with ALF deemed not appropriate for liver transplanta-tion, despite poor prognostic criteria and/or providing greaterstability whilst awaiting transplantation.

Any intervention that aims to replace the failing liver shouldsecure metabolic and excretory functions, replace liver-derivedproteins and peptides and attenuate the severity of any innateimmune activation following acute hepatocyte injury [10].

Most studies of extracorporeal liver support have been basedupon dialysis techniques assuming that ALF can be treated by adevice that corrects blood’s composition. Thus far, such proce-dures have been of limited success and currently no study hasbeen able to demonstrate an improvement in survival in patientswith ALF [11–13].

Plasma replacement or exchange therapy with fresh frozenplasma is an established therapy used for otherimmunologically-driven disorders [14]. Case series of high-volume plasma exchange (HVP) in patients with ALF have beenshown to be safe [15,16], to decrease the severity of hepaticencephalopathy, decrease vasopressor requirements [17,18]and with a signal of accelerated hepatic nitrogen turnover [19,20].

The mechanism of action of therapeutic plasma exchange inameliorating the course of various diseases is putative and per-tains to the removal of plasma cytokines and adhesion molecules,replacement of plasma factors, and immune modulation. In thisstudy we hypothesised that HVP would reduce mortality inpatients with ALF by attenuating the development of MOF. Theprimary endpoint was transplant-free survival during the hospitalstay. The secondary endpoint included survival after transplanta-tion with or without HVP using an intention-to-treat analysis. Inaddition we performed a proof-of-principle study evaluating theeffects of HVP on circulating immune cells and leucocyte subsets.

Patients and methods

This study was conducted in two parts. Study A: ALF patients randomised to stan-dard medical treatment (SMT) and to SMT plus HVP. Study B: examined the effectof plasma obtained before and after the first HVP session from ALF patients on thecirculating immune cell subsets innate and native immune system.

Study A

Trial designThe eligibility criteria were age greater than 18 years and a diagnosis of ALF withat least grade 2 hepatic encephalopathy [21]. All aetiologies including acuteWilson’s disease, Budd-Chiari syndrome and acute presentation of autoimmunehepatitis (without clinical or radiological evidence of cirrhosis or chronic liverinjury in the later two cohorts) were eligible for inclusion. Twenty-six patientswere listed for liver transplantation (LTx) in the HVP treated group vs. 30 in thein the control group (NS). Before enrolment into the study, all patients had reviewof possible aetiological factors and clinical examination. A diagnostic ALF screenwas secured incorporating serological tests, hepatitis serology, autoimmunemarkers, and microbiological cultures, in addition to ultrasound imaging and/orCT imaging as determined by the treating clinician.

70 Journal of Hepatology 20

Exclusion criteria included withdrawal of assent, alcoholic hepatitis, primarynon-function of liver graft or graft dysfunction, any form of known chronic liverdisease, known malignancies, liver resections with liver failure, hypoxic hepatitisand malignancy presenting as ALF. Patients with clinical suspicion of irreversiblebrain injury (fixed dilated pupils unresponsive to standard interventions) or braindeath were excluded from enrolment.

The protocol was approved by the institutional review board or ethics com-mittee at each centre alongside appropriate adherence to all applicable lawsand regulations governing clinical research. Assent/consent for entry into the trialwas obtained from the patient́s nominated representative.

RandomisationEntry into the study was required within 24 h of the development of grade 2encephalopathy. When a patient was identified as suitable for study entry, aresearch co-ordinator was telephoned in the co-ordinating centre (Copenhagen)and SMT or SMT plus HVP was defined from a pre-made randomization code thatwas blinded to the co-ordinator (transplant coordinating staff) and the investiga-tors by using opaque envelopes.

Eligible and consented patients were randomly assigned, in a 1:1 ratio, toreceive either SMT or SMT plus HVP with no block randomisation. The studywas not blinded.

At entry to intensive care, at randomization, study entry and during the studyperiod, patients were defined as fulfilling poor prognostic criteria based on theKing’s College Criteria [1–4]. Reasons for not being active on the urgent transplantlist despite fulfilling poor prognostic criteria were documented.

Patients and methodsAll patients were treated in three specialised liver intensive care units with SMT asoutlined in the study protocol (ClinicalTrials Gov Number NCT00950508). Trans-fusion triggers were as per local guidelines, coagulation support using plasmawas not routinely offered in the SMT group. Use of intravenous insulin, hypertonicdextrose and enteral feeding were all utilised to maintain euglycaemia.

Intubation and ventilation was undertaken for standard indications in addi-tion to development of grade 3 encephalopathy, with infusion of sedation (propo-fol or midazolam and opiates) to facilitate effective ventilation. H2 antagonist orproton pump inhibitors, antibiotics and antifungal regimes followed local guide-lines with microbiological input.

Advanced haemodynamic monitoring was undertaken, with clinical goals ofeuvolaemia with mean arterial pressure maintained at greater than 60 mmHg.First choice of pressor was norepinephrine, with dobutamine as first line ino-trope. The use of dopamine, adrenaline and terlipressin were allowed as addi-tional vasopressors.

N-Acetylcysteine, was given to all patients irrespective of aetiology [22] andgiven for a maximum of five days. The primary mode of renal replacement therapywas continuous haemofiltration. Intermittent haemodialysis was allowed only ifthe patient showed haemodynamically stability and there was no concern regard-ing cerebral deterioration or cerebral oedema. Anticoagulation of extracorporealcircuits followed local guidelines with use of low dose heparin, epoprostenol ornothing. Cerebral CT scan was not required before enrolment into the study.

Intracranial pressure (ICP) monitoring was instituted if the attending clini-cian felt it was indicated. An episode with a sustained increase in ICP >20 mmHgfor more than 5 min was defined as intracranial hypertension.

Plasma exchange procedureThe volume of HVP exchanged was stipulated as 15% of ideal body weight (rep-resenting 8–12 L per day per procedure); patient plasma was removed at a rate of1–2 L per hour with replacement with fresh frozen plasma in equivalent volume.The HVP procedure was undertaken on three consecutive days but with no fixedtime interval between each treatment.

Data collectionPhysiological and laboratory parameters were collected prospectively during a12 year period (1998–2010), allowing retrospective calculation of both sequentialorgan failure assessment score (SOFA) score [23] and CLIF-SOFA score [24] andSIRS parameters [25]. Outcome was defined as survival to hospital discharge ordeath; data was collected on a predefined CRF.

StatisticsThe sample size calculation was based upon retrospective observations of mortal-ity of patients with ALF. Assuming a 20% improvement in survival with the inter-vention, for a significance level of rate of 5% with a beta-value of 20% wecalculated that 182 patients should be enrolled. Two predefined interim analyseswere planned at 60 and 120 patients examining safety and futility endpoints.

16 vol. 64 j 69–78

JOURNAL OF HEPATOLOGY

The main analysis was performed in respect to the primary endpoint of

transplant-free survival (determined at the time of discharge from hospital). Cen-soring was defined as the time-point liver transplantation was performed.

Student’s t test was used to compare baseline data between the two ran-domised groups of patients. For data that did not follow a normal distributionthe significance of differences were tested using Mann–Whitney (in-betweengroups) or Wilcoxon tests (matched data). Intention-to-treat (ITT) analysis wasused to compare the survival rates in the two groups. Data are presented as meanand standard deviation (SD) when normally distributed data and otherwise asmedian and interquartile range (IQR). X2 tests and Kaplan-Meier curves examinedsurvival using Cox analysis.

Study B

In this proof-of-principle study the effect of HVP on circulating innate immunecell function was studied in eight ALF patients who underwent early HVP (within48 h following ICU admission), 12 ALF patients undergoing late HVP (>48 h) and11 control ALF patients receiving best supportive care.

Blood samples were taken before and approximately 12 h after the first HVP;control samples were taken at admission and on day 3–5. Full blood count, inter-national normalized ratio (INR), liver and renal function tests, lactate, ammoniaand clinical variables were entered into a database.

Immunophenotyping and function of monocytes, lymphocytes and neu-trophils were performed following 24 h culture of healthy peripheral bloodmononuclear cells in X-vivo medium (Lonza, Switzerland) and 25% plasmaderived from ALF patients. Monocyte TNFa/IL-6 production in response to LPS(100 ng/ml; 6 h) and immunophenotyping (CD14, CD16, HLA-DR, CD86, MERTK,CD163, CD64, CCR7, AnnexinV) were assessed by intra- and extracellular stainingusing flow cytometry.

Lymphocyte phenotype (CD4, CD8, CD56, CD127, CD25, FoxP3) and LPS-induced production of IFN-c, IL-2, TGF-b, IL-10, IL-4 were assessed followingthe same protocol as per monocytes.

Healthy polymorphonuclear cells were isolated using Polymorphprep solu-tion (density 1.113 g/ml; Axis-Shield, Norway). Contaminating erythrocytes wereremoved with Pharm Lysis (Becton Dickinson (BD), USA). Cells were incubatedwith 25% plasma from ALF patients and 1 ll/ml Golgi Stop (BD) for 2 h. Phenotyp-ing (CD16, CD11b, CD62L) and cytokine production (IL-8) were assessed asdescribed above.

Antibodies for flow cytometry as cited were purchased from BD, UK; eBio-science, UK; R&D Systems, UK; Biolegend, UK. We used a FACS Canto II flowcytometer, BD. Flowlogic (Iniviai, Australia) and Flow Jo v10.0.7 (Tree Star Inc,USA) were used for data analyses.

DAMPS, endothelial activation markers and cytokines were measured.Histone-associated DNA was quantified using cell death detection ELISAplus, Roche, Switzerland. Angiopoietin-2 was measured using ELISA (R&D Sys-tems, UK). Plasma cytokines were analysed using Meso Scale Discovery (MSD;USA).

Statistical analysis

For data that did not follow a normal distribution, the significance of differenceswere tested using Mann–Whitney or Wilcoxon tests. Graphs were drawn usingPrism 6.0c (GraphPad, La Jolla, CA).

Results

Study A

A total of 183 patients met the inclusion criteria in three livertransplantation centres (London (n = 25), Helsinki (n = 15) andCopenhagen (n = 143)) in the period from 1998 to 2010. We can-not account for all those screened as this was not prospectivelydefined or recorded, and the CONSORT criteria was not definedat the time of planning this study (1995 to 1998) (Supplemen-tary. Fig. 1). The patients were randomly assigned to receiveSMT (91 patients) or SMT plus HVP (92 patients). Consent waswithdrawn in one patient from the patient next-of-kin after thefirst HVP procedure due to regret of consent.

Journal of Hepatology 20

182 patients were available for analysis, 90 patients in theSMT group and 92 patients in the HVP group. Baseline character-istics were similar in the two groups (Table 1). There was apredominance of Caucasian ethnicity with more females thanmales. The majority of the patients had concomitant acute kidneyinjury and/or were mechanically ventilated with need ofvasopressor support at the time of randomisation. Antibioticswere given to 88.1% in the SMT group and 94.6% in the HVPgroup. Comparing the two groups at baseline there were no dif-ferences between temperature, mean arterial pressures, needfor vasopressors (Table 2), arterial blood gases, lactate, glucoseor ammonia levels (NH3). The severity of liver failure was similarbetween the two groups with regard to INR, bilirubin andaminotransferases (ALT) (Tables 2 and 3).

HVP dataAll patients received at least one treatment session with HVPapart from one patient who was transplanted before HVP couldbe initiated. All the 182 enrolled patients were included in theITT analysis.

The mean number of HVP treatments per patient was2.4 ± 0.8. The mean plasma volume exchanged during the firstsession was 9.3 ± 1.3 L, 9.2 ± 1.2 L for session 2 and 9.0 ± 1.2 Lsession 3. The mean administered dose was 0.33 ± 0.12 L/kg ofactual body weight per treatment.

Comparison of HVP and SMT groupMean length of hospital stay was 21.9 ± 28.3 days in the HVPgroup, compared to 41.8 ± 27.9 days in the SMT group(p = 0.98). Forty-six (50%) of the patients in the HVP group and44 (49%) in the control group were listed for liver transplantation.24 (26.1%) patients in the HVP group underwent transplantationas compared to 32 (35.6%) patients in the control group (p = 0.17).The mean time to transplantation following listing was4.6 ± 0.6 days in the HVP treated group and 3.7 ± 1.5 in the SMTgroup (p = 0.75).

Survival to hospital discharge was 58.7% for patients treatedwith HVP vs. 47.8% for the control group (HR for high-volumeHVP vs. SMT with stratification for liver transplantation: 0.56;95% CI 0.36 to 0.86; p = 0.0083) at hospital discharge (Fig. 1).

In patients who were transplanted, HVP prior to transplanta-tion did not improve survival compared with patients whoreceived SMT alone (Fig.2).

