Gastroenterology 2019;157:149–162 CLINICAL—LIVER Effects of Albumin Treatment on Systemic and Portal Hemodynamics and Systemic Inflammation in Patients With Decompensated Cirrhosis Javier Fernández,1,2,* Joan Clària,1,2,* Alex Amorós,1 Ferrán Aguilar,1 Miriam Castro,2 Mireia Casulleras,2 Juan Acevedo,3 Marta Duran-Güell,2 Laura Nuñez,4 Montserrat Costa,4 Mireia Torres,4 Raquel Horrillo,4 Luis Ruiz-del-Árbol,5 Cándido Villanueva,6 Verónica Prado,2 Mireya Arteaga,2 Jonel Trebicka,1,7 Paolo Angeli,1,8 Manuela Merli,9 Carlo Alessandria,10 Niels Kristian Aagaard,11 German Soriano,12 François Durand,13 Alexander Gerbes,14 Thierry Gustot,15 Tania M. Welzel,16 Francesco Salerno,17 Rafael Bañares,18 Victor Vargas,19 CLINICAL LIVER Agustin Albillos,5 Aníbal Silva,2 Manuel Morales-Ruiz,2 Juan Carlos García-Pagán,2 Marco Pavesi,1 Rajiv Jalan,20 Mauro Bernardi,21 Richard Moreau,1,13,22 Antonio Páez,4 and Vicente Arroyo1 1 EF Clif, EASL-CLIF Consortium and Grifols Chair, Barcelona, Spain; 2Hospital Clínic, IDIBAPS and CIBERehd, Barcelona, Spain; 3South West Liver Unit, Derriford Hospital, Plymouth, United Kingdom; 4Bioscience Research Group, Grifols, Barcelona, Spain; 5Department of Gastroenterology, Hospital Ramón y Cajal and CIBERehd, Madrid, Spain; 6Department of Gastroenterology, Hospital de Sant Pau and CIBERehd, Barcelona, Spain; 7Department of Internal Medicine, University Hospital of Bonn, Bonn, Germany; 8Unit of Internal Medicine and Hepatology, Department of Medicine, DIMED, University of Padova, Padova, Italy; 9Department of Clinical Medicine, Sapienza University of Rome, Rome, Italy; 10Division of Gastroenterology and Hepatology, San Giovanni Battista Hospital, Torino, Italy; 11Department of Hepatology and Gastroenterology, Aarhus University Hospital, Aarhus, Denmark; 12Department of Gastroenterology and Hepatology, Hospital of Santa Creu i Sant Pau and CIBERehd, Barcelona, Spain; 13Service d’Hépatologie, Hôpital Beaujon, Assistance Publique-Hôpitaux de Paris, Clichy, France; 14Department of Medicine II, Liver Centre Munich, University Hospital, LMU Munich, Munich, Germany; 15Liver Transplant Unit, Erasme Hospital (ULB), Brussels, Belgium; 16Medical Department I, Goethe University, Frankfurt, Germany; 17Department of Internal Medicine, Policlinico IRCCS San Donato, Milano, Italy; 18 Department of Gastroenterology, Hospital Gregorio Marañon, and CIBERehd, Madrid, Spain; 19Department of Internal Medicine, Hospital Vall d’Hebron and CIBERehd, Barcelona, Spain; 20Liver Failure Group, Institute for Liver Disease Health, University College London, Royal Free Hospital, London, United Kingdom; 21Department of Medical and Surgical Sciences, University of Bologna, Bologna, Italy; 22Inserm, Université Paris Diderot-Paris 7, Centre de Recherche sur l’Inflammation, Paris, France This article has an accompanying continuing medical education activity, also eligible for MOC credit, on page e17 (https://www. gastrojournal.org/cme/home). Learning Objective: Upon completion of this CME activity, successful learners will be able to (1) discuss the clinical relevance of normalizing serum albumin concentration in patients with decompensated cirrