CS

GM

TaohcdTa(tmvws

t

©3

T

ytokine Level Modifications: Molecular Adsorbent Recirculatingystem Versus Standard Medical Therapy

. Novelli, M.C. Annesini, V. Morabito, P. Cinti, F. Pugliese, S. Novelli, V. Piemonte, L. Turchetti,. Rossi, and P.B. Berloco

ABSTRACT

Introduction. Acute-on-chronic liver failure (ACLF) is a systemic inflammatory reaction,which is characterized by a predominantly proinflammatory cytokine profile, causing thetransition from stable cirrhosis to ACLF. The aim of the present study was to evaluate thechanges in several cytokines associated with inflammatory liver disease and liver regener-ation among 15 ACLF patients treated with the Molecular Adsorbent RecirculatingSystem (MARS) compared with 15 patients treated with standard medical therapy (SMT).The subjects showed various disease etiologies but similar values for Model End-stageLiver Disease scores.Methods. In the MARS group, 15 (10 male and 5 female) patients were treated withMARS (Gambro). The number of MARS applications was nine; the length of applicationswas 8 hours. In the SMT group; 15 (10 male and 5 female) patients were treated with SMT.The patients were monitored for 30 days from inclusion with a survival follow-up at 3months. Statistical results were calculated with SPSS14.0 (SPSS Inc, Chicago, Ill). A P �.07 was considered significant.Results. In the MARS group, we observed significant changes in the levels of Interleukin(IL)-6, IL-1, IL-10, and tumor necrosis factor (TNF)-� in association with improvedhepatocyte growth factor. Patient survival at 3 months was 60%. The SMT group showedonly a significant change in TNF-� (P � .03). Patient survival at 3 months was 30%.Conclusion. The MARS liver support device corrected pathophysiologies of ALF and

may be used to enhance spontaneous recovery or as a bridge to transplantation.

ncsade

TdDIL

mS

HE LIVER IS COMPOSED of “stable” tissue thatdoes not regenerate in normal circumstances but

cquires this ability after certain forms of damage. What webserve is not a true regenerative process but rather ayperplastic response that involves virtually every hepato-yte of the remaining healthy liver.1,2 Cell reproductionuring hepatic regeneration is stimulated by several factors.umor growth factor-�, hepatocyte growth factor (HGF),nd the interaction of certain cytokines such as interleukinIL) 6 play major role in growth during liver regenera-ion.3,4 Proliferation of hepatocytes also occurs followingassive hepatic necrosis or apoptosis caused by ischemia,

iral infections, or hepatic toxins. It is, however, unclearhy liver regeneration in these cases does not seem to be as

yncronized as it is after hepatic resection.5

Hepatic regeneration starts during the active inflamma-

ory phase but reaches completion only after removal of the R

2009 by Elsevier Inc. All rights reserved.60 Park Avenue South, New York, NY 10010-1710

ransplantation Proceedings, 41, 1243–1248 (2009)

oxious stimulus. The systemic inflammatory responses areomplex and connected to many biological factors. Theystem’s complexity and redundancy are both strong pointsnd weaknesses. In a liver with preexisting pathologies,efective apoptosis can occur in vunerable hepatocytes viaxtrinsic means, receptor-mediated death, or alternatively

From the Dipartimento “P. Stefanini” Chirurgia Generale erapianti d’Organo (G.N., V.M., P.C., M.R., P.B.B.), Dipartimentoi Scienze Anestesiologiche, Medicina Critica e del Dolore (F.P.),ipartimento di Ingegneria Clinica (S.N.), and Dipartimento di

ngegneria Chimica, dei Materiali e dell’Ambiente (M.C.A., V.P.,.T.), La Sapienza Università di Roma, Rome, Italy.Address reprint requests to Prof Gilnardo Novelli, Diparti-ento “P. Stefanini” Chirurgia Generale e Trapianti d’Organo, Laapienza Università di Roma, Viale del Policlinico 155, 00161,

ome, Italy. E-mail: novelligilnardo@virgilio.it

0041-1345/09/$–see front matterdoi:10.1016/j.transproceed.2009.03.035

1243

tciktavptAusAfbmlMrpwtd

M

MM

F(e3ussncavnep(bfP

M

Mat

occbmopcSma

M

Dbhsmcrtotattpvtaaa

EMSS

S

IS

MIIITH

tEhr

1244 NOVELLI, ANNESINI, MORABITO ET AL

hrough intercellular stress causing alterations and insuffi-ient regenerative capacity. This is probably due to anmbalance between the prevalent proinflammatory cyto-ines and anti-inflammatory cytokines. This imbalance de-ermines the passage from a stable chronic hepatic illness ton acute one. The interest in artificial extracorporeal de-ices is therefore justified as they can help to detoxify andurify these substances in a relatively short period. Today,he most widely used artificial device is the Moleculardsorbent Recirculating System (MARS), which has beensed in almost 5000 treatments. Large controlled clinicaltudies have shown it to be the most reliable treatment.lthough several investigators have described clinical ef-

ects of MARS on cytokine removal,6–10 some results haveeen conflicting. Ilonen et al11 showed that cytokine re-oval was not significant in a study performed on 49 acute

iver failure (ALF) patients after 22 hours of continuousARS treatment. Sen et al in 200412 reported another

andomized study on nine acute-on-chronic liver failureatients who underwent four 8-hour MARS treatmentsith no significant removal of cytokines. The explanation of

hese results is probably based on excessive cytokine pro-uction compared to detoxification.The aim of this study was to evaluate the efficiency ofARS to increase cytokine removal by:

(1) increasing the length of single treatments;(2) increasing number of treatments;(3) optimizing flow speed; or(4) optimizing albumin action in patient/circuit ratio

through application of a mathematical model. Oursecondary endpoint was to evaluate patient out-comes at 3-month follow up.

ATERIALS AND METHODSARS Group

ifteen (10 male and 5 female) patients were treated with MARSGambro, Stockholm, Sweden) in association with dialysis machinequipment. The mean age of these patients was 48.5 years (range �5–61). Five patients were excluded from the study because onenderwent liver transplantation (four treatments); two died fromepsis (three and two treatments respectively); and two patientsuccumbed to multiorgan failure (three treatments). The meanumber of MARS applications was nine and the length of appli-ations was 8 hours. Six consecutive-day treatments and threelternate day treatments were performed over 12 days total. Serumalues related to inflammation (IL-1, IL-6, IL-10, and tumorecrosis factor alpha [TNF-�]), to regenerative function (HGF), toxcretory and detoxification function (bilirubin), to synthetic ca-acity (international normalized ratio value), and to renal functioncreatinine) were measured using standard laboratory proceduresefore and after MARS treatments. The patients were evaluatedor 30 days from inclusion with a survival follow-up at 3 months.atient characteristics at baseline are reported in Table 1.

ARS Application

ARS utilizes albumin as a molecular adsorbent to removelbumin-bound liver toxins from patient blood. All patients were

reated via a central venous catheter. For anticoagulation, a bolus t

f 3000 U of unfractionated heparin was injected into the extra-orporeal system at the beginning of the treatment, followed byontinuous infusion dosed to achieve an activated partial throm-oplastin time of 50 seconds. Blood flow rate was 190 to 220L/min, depending on the hemodynamic status of the patient and

n mathematical model. Blood passed through an albumin non-ermeable, high-flux dialysis membrane. A closed-loop dialysateircuit containing �600 mL 20% stored human albumin (GrifolsA, Barcelona, Spain) was driven by a roller pump of the MARSonitor at 190 to 220 mL/min. Infusions of frozen plasma were

llowed during the treatments.

ARS Session Model Scheme

uring the treatments, the concentrations of albumin and albumin-ound toxins change both in the circuit and in the patient fromour to hour. We observed ideal concentrations for every patientupported by the study using the clinical data and the mathematicalodel (Fig 1). Detoxification directly correlated with albumin

oncentrations: the greater the concentration, the more toxinsemoved. The ratio of albumin concentrations in the patient and inhe circuit, however, must remain constant. The duration of theseptimal albumin levels is necessary for the removal of all specificoxins; this treatment was proportional to the concentration oflbumin in the blood and the circuit. Changes in albumin concen-rations and lengths of treatment did not alter safety. From time toime, we increased the albumin in blood or in the circuit. In sevenatients, we decided to administer one or more 50-mL albuminials (20%) before the start of each treatment due to the results ofhe mathematical model obtained from clinical status (serumlbumin � 2.4 � 1.4 mg/dL). Together with the 600-mL standard,s indicated by producers, we also included one or more 50-mLlbumin vials (20%) in the circuit, to optimize the extracorporeal