The survival of those patients who fulfilled poor prognosticcriteria but were not listed for transplantation due to contraindi-cations (such as severe psychiatric disease or medical co-morbidity) was significantly higher in those who received SMTplus HVP (n = 28) as compared to those in receipt of SMT alone(n = 36) (Fig. 3). There was a non-significant trend in those listedfor transplantation but not offered a graft (due to development ofcontraindications or an improved clinical condition such thatde-listing occurred) to have improved outcome in the group ofpatients treated with HVP (n = 22) compared to SMT (n = 12)(Cox: p = 0.13).

The survival in paracetamol-intoxicated patients was similarcompared to non-paracetamol-intoxicated patients both in thecontrol group and in patients treated with HVP (data not shown).

Clinical variablesINR, bilirubin, ALT and arterial ammonia concentration, but notlactate, all decreased significantly during the first 7 days in the

16 vol. 64 j 69–78 71

Table 1. Clinical characteristics of the two groups studied at the time of randomization.

SMT (n = 90) HVP (n = 92) p valueGender (females) 57 (63.3%) 66 (71.7%) 0.23Age (years) 45 (36-56) 46 (33-56) 0.84Weight (kg) 70 (61-81) 67 (60-75) 0.19Height (cm) 170 (165-180) 169 (161-175) 0.24Hyper/Acute/Subacute 69/13/8 65/22/5 0.16Etiology* 58/5/17/6/2/2 50/6/22/11/0/3 0.39HE grade (2/3/4) 6/26/58 7/28/55 0.49Mechanical ventilated 66 (77.7%) 77 (86.5%) 0.13Episodes with systolic BP <80 mmHg 23 (27.1%) 25 (28.1%) 0.88 Need for vasopressors 51 (60.7%) 45 (51.1%) 0.23Oliguria 43 (51.2%) 52 (59.8%) 0.11Oozing 14 (16.9%) 14 (16.1%) 0.89Hyperthermia ( >38.0 °C) 6 (7.2%) 13 (14.8%) 0.12Suspected infection 26 (30.1%) 29 (33.0%) 0.78

SMT, standard medical therapy; HVP, plasma exchange; ICP, intracranial pressure, HE, hepatic encephalopathy. *Given as ALF induced by: paracetamol/acute viral hepatitis/unknown aetiology/toxic hepatitis/acute Budd-Chiari/various.

Research Article

HVP group compared to the SMT group (Table 3). There were sig-nificantly fewer patients in the HVP group who required renalreplacement therapy from day 1–7 compared to the controlgroup (47% vs. 68%), OR 0.42 (CI 0.23–076: p <0.0045); despitesimilar creatinine at enrolment. Plasma creatinine wasunchanged from day 0 to day 7 in the HVP group (192 ± 153 to150 ± 48 mmol/L) but increased from 226 ± 181 to286 ± 230 lmol/L in the SMT group (p <0.0001).

Mean arterial pressure increased significantly in the HVPgroup compared to the control group alongside a significantdecrease in vasopressor requirement (Table 2). Other physiologi-cal parameters were unchanged (Table 2).

ICP was measured in 16 patients in the HVP group and 16 inthe control group during the first 7 days of the study; there wereno differences in measured ICP or sustained increases in ICPbetween the two groups in this study period (Table 2).

SIRS and SOFA scoresSIRS score decreased in the HVP group from day 2 when com-pared to baseline and in comparison to the SMT group (Table 4).The SOFA and CLIF-SOFA also decreased significantly from base-line in the HVP group and in comparison with the SMT group.SIRS, SOFA and CLIF-SOFA scores did not change significantly overtime in the SMT group (Table 4).

Adverse eventsCardiac arrhythmia, pancreatitis, worsening gas exchange, ARDS,or transfusion-related acute lung injury, culture positive infectionor haemorrhage were not statistically significant differentbetween the two groups (Tables 2 and 3).

Study B

The patient characteristics in this proof-of-principle study did notdiffer between HVP and the control group in terms of diseaseseverity, organ failure scores, liver and renal function or SIRS cri-teria. Aetiologies for ALF included mainly paracetamol overdosen = 20, acute hepatitis of unknown aetiology (n = 3), hepatitis B

72 Journal of Hepatology 20

(n = 2), hepatitis A (n = 1), Wilson’s (n = 1) and other (n = 3),and were well balanced between the HVP and no HVP groups.

Monocyte activation and circulating levels of DAMPSHVP significantly reduced the level of circulating histone-associated DNA (Fig. 4A). To evaluate the impact of HVP on mono-cyte function, we characterised the phenotype and function ofmonocytes before and after HVP. Production of the pro-inflammatory mediator TNF-awas significantly attenuated follow-ing the first course of HVP compared to patients not receiving HVP(p = 0.0004), independent of the timing of HVP. IL-6 productionwassignificantly reduced in patients undergoing HVP within 48 h ofadmission, but not in those undergoing late (>48 h) HVP(p = 0.0078 vs. p = 0.9697) (Fig. 4B). Concomitantly, phenotypicmarkers of monocyte activation were modulated following HVPas evidenced by reduced surface expression of CD163(p = 0.0250), CD64 (p = 0.0042), CCR7 (p = 0.0008). HVP did notmodulate the total number of circulatingmonocytes or their degreeof apoptosis (Fig. 4C). In contrast to HVP group, levels of monocyte-derived cytokines (TNF-a, IL-6) and activation marker expressionwere unaltered in the non-HVP patient group (Fig. 4B,C).

Neutrophil inflammatory cytokine production following HVPFollowing HVP, we detected a significant reduction in IL-8 pro-duction from neutrophils (p = 0.0013) (Fig. 4D). Simultaneously,expression of L-selectin (CD62L), an adhesion and homing recep-tor important for neutrophil trafficking and endothelial adhesion,is also significantly reduced (p <0.0001) (Fig. 4E).

Lymphocyte phenotype is unaffected while cytokine production fromCD4-T cells and NK/NK-T cells are reduced following HVPThe numbers of circulatory lymphocytes as well as sub-differentiation of lymphocytes including CD4- and CD8-T cells,NK-T cells, NK-cells and regulatory T cells did not change follow-ing HVP (Supplementary Fig. 2A). However, similar to datadescribed above, production of certain cytokines involved inmodulation of the inflammatory response such as IL-10(p = 0.0003), IL-4 (p = 0.001) and TGFb (p <0.0001) were signifi-cantly reduced (Supplementary Fig. 2B).

16 vol. 64 j 69–78

Table 2. Changes in clinical variables and blood gases in patients treated with plasma exchange (HVP) vs. control group at baseline (day 0) to day 7.

SMT (n = 90) HVP (n = 92)Day Median IQR Median IQR p value

Temperature (°C) 0 36.7 [36.0-37.3] 36.8 [35.9-37.5] 0.741 36.7 [36.1-37.6] 36.9 [35.8-37.6] 0.552 36.9 [35.8-37.5] 36.8 [36.2-37.5] 0.903 37.9 [36.0-37.7] 37.2 [36.4-38.0] 0.247 37.3 [36.4-37.9] 37.7 [36.5-38.0] 0.43

MAP (mmHg) 0 75 [69-85] 75 [70-88] 0.661 75 [69-84] 90** [80-100] <0.00012 74 [68-80] 85** [75-100] <0.00013 75 [65-85] 88# [74-100] <0.0017 80 [68-92] 80 [70-94] 0.51

NA (μg/kg/min) divided by 100 0 5 [0-10] 4 [0-10] 0.511 5 [0-10] 0** [0-3] <0.00012 5# [0-16] 0** [0-2] <0.00013 4 [0-14] 0** [0-0] <0.00017 1 [0-14] 0# [0-3] 0.06

ICP (mmHg) 0 9 [7-12] 15 [7-21] 0.171 14# [12-22] 9 [8-13] 0.042 15# [7-18] 11 [10-20] 0.953 9 [5-25] 12 [8-16] 0.847 12 [4-22] 12 [7-20] 0.95

Heart rate (beats/min) 0 95 [82-106] 98 [86-112] 0.201 97 [82-110] 96 [85-110] 0.982 99 [85-110] 91* [80-100] 0.013 100# [88-113] 97 [88-113] 0.107 99 [90-110] 90# [82-106] 0.30

PaO2 (kPa) 0 12.5 [10.4-14.8] 13.1 [11.1-15.6] 0.171 12.3 [10.8-15.6] 11.9# [10.4-14.4] 0.162 12.3 [10.7-13.8] 11.8# [10.4-14.3] 0.433 11.6 [10.8-13.5] 11.9# [10.7-14.0] 0.767 12.0 [10.6-13.2] 11.7# [10.4-13.4] 0.97

PaCO2 (kPa) 0 4.69 [4.10-5.12] 4.59 [4.03-5.25] 0.661 4.69 [4.08-5.29] 4.95** [4.49-5.76] 0.012 4.78 [4.11-5.48] 5.02** [4.60-5.80] 0.013 4.88 [4.41-5.30] 5.10* [4.39-5.75] 0.067 4.70# [4.30-5.33] 4.87# [4.30-5.58] 0.34

PaO2/FiO2 0 302 [197-376] 306 [229-397] 0.401 284 [192-364] 271# [197-341] 0.572 280 [201-353] 273# [205-357] 0.933 273 [200-355] 258# [188-374] 0.977 260 [187-339] 259 [164-367] 0.89

MAP, mean arterial pressure; NA, noradrenalin infusion rate; PaO2, arterial partial pressure for oxygen; PaCO2, arterial partial pressure for carbon dioxide.#p <0.05 compared to day 0 (baseline) within the same randomized group.*p <0.001 compared to day 0 (baseline) within the same randomized group.**p <0.0001 compared to day 0 (baseline) within the same randomized group.p values in the table compares HVP with SMT.

JOURNAL OF HEPATOLOGY

Evolution of surrogate parameters for endothelial activation,inflammatory response and organ failure following HVPTo date, changes in hepatic and systemic inflammatory responseare not quantifiable using circulatory biomarkers. We analysedevolution of plasma cytokines concentration and endothelial acti-vation markers as well as markers of disease severity, organ dys-function, and SIRS.

Journal of Hepatology 20

Angiopoietin-2, a marker of tissue endothelial dysfunctionand leakage, was significantly reduced following HVP(p = 0.0049) (Supplementary Fig. 2C), corroborating HVP mayinfluence the interaction of tissue immune cell and endothelialactivation. Circulatory cytokines, highly released by a huge vari-ety of immune effector and endothelial cells, were not clearedby HVP (Supplementary Fig. 2D).

16 vol. 64 j 69–78 73

Table 3. Biochemistry changes in patients with plasma exchange (HVP) vs. control group at baseline (day 0) to day 7.

SMT (n = 90) HVP (n = 92)Day Median IQR Median IQR p value

Glucose (mmol/L) 0 8.4 [6.8-11.7] 7.3 [6.0-10.2] 0.091 8.0 [6.5-11.6] 7.6 [5.8-10.6] 0.332 8.2 [6.3-10.4] 7.2 [5.8-10.2] 0.153 8.1 [6.4-10.4] 7.6 [6.6-10.2] 0.957 10.3 [7.7-11.6] 7.2# [6.2-9.4] 0.01

Lactate (mmol/L) 0 3.1 [2.3-5.3] 3.2 [2.1-5.7] 0.761 3.0 [2.0-4.4] 3.0 [2.2-4.7] 0.702 2.8 [1.9-4.6] 2.6# [2.0-3.5] 0.603 2.7 [1.9-4.0] 2.7# [1.7-3.5] 0.477 2.1 [1.8-3.0] 2.3# [1.8-3.3] 0.62

NH3 (μmol/L) 0 133 [90-178] 138 [95-172] 0.831 128 [98-171] 112** [81-134] <0.012 131 [101-174] 97** [73-127] <0.0013 126 [89-166] 106* [70-141] 0.077 120 [87-167] 113 [71-156] 0.81

INR 0 4.8 [3.7-7.1] 5.5 [4.1-7.4] 0.111 4.4# [3.0-6.2] 1.7** [1.3-2.3] <0.00012 4.2# [2.7-5.6] 1.7** [1.4-2.1] <0.00013 3.7# [2.7-5.4] 1.7** [1.5-2.2] <0.00017 3.5# [2.7-4.4] 2.0** [1.7-2.6] <0.0001

Bilirubin (μmol/L) 0 161 [84-359] 162 [100-387] 0.361 176# [102-362] 127** [80-198] <0.012 195# [119-348] 123* [77-221] <0.0013 257# [138-398] 135# [100-212] <0.0017 290** [165-411] 183 [116-270] <0.01

ALT (IU/L) 0 1788 [600-4566] 2502 [784-4876] 0.181 1318** [497-3313] 372** [131-1231] <0.00012 851** [309-2159] 196** [72-479] <0.00013 651** [275-1507] 108** [49-255] <0.00017 489** [211-1094] 114** [55-197] <0.0001

WCC (x109/L) 0 10.0 [6.9-16.0] 9.0 [5.2-13.5] 0.051 9.4 [6.0-14.0] 7.9 [4.7-11.0] 0.032 8.4 [6.4-15.5] 7.7 [4.9-11.5] 0.103 9.8 [7.0-18.0] 9.4 [6.4-12.0] 0.137 12.0 [8.0-16.0] 9.7 [7.0-14.0] 0.14

Platelets (x109/L) 0 94 [60-163] 96 [49-180] 0.941 89# [56-141] 67** [37-123] 0.022 80* [53-119] 55** [42-93] 0.023 76* [42-110] 54** [31-83] 0.037 75 [49-114] 59** [39-79] 0.01

NH3, arterial ammonia concentration; INR, international normalization ratio; ALT, aminotransaminases; WCC, white cell count.#p <0.05 compared to day 0 (baseline) within the same randomized group.*p <0.001 compared to day 0 (baseline) within the same randomized group.**p <0.0001 compared to day 0 (baseline) within the same randomized group.p value in the table compares HVP with SMT.