Table 1. Brochemical Parameters and Patient Characteristicsat Baseline

MARS Group (n � 10) SMT Group (n � 10)

tiology 4 pt HBV, 6 pt alcohol 4 pt HBV, 6 pt alcoholean age 49.3 (range 37–59) 50.7 (range 39–61)ex 6 female/4 male 5 female/5 maleerum bilirubin(mg/dL)

20.3 (range 17–24) 20.3 (range 17–25)

erum creatinine(mg/dL)

1.2 (range 0.7–2.7) 1.3 (range 1.7–2.1)

NR 1.7 (range 1.1–2.7) 1.7 (range 1.6–2.2)erum albumin(mg/dL)

2.1 (range 1.9–2.6) 2.3 (range 1.9–2.8)

ELD score 25.6 (range 20–30) 27 (range 20–30)L-1 (pg/mL) 46.6 (range 30–78) 40 (range 30–69)L-6 (pg/mL) 81 (range 64–103) 67.6 (range 49–95)L-10 (pg/mL) 18.6 (range 11–25) 22.4 (range 12–30)NF-� 26.5 (range 18–39) 30.4 (range 21–38)GF 6.3 (range 3.9–8.8) 5.1 (range 2.8–8.6)

MARS, molecular adsorbent recirculating system; SMT, standard medicalherapy; Pt, patient; INR, international normalized ratio; MELD, Model fornd-stage Liver Disease; TNF, tumor necrosis factor; IL, Interleukin; HGF,epatic growth factor; HBV, cirrhosis virus B related; Alcohol, cirrhosis alcoholelated.

reatment.

M

TMpfHmtsal

S

Fmatv3dewrfbaiP

C

BtawBm(

S

Scr

bt

RB

AwMfi.gwopt

R

Atic

P

Mpneis(

C

PiacpMi

CYTOKINE LEVEL MODIFICATIONS 1245

ARS Safety

he nonbiological hepatic support MARS is safe and tolerable.ARS treatment can be considered biologically compatible, com-

rehensive, and highly secure. We observed some local complications:our patients presented transitory hypotension (78 � 2.0/50 � 3.5 mm

g) that was corrected through parenteral administration ofannitol (20%), physiological solution, and cortisone. Five pa-

ients with transitory hypoglycemia were treated with glucoseolutions (33%). The thrombocytremia was controlled throughdministration of platelets before the start of treatment when theevels were less than 30,000 mm3 (two patients).

tandard Medical Therapy Group

ifteen (10 male and 5 female) patients were treated with standardedical therapy (SMT), prophylaxis against bacterial infections,

nd judicious use of diuretics. The precipitating factors were alsoreated appropriately. We also administered one 50-mL albuminial (20%) daily. The mean age of the patients was 51 years (range7–64). Five patients were excluded from the study, because twoied of sepsis, one succumbed to multiorgan failure. We alsoxcluded the patient with the best results and the patient with theorst results to achieve a more homogeneous group. Serum values

elated to inflammation (IL-1, IL-6, IL-10, TNF-�), to regenerativeunction (HGF), to excretory and detoxification functions (biliru-in), to synthesis capacity (international normalized ratio value),nd to renal function (creatinine) were analyzed from early morn-ng values collected throughout the observation period of 30 days.atient characteristics at baseline are reported in Table 1.

ytokine Assays

lood samples drawn at baseline, at the end of fifth and ninthreatment, and at 30 days were centrifuged after 30 minutes; serumliquots were stored frozen at �70°C for later cytokine assays IL-6as analyzed by a chemoluminescence assay (Immulite 2000, DPCiermann GmbH, Bad Nauheim, Germany). Enzyme-linked im-unosorbent assay was used to determine IL-10 and TNF-�

Quantikine, R&D Systems, Minneapolis, Minn).

tatistical Analysis

tatistical results were calculated with SPSS14.0 (SPSS Inc, Chi-ago, Ill). A P value � .07 was considered significant. The cytokine

esults were compared using the Wilcoxon matched-pairs test c

etween ninth extracorporeal treatment of MARS group andhirtieth day of the SMT group.

ESULTSiochemistry

significant improvement in serum bilirubin (P � .0009)ith a mean score of 11.38 mg/dL was seen among theARS group over nine treatments. The data were con-

rmed at 30 days with a mean score 5.59 of mg/dL (P �0008). There were no significant changes in the SMTroup. Changes in serum albumin and prothrombin timeere not considered significant in the MARS group becausef the administration of albumin and of fresh frozenlasma. In the SMT group, we observed improved pro-hrombin time (P � .02).

enal Function

significant change in renal function was observed amonghe MARS group over nine treatments: mean serum creat-nine value was 1.2 mg/dL (P � .01) versus a nonsignificanthange in the SMT group at 30 days.

rognostic Markers

odel for End-stage Liver Disease (MELD) scores im-roved significantly among the MARS group after theineth treatment (P � .0002) with a mean score of 19,specially at 30 days (P � .0005), associated with anmproved clinical status with a mean score of 14. Noignificant improvement was observed in the SMT groupP � .8) at the end of the observation period.

ytokines

lasma levels of proinflammatory cytokines were elevatedn most patients in both groups; there were no elevations ofnti-inflammatory cytokines. Figure 2 shows significanthanges in the MARS group. Using the Wilcoxon matched-airs test, we compared the results between the ninethARS treatment and the SMT group at 30 days as reported

n Fig 3. In the SMT group, we observed only a significant

Fig 1. MARS session model.Scheme of MARS session model.AD, albumin dialysis membranemodule; AR, anionic resin col-umn; AC, activated carbon col-umn; TD, conventional dialysismembrane module; MARS, Mo-lecular Adsorbent RecirculatingSystem.

hange in TNF-� (P � .03).

S

Aa1ta2

D

Rbowt

a

ti

wcpb

FmMt

1246 NOVELLI, ANNESINI, MORABITO ET AL

urvival

t 3 months of follow-up, six MARS group patients werelive. Patient 4 died at 1 month and three patients (5, 6, and0) died at 3 months of follow-up. At 3 months of follow-up;hree SMT group patients were alive. Three patients (5, 6,nd 4) died at 1 month; one, (8) at 2 months, and two (1 and) at 3 months of follow-up.

ISCUSSION

eports have shown discordant results of cytokine removaly MARS, which may be based on greater cytokine turn-ver than effective detoxification. This is why our objectiveas to try to improve detoxification and optimize patient

reatment by applying a mathematical model.A mathematical model has been previously proposed for

ig 2. Change of cytokines (pg/L) over 30-day study period inARS group valued with studenttest.

lbumin-bound toxin removal in MARS13–15 (Fig 1). This m

wo-compartment model (intravascular and extravascular)s represented by the following equations:

Toxin mass balance in the extravascular compartment:

VedcB,i

dt� �K(cB,i � cB,e)

Toxin mass balance in the intravascular compartment:

VidcB,i

tot

dt� K(cB,i � cB,e) � QB(cB,tot

in � cB,1tot ) � IB

here cB,e and cB,i stand for free (not albumin-bound) toxinoncentrations in the extravascular and intravascular com-artment, respectively; cB,i

tot is the total (free and albumin-ound) toxin concentration in the intravascular compart-

ent; cB,tot

in stands for the total toxin concentration in blood

dtgme

fpapcnwtdcwtoaeFaiF

aotim

dsc

tciet(hWknAwtbmmsrrtatiase

CYTOKINE LEVEL MODIFICATIONS 1247

ownstream of MARS membrane module; K is a massransfer coefficient; QB is the blood flow rate; IB is the toxineneration rate per unit volume of intravascular compart-ent; Vi and Ve are the volumes of the intravascular and

xtravascular compartments, respectively.Computer simulations of treatment sessions were per-

ormed by means of the aforementioned model, assumingarameter values referred to cytokines: K � 0.361 min�1

s reported by Berk et al16 Vi � 3 L and Ve � 5/3 Vi asreviously assumed by Magosso et al.17 Endogenousytokine production was neglected, setting IB � 0. Fi-ally, an average blood flow rate of QB � 190 mL/minas assumed on the basis of the actual operating condi-

ions, namely, (an equal flow rate to that in the albuminialysate circuit. The results showed that albumin con-entrations in both patient plasma and MARS circuitere important parameters that affected the efficiency of