Research Article

Clinical parameters such as liver enzymes (AST, p = 0.0078;GGT, p = 0.0391), kidney function tests (creatinine, p = 0.0078),lactate, as a marker of tissue injury (p = 0.0156), and MELD score(p = 0.0156) improved following early HVP (Fig. 4F).

74 Journal of Hepatology 20

Discussion

ALF is a rare condition with significant heterogeneity, this, alongwith transplantation renders study design complex and the timeto enrol appropriate patient numbers is prolonged, as seen in this

16 vol. 64 j 69–78

Cum

ulat

ive

prop

ortio

nsu

rviv

ing

(%)

1009080706050403020100

0 7 14 21 28 35 42 49 56 63 70 77 84Time (days)

HPV (n = 92)

SMT (n = 90)

Fig. 1. Main results of the intention-to-treat analysis survival data in thestandard medical treated group (SMT) compared to the high-volume plasmaexchange (HVP) treated group (LogRank: p = 0.0058).

Cum

ulat

ive

prop

ortio

nsu

rviv

ing

(%)

1009080706050403020100

0 7 14 21 28 35 42 49 56 63 70 77 84Time (days)

-HVP/+LTx (n = 32)

+HVP/+LTx (n = 24)+HVP/-LTx (n = 68)

-HVP/-LTx (n = 58)

Fig. 2. Survival in the groups, in the two groups receiving SMT (standardmedical treated group) with and without emergency transplantation (�HVP+LTx vs. +HVP�LTx) and the two group receiving SMT with and withoutemergency transplantation (�HVP�LTx vs. +HVP�LTx) (LogRank: p = 0.0058)and Cox proportional hazard: LTx: p <0.0001; HVP: p = 0.0076).

Cum

ulat

ive

prop

ortio

nsu

rviv

ing

(%)

1009080706050403020100

0 7 14 21 28 35 42 49 56 63 70 77 84Time (days)

HVP (n = 28)

SMT (n = 36)

Fig. 3. Survival data in patients fulfilling transplant criteria that receivedstandard medical treated (SMT) only, compared to the SMT treated group thatalso received high-volume plasma exchange (HVP) (Cox: p = 0.03).

JOURNAL OF HEPATOLOGY

study. One of the criticisms of this study is the time taken to enrolappropriate numbers, and the effects of change in standard ofcare. However, this was a randomised controlled trial andimprovement of care would be equally improved in both groups.Furthermore when we compared outcomes between the first andsecond half of this study there was no difference in survival (Sup-plementary Fig. 3). Medical care was highly similar in the threecentres and we found no differences in survival between centres,neither in the control and treated groups (Supplementary. Fig. 4).

The aim of this open, prospective, randomised, controlledstudy was to examine if HVP reduces mortality in patients withALF. We found that exchange of plasma in ALF patients with freshfrozen plasma increases transplant-free survival after 3 months(Fig. 1). The data shows that the main effect of HVP on survivalis achieved in the patients that did not undergo emergency livertransplantation (Fig. 2). The data also shows that in those patients,who fulfilled criteria for poor prognosis but were not listed fortransplantation due to contraindications, also demonstrated asignificant increase in survival after HVP therapy (Fig. 3). Interest-ingly, patients listed for transplantation, but not offered a liver

Journal of Hepatology 20

graft also tended to have a better outcome though the numberof studied patients was low. Spontaneous survival rate was lowestin the group of patients that were not treated with HVP butfulfilled poor prognostic criteria and not listed for emergency livertransplantation (Fig. 2). The survival rate in this subgroup wascomparable with previous reports in the literature [3].

HVP may have a beneficial effect by delivery of physiologicallyimportant substances contained in fresh frozen plasma. Equallythe effect of HVP could also result from the removal of toxicfactors. It is also possible that the extensive release of intracellu-lar material from the necrotic liver and platelet destruction resultin microparticles and microcirculation failure [26]. The beneficialeffect of HVP, particularly in regard to systemic haemodynamicsmay be related to removal of circulating filaments, proteins orvasoactive substances [17,20].

The positive effect of HVP on survival observed in this study,as supported by the data of study B, would suggest early applica-tion modulates a pro-inflammatory ‘‘storm” and limits the anti-inflammatory response, providing a window of homeostasis forthe liver to regenerate as illustrated by improved survival andimproved SIRS and SOFA scores.

We did not observe a difference in arterial lactate concentra-tion in either group neither at baseline nor during treatment sug-gesting that HVP, at least over the time period of the study didnot reduce the lactate production (microcirculation and muscle)nor significantly affect metabolic clearance in the liver. Thisobservation is in contrast to the decrease in SIRS markers seenin the HVP group, which one might postulate could be associatedwith a decrease in muscle lactate production [27].

We also conducted a proof-of-principle study to assess howALF plasma pre- and post-HVP altered the function of circulatingimmune cell subsets. Our novel data suggests HVP dampensinnate immune responses through removal of circulating DAMPs,such as histone-associated DNA. These non-infectious inflamma-tory molecules promote innate immune activation followingacute hepatocellular death [28–30] through the activation ofpro-inflammatory responses of circulating monocytes and neu-trophils. Importantly we show that both monocyte and neu-trophil derived production of pro-inflammatory mediators wasmarkedly attenuated following HVP. As shown in Fig. 4F, wehypothesise that HVP modulates the tissue destructive functionsand migratory capabilities of circulating innate immune cells,ameliorating the severity of hepatic and MOF. However, furtherwork is required to elucidate the precise mechanisms ofhow HVP modulates the function circulating, tissue residentinnate immune cells, severity of acute tissue injury and organdysfunction.