he treatment. Even if the effects of these parameters onverall treatment efficiency is not straightforward, it can beffirmed that increased plasma concentration of albumin cannhance extraction of albumin-bound toxins from tissues.urthermore, the simulations showed that an increase inlbumin concentration in the MARS circuit solution canmprove albumin-bound toxin transfer across the MARS-LUX membrane.On the basis of these results in the cases presented in this

rticle, we decided to administer one or more 50-mL vialsf albumin (20%) before beginning each treatment. Fur-hermore, together with the 600-mL standard solution, asndicated by the manufacturer, we also included one or

ore 50-mL albumin vials (20%) in the MARS circuit. The (

uration of each treatment was 8 hours. Mathematicalimulations showed that this treatment was effective forytokine detoxification.

At the start of our study, patients showed higher concen-rations of proinflammatory compared to anti-inflammatoryytokines. Balanced levels of interleukin (pro- and anti-nflammatory) are a favorable “quad vitam” prognosticlement. Our results showed that removal of proinflamma-ory cytokines was already significant by the fifth treatment40 hours), rising steadily to the nineth treatment (72ours) with stable values up to 30 days of observation.ilcoxon match-pairs test showed detoxification of cyto-

ines and increase in HGF were more significant at theineth MARS treatment than at 30 days for the SMT group.mong these tests, only IL-6 was not significant. Consistentith these results, IL-10 remained stable in both groups

hroughout the observation period. Can this improvemente determined only by the application of the mathematicalodel? Time (t) was also influential. Cytokine removal wasore efficient when correlated with the length/number of

essions; in some cases it resulted in an increased filtrationate in correlation with optimization of pump volume flowate. The binding strength of cytokine/albumin was cap-ured by the carbon filter 104 L/mol. The affinity betweenlbumin and cytokine was weak, thus allowing us to increasehe volume flow rate and decrease the time. The othermportant aspect is patient albumin concentration. Albumindministration to patients before beginning treatment re-ulted in greater toxin capture. Interleukin accumulation,specially IL-6, resulted in lowering of albumin in the blood

Fig 3. A Wilcoxon matched-pairs test showing the compari-son of cytokine (pg/mL) levelsbetween the nineth session ofMARS group and 30 days of SMTgroup. The decrease of cytokineproinflammatory levels in the MARSgroup (72 hours of treatment in 12days) was matched by a corre-sponding improvement of cyto-kine anti-inflammatory and HGF.In this comparison, we observeda nonsignificant result for IL-6.Mars, Molecular Adsorbent Re-circulating System; IL, interleukin;TNF, tumor necrosis factor; HGF,hepatocyte growth factor; SE, st-andard error of the mean; SD,standard deviation.

from 42 g/L to 22 g/L) and deficient liver functions. If we

iawwicwratwsLml

dhtedfwoc

mmsb3sstabtpcomo

bdHusa

as

R

R

ls

e

ic

a

cc

rbO

mdA

mpM

r3

a2

er

ta3

oc

tap

k

bs

1248 NOVELLI, ANNESINI, MORABITO ET AL

ncreased albumin concentrations in the circuit proportion-lly, we obtained greater detoxification; however, saturationas reached in a shorter time. To avoid this phenomenone regulated pump flow rate to 190 to 220 mL/h and

ncreased number of treatments to achieve better detoxifi-ation in 8 hours (�l). Absorption was faster and the columnas able to detoxify the solution until saturation was almost

eached (break point) and the solute could no longer bebsorbed. Effluent concentration was reduced for longerhan �l. In fact, after this time lapse, albumin in the circuitas less pure, although still able to clear some but not all

ubstances. Binding strength of bilirubin and albumin is 108

/mol, which is high; therefore, lower volume flow rates andore time are necessary for clearance. This finding high-

ights how clearance depends on the time/flow ratio.Both the MARS and the SMT group patients who died

uring follow-up at the end of observation period showedigher cytokine concentrations. Among SMT patients, cy-okine removal occurred slowly, never exceeding massivexisting turnover. Daily administration of 50 mL albuminid not substantially change the situation. At 3 months ofollow-up, the survival rate of patients treated with MARSas 60% compared with 30% for SMT patients. At the endf the observation period, MARS patients showed signifi-ant improvement in MELD scores.