We also confirmed that arterial pressure increases concomi-tant with a decline in the need for vasopressor support [10,17],raising the possibility of clearance of vasoplegic mediators orimproved receptor activation. Angiopoietin-2, a marker of tissueendothelial dysfunction and leakage, was significantly reducedfollowing HVP, suggesting that HVP may influence the interactionof tissue immune cell and endothelial activation. Interestinglythe arterial ammonia concentration, known to inhibit normalphagocytic function also decreased [31–33].

In association with the observed tendency to a decline inwhite cell count there was a consistent decrease in the SIRSand SOFA scores in study A that may reflect an immune-modulatory effect of HVP potentially limiting progression ofMOF as mentioned above (Table 4). Indeed, such an anti-

16 vol. 64 j 69–78 75

Table 4. The sequential organ failure assessment score (SOFA), the CLIF- SOFA score and the systemic inflammatory response syndrome score (SIRS) in patients withplasma exchange (HVP) vs. the control group at baseline (day 0) to day 7.

SMT (n = 90) HVP (n = 92)Day Median IQR Median IQR p value

SOFA-score 0 14 [11-18] 13 [11-18] 0.591 15* [13-19] 12# [10-16] <0.012 16# [14-19] 13 [10-16] <0.0013 16# [14-19] 13 [11-16] <0.00017 17 [14-19] 13 [11-17] 0.05

CLIF-score 0 17 [14-20] 16 [14-19] 0.671 18# [15-20] 13** [12-17] <0.00012 18# [15-20] 13** [12-16] <0.00013 18 [15-21] 13* [12-16] <0.00017 18 [15-21] 13# [12-18] 0.01

SIRS-score 0 2 [1-3] 2 [2-3] 0.131 2 [1-2] 1** [1-2] 0.242 2 [1-3] 1** [0-2] <0.0013 2 [1-3] 1* [1-2] 0.017 2 [1-3] 1* [1-2] 0.06

#p <0.05 compared to day 0 (baseline) within the same randomized group.*p <0.001 compared to day 0 (baseline) within the same randomized group.**p <0.0001 compared to day 0 (baseline) within the same randomized group.p value in the Table compares HVP with SMT.

Research Article

inflammatory effect of HVP may also explain why the HVP trea-ted group of patients had a statistically significant lower needfor renal replacement therapy during their course of illness[31,34], assuming acute kidney injury to be largely inflammatorydriven in nature.

Cerebral oedema and high ICP evolve in about 15–20% ofpatients with ALF [35,36]. Here we confirm this observation.Although the incidence of this complication has declined overthe last three decades [3] previous studies have suggested a ben-eficial effect of HVP on cerebral haemodynamics [18]. In thisstudy however, the ICP remained similar in the two study groupsthroughout the treatment period. Given the small number ofpatients that developed this complication, the study would notbe powered to observe any potential benefit.

The decision to delineate three treatments over three days inthe protocol was based on the assumption that it would provide atime frame in which hepatic stunning could be reversed and liverregeneration be established. The dose of HVP was a pragmaticdecision based on results from previous studies and recognitionof the strain on the blood banks if larger volumes or more pro-longed treatment periods, had to be undertaken. Furthermore,prolonged duration of HVP might contribute to functionalimmunosuppression and increased sepsis. In this study weexchanged about 23 litres per treated patient, close to the mini-mum dose we aimed to achieve. We cannot refute that moreintensive or continuous HVP [37] would have improvedtransplant-free survival even further or that a lower dose maybe equally effective.

In this study, the timing of the HVP intervention in ALFpatients was based upon our previous experience. We cannotrefute if even earlier intervention with HVP, i.e. before develop-ment of overt encephalopathy may have improved survival fur-ther or would have put a patient at risk of complications of anintervention before the development of one of the pivotal organfailures in defining prognosis. Earlier use of HVP may alsoobscure the development of poor prognostic criteria and hence

76 Journal of Hepatology 20

deprived patients of being listed for liver transplantation. Equally,later use of HVP may worsen outcome if one of the postulatedmechanisms is decreasing inflammatory phenotype and suppres-sion of cytokine response. At this time immune-paresis has devel-oped and HVP may be detrimental. The role of plasma proteinsand their effect on cellular immune function remains in itsinfancy [38].

We feared that sepsis, brain oedema and coagulopathy withassociated drop in temperature would be the main complicationsin the HVP treated group, both potentially by virtue of immunemodulation and putative removal of antibiotics and antifungalsresulting in impaired levels. However, we found no statistical sig-nificant difference between treated and non-treated patientswith regard to these complications. Another concern would bethe small but not impossible risk of transmission of (viral) dis-eases with the use of fresh frozen plasma. We considered thatthis risk may have been low as blood products in our institutionsare collected under strict control of transmittable diseases.Though, we did not suspect or see any cases of viral transmitteddiseases in surviving treated patients, we did not systematicallytest for this potential complication.

This study shows, for the first time in patients with ALF animproved transplant free survival by using a rather primitiveform of extracorporeal liver support utilizing HVP. This pro-vides rationale for further comparative studies between HVPand other liver support systems to explore the immunologicaleffects of these therapies. In addition, further work may berequired to determine the optimal dosing and timing of HVP.However, there are limitations for undertaking further studies.Increasingly, in Europe and the US the time to transplantationis short [2,3]. This limits the time available for any treatmentintervention to show benefit or otherwise as was observed inthe FUMAR study [12]. As reported in the sub-analysis in thislatter study [12] we also show that the effect of HVP was pri-marily found in patients that were not transplant candidates/not transplanted.

16 vol. 64 j 69–78

C

D F

E

BA

pre post BL d3-50

10000

20000

30000

40000

50000

mU

/ml

TPE no TPE

n.s.*

Histone-associated DNA

pre post BL d3-50

2000

4000

6000

8000

MFI

***

n.s.

TPE no TPE

TNF-α TNF-α

pre post pre post0

2000

4000

6000

8000

MFI *

*

TPE <48h TPE >48h

IL-6

pre post BL d3-50

2000

4000

6000

8000

10000

MFI n.s.

TPE no TPE

n.s.

IL-6

pre post pre postTPE <48h TPE >48h

0

2000

4000

6000

8000

10000

MFI

*

n.s.

pre post BL d3-5

MFI

0

500

1000

1500

2000

2500 *

n.s.

TPE no TPETPE post BL d3-5

0

1000

2000

3000

4000

MFI

**

n.s.

TPE no TPEpre post BL d3-5

0

1000

2000

3000

MFI

**

n.s.

TPE no TPEpre post pre post

CD163 CD64 CCR7

0.0

0.1

0.2

0.3

0.4

0.5

x109 /L

n.s.

Monocytes

0

200

400

600

800

1000

MFI

n.s.Annexin V

0

500

1000

1500

MFI

pre post BL d3-5TPE no TPE

n.s.**

0

5

10

15

20

x109 /L

pre post

n.s.

0

10

20

30

40

50

% o

f neu

troph

ils

pre post BL d3-5TPE no TPE

n.s.

**

IL-8

L-selectin (CD62L)

Neutrophils

Inflammatorymediators

Inflammation Tissue damage Hepatic vein

Portal vein

Hepatocytedeath

Hepatocyteregeneration?

DAMPs(DNA, histones, HMGB1)

Macrophageactivation

1

2

2

CD64

TNF IL-6

CD163CCR7

Monocyte activation“Pro-inflammatory”macrophage

“Pro-inflammatory”monocytes

CD64

TNF IL-6

CD163CCR7

DAMPs

SIRS

“Pro-inflammatory”neutrophils

CD62LIL-8

4

CD64

TNF IL-6

CD163CCR7

CD62LIL-8

Liver

Circulation“Attenuated”

monocytes & neutrophils

HVP

Removal of DAMPs

Removal of other mediators?

3

Neutrophilextravasatation

TNF IL-6

IL-8

TNF IL-6

Inflammation Tissue damage

4

4

Extrahepatic tissues(kidneys, lungs,peritoneum, etc.)

“Attenuated” tissuedestruction

Fig. 4. Modulation of innate immune cells following high-volume plasma exchange HVP. In (A)–(E) HVP is termed TPE. (A) Circulating histone-associated DNA levelsare reduced following HVP (n = 12 HVP- vs. n = 7 no HVP patients; mU/ml). (B) LPS-induced pro-inflammatory cytokine production by monocytes is reduced following HVP(HVP n = 20; no HVP n = 11; early HVP [<48 h following admission] n = 8; late HVP [>48 h following admission] n = 12; MFI, mean fluorescence intensity). (C) Expression ofmonocyte activation markers (CD163, CD64, CCR7) is reduced following HVP (HVP n = 16; no HVP n = 8; MFI); Numbers of monocytes (�109/L) and rate of apoptosis(Annexin, MFI) remain unchanged. (D) Neutrophil IL-8 production is reduced following HVP (HVP n = 16, no HVP n = 8; % of neutrophils). (E) Expression of L-selectin isreduced on neutrophils (MFI), number of neutrophils is unchanged (�109/L). Wilcoxon tests; *p <0.05; **p <0.01, ***p <0.001. (F) Proposed model for how HVP dampenstissue innate immune responses: Hepatocyte death in ALF triggers release of DAMPs that activate innate immune cells in the liver and the circulation, and subsequentlyleads to tissue inflammation and SIRS. HVP leads to a reduction of circulatory DAMPs and to an attenuation of the pro-inflammatory profile of innate immune cells,conferring reduced tissue destructive capabilities.

JOURNAL OF HEPATOLOGY

To conclude HVP improves clinical and paraclinical variablesand increases liver transplant-free survival in patients with ALF.Based upon our subsequent proof-of-concept study we suggestthat the effect of HVP is mediated by dampening circulatinginnate immune responses.

Journal of Hepatology 20

Conflict of interest

The authors who have taken part in this study declared that theydo not have anything to disclose regarding funding or conflict ofinterest with respect to this manuscript.

16 vol. 64 j 69–78 77

Research Article

Acknowledgement

The authors thank the national institutions, staff and blooddonors who supported this study without any free.For the proof-of-concept study, the authors would like toacknowledge the funding support from NIHR Imperial Biochemi-cal Research Centre (BRC), Medical Research Council (MRC), Euro-pean Association for the Study of the Liver (EASL) and RosetreesCharitable Trust.(ClinicalTrials.gov number NCT00950508).

Supplementary data

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.jhep.2015.08.018.

References

[1] O’Grady JG, Schalm SW, Williams R. Acute liver failure: redefining thesyndromes. Lancet 1993;342:273–275, [Erratum, Lancet 1993;342:1000].

[2] Lee WM, Stravitz RT, Larson AM. Introduction to the revised AmericanAssociation for the Study of Liver Diseases Position Paper on acute liverfailure 2011. Hepatology 2012;55:965–967.

[3] Bernal W, Wendon J. Acute liver failure. N Engl J Med 2013;369:2525–2534.[4] Chung RT, Stravitz RT, Fontana RJ, Schiodt FV, Mehal WZ, Reddy KR, et al.

Pathogenesis of liver injury in acute liver failure. Gastroenterology2012;143:e1–e7. http://dx.doi.org/10.1053/j.gastro.2012.07.011.

[5] Rolando N, Wade J, Davalos M, Wendon J, Philpott-Howard J, Williams R. Thesystemic inflammatory response syndrome in acute liver failure. Hepatology2000;32:734–739. http://dx.doi.org/10.1053/jhep.2000.17687.

[6] Vaquero J, Polson J, Chung C, Helenowski I, Schiodt FV, Reisch J, et al.Infection and the progression of hepatic encephalopathy in acute liverfailure. Gastroenterology 2003;125:755–764.

[7] Antoniades CG, Berry PA, Wendon JA, Vergani D. The importance of immunedysfunction in determining outcome in acute liver failure. J Hepatol2008;49:845–861. http://dx.doi.org/10.1016/j.jhep.2008.08.009.

[8] Bernuau J, Rueff B, Benhamou JP. Fulminant and subfulminant liver failure:definitions and causes. Semin Liver Dis 1986;6:97–106.

[9] O’Grady JG, Alexander GJ, Hayllar KM, Williams R. Early indicators ofprognosis in fulminant hepatic failure. Gastroenterology 1989;97:439–445.

[10] Tygstrup N, Larsen FS, Hansen BA. The value of high volume plasmapheresisin acute liver failure. In: Lee W, Williams R, editors. Acute liver failure. NewYork: Saunders; 1996. p. 267–277.

[11] O’Grady JG, Gimson AES, O’Brien CJ, Pucknell A, Hughes RD, Williams R.Controlled trials of charcoal hemoperfusion and prognostic factors infulminant hepatic failure. Gastroenterology 1988;94:1192.

[12] Saliba F, Camus C, Durand F, et al. Albumin dialysis with a noncell artificialliver support device in patients with acute liver failure: a randomized,controlled trial. Ann Intern Med 2013;159:522–531.