As widely stated in the literature, MELD represents theost widely accepted model for short- and medium-termortality prediction in ALF, but it is clinically limited. A

tudy performed by Yu et al18 showed that a MELD scoreetween 20 and 30 was associated with a 56% to 60%-month mortality. Patients with better outcomes showedignificant reductions in cytokine levels, decreased MELDcores, and improved clinical status. However, we believehat there was no correlation between cytokine reductionnd MELD improvement. As shown in the literature,ilirubin is removed by the resin column, creatinemia viahe dialysate filter, and international normalized ratio im-roves not only due to hepatic synthesis but also because ofontinuous administration of blood components through-ut session. We believe that MELD score may show an evenore significant association if combined with determination

f cytokine concentrations.In conclusion, treatments can only be perfected when,

ased on parameters and single patient variables, theynamics of the removal of each cytokine are known.owever, personalized mathematical models have allowed

s to obtain better detoxification from the device. Furthertudies with greater patient numbers are necessary to define

pplication standards based on biological knowledge about

tM

lbumin, various cytokine functions, and development moreelective devices.

EFERENCES

1. Taub R: Liver regeneration: from myth to mechanism. Natev Mol Cell Biol 5:836, 20042. Higgins GM, Anderson RM: Experimental pathology of the

iver. Restoration of the liver of the white rat following partialurgical removal. Arch Pathol 12:186, 1931

3. Michalopoulos GK, DeFrances MC: Liver regeneration. Sci-nce 276:60, 1997

4. Sigal SH, Rajvanshi P, Gorla GR, et al: Partial hepatectomy-nduced polyploidy attenuates hepatocyte replication and activatesell aging events. Am J Physiol 276:G1260, 1999

5. Koniaris LG, McKillop IH, Schwartz SI, et al: Liver regener-tion. J Am Coll Surg 197:634, 2003

6. Heemann U, Treichel U, Loock J, et al: Albumin dialysis inirrhosis with superimposed acute liver injury: a prospective,ontrolled study. Hepatology 36:949, 2002

7. Stange J, Mitzner SR, Risler T, et al: Molecular adsorbentecycling system (MARS): clinical results of a new membrane-ased blood purification system for bioartificial liver support. Artifrgans 23:319, 19998. Mitzner SR, Klammt S, Peszynski P, et al: Improvement ofultiple organ functions in hepatorenal syndrome during albumin

ialysis with the molecular adsorbent recirculating system. Therpher 5:417, 20019. Sorkine P, Ben Abraham R, Szold O, et al: Role of theolecular adsorbent recycling system (MARS) in the treatment of

atients with acute exacerbation of chronic liver failure. Crit Careed 29:1332, 200110. Lamesch P, Jost U, Schreiter D, et al: Molecular adsorbant

ecirculating system in patients with liver failure. Transplant Proc3:3480, 200111. Ilonen I, Koivusalo A-M, Repo H, et al: Cytokine profiles in

cute liver failure treated with albumin dialysis. Artif Organs 32:52,00812. Sen S, Davies NA, Mookerjee RP, et al: Pathophysiological

ffects of albumin dialysis in acute-on-chronic liver failure: aandomized controlled study. Liver Transpl 10:1109, 2004

13. Annesini MC, Piemonte V, Turchetti L: Albumin-boundoxin removal in liver support devices: case study of tryptophandsorption and dialysis. Chemical Engineering Transactions 14:65, 200814. Annesini MC, Piemonte V, Turchetti L: Artificial liver

ptimization: analysis of albumin bound toxin adsorption. Chemi-al Engineering Transactions 11:551, 2007

15. Annesini MC, Piemonte V, Turchetti L: Albumin boundoxins removal in liver support devices: case study of bilirubindsorption and dialysis. New York: Nova Scince Publishers, Inc; (inress)16. Berk PD, Howe RB, Bloomer JR, et al: Studies of bilirubin

inetics in normal adults. J Clin Invest 48:2176, 196917. Magosso E, Ursino M, Coli L, et al: A modeling study of

ilirubin kinetics during molecular adsorbent recirculating systemession. Artif Organs 30:285, 2006

18. Jian-Wu Yu, Gui-Qiang Wang, Shu-Chen Li: Prediction of

he prognosis in patients with acute-on-chronic hepatitis using the

ELD scoring system. J Gastroenterolo Hepatol 21:1519, 2006