[13] Demetriou AA, Brown Jr RS, Busuttil RW, et al. Prospective, randomized,multicenter, controlled trial of a bioartificial liver in treating acute liverfailure. Ann Surg 2004;239:660–667.

[14] Winters JL. Plasma exchange: concepts, mechanisms, and an overview of theAmerican Society for Apheresis guidelines. Hematology Am Soc HematolEduc Program 2012;2012:7–12. http://dx.doi.org/10.1182/asheducation-2012.1.7.

[15] Kondrup J, Almdal T, Vilstrup H, Tygstrup N. High volume plasma exchangein fulminant hepatic failure. Int J Artif Organs 1992;15:669–676.

[16] Nakamura T, Ushiyama C, Suzuki S, et al. Effect of plasma exchange on serumtissue inhibitor of metalloproteinase 1 and cytokine concentrations inpatients with fulminant hepatitis. Blood Purif 2000;18:50–54.

[17] Larsen FS, Hansen BA, Ejlersen E, Mogensen T, Tygstrup N, Secher NH.Systemic vascular resistance during high volume plasmapheresis in patientswith fulminant hepatic failure: relationship with oxygen consumption. Eur JGastroenterol Hepatol 1995;7:892.

78 Journal of Hepatology 20

[18] Larsen FS, Hansen BA, Ejlersen E, et al. Cerebral blood flow, oxygenmetabolism and transcranial Doppler sonography during high volumeplasmapheresis in fulminant hepatic failure. Eur J Gastroenterol Hepatol1996;8:261–265.

[19] Clemmesen JO, Kondrup J, Nielsen LB, Larsen FS, Ott P. Effects of high-volume plasmapheresis on ammonia, urea, and amino acids in patients withacute liver failure. Am J Gastroenterol 2001;96:1217–1223.

[20] Clemmesen JO, Gerbes AL, Gülberg V, Hansen BA, Larsen FS, Skak C, et al.Hepatic blood flow and splanchnic oxygen consumption in patients withliver failure. Effect of high-volume plasmapheresis. Hepatology1999;29:347–355.

[21] Ferenci P, Lockwood A, Mullen K, Tarter R, Weissenborn K, Blei AT. Hepaticencephalopathy–definition, nomenclature, diagnosis, and quantification:final report of the working party at the 11th World Congresses ofGastroenterology, Vienna, 1998. Hepatology 2002;35:716–721.

[22] Lee WM, Hynan LS, Rossaro L, et al. Intravenous N-acetylcysteine improvestransplant-free survival in early stage non-acetaminophen acute liverfailure. Gastroenterology 2009;137:856–864.

[23] Vincent JL, Moreno R, Takala J, et al. The SOFA (Sepsis-related Organ FailureAssessment) score to describe organ dysfunction/failure. On behalf of theWorking Group on Sepsis-Related Problems of the European Society ofIntensive Care Medicine. Intensive Care Med 1996;22:707–710.

[24] Moreau R, Jalan R, Gines P, Pavesi M, Angeli P, Cordoba J, et al. Acute-on-chronic liver failure is a distinct syndrome that develops in patients withacute decompensation of cirrhosis. Gastroenterology 2013;144:1426–1437,1437, e1–e9.

[25] American College of Chest Physicians/Society of Critical Care MedicineConsensus Conference: definitions for sepsis and organ failure and guide-lines for the use of innovative therapies in sepsis. Crit Care Med1992;20:864–874.

[26] Stravitz RT, Bowling R, Bradford RL, Key NS, Glover S, Thacker LR, et al. Roleof procoagulant microparticles in mediating complications and outcome ofacute liver injury/acute liver failure. Hepatology 2013;58:304–313.

[27] Levy B, Gibot S, Franck P, Cravoisy A, Bollaert PE. Relation between muscleNa+K+ ATPase activity and raised lactate concentrations in septic shock:aprospective study. Lancet 2005;365:871–875, [Erratum in: Lancet2005;366:122].

[28] Takano K, Shinoda M, Tanabe M, Miyasho T, Yamada S, Ono S, et al.Protective effect of high-mobility group box 1 blockade on acute liver failurein rats. Shock 2010;34:573–579. http://dx.doi.org/10.1097/SHK.0b013e3181df0433.

[29] Craig DGN, Lee P, Pryde EA, Masterton GS, Hayes PC, Simpson KJ. Circulatingapoptotic and necrotic cell death markers in patients with acute liver injury.Liver Int 2011;31:1127–1136. http://dx.doi.org/10.1111/j.1478-3231.2011.02528.x.

[30] Wen Z, Liu Y, Li F, Ren F, Chen D, Li X, et al. Circulating histones exacerbateinflammation in mice with acute liver failure. J Cell Biochem2013;114:2384–2391. http://dx.doi.org/10.1002/jcb.24588.

[31] Selvapatt N, Singanayagam A, Wendon J, Antoniades CG. Understandinginfection susceptibility in patients with acute-on-chronic liver failure.Intensive Care Med 2014;40:1363–1366. http://dx.doi.org/10.1007/s00134-014-3349-x.

[32] Shawcross DL, Wright GA, Stadlbauer V, Hodges SJ, Davies NA, Wheeler-Jones C, et al. Ammonia impairs neutrophil phagocytic function in liverdisease. Hepatology 2008;48:1202–1212.

[33] Taylor NJ, Nishtala A, Manakkat Vijay GK, Abeles RD, Auzinger G, Bernal W,et al. Circulating neutrophil dysfunction in acute liver failure. Hepatology2013;57:1142–1152.

[34] Tritto G, Davies NA, Jalan R. Liver replacement therapy. Semin Respir CritCare Med 2012;33:70–79.

[35] Murphy N, Auzinger G, Bernal W, Wendon J. The effect of hypertonic sodiumchloride on intracranial pressure in patients with acute liver failure.Hepatology 2004;39:464–470.

[36] Bernal W, Hall C, Karvellas CJ, Auzinger G, Sizer E, Wendon J. Arterialammonia and clinical risk factors for encephalopathy and intracranialhypertension in acute liver failure. Hepatology 2007;46:1844–1852.

[37] Nakae H, Eguchi Y, Saotome T, et al. Multicenter study of plasma diafiltrationin patients with acute liver failure. Ther Apher Dial 2010;14:444–450.

[38] Antoniades CG, Khamri W, Abeles RD, Taams LS, Triantafyllou E, PossamaiLA, et al. Secretory leukocyte protease inhibitor: a pivotal mediator of anti-inflammatory responses in paracetamol acetaminophen-induced acute liverfailure. Hepatology 2014;59:1564–1576.

16 vol. 64 j 69–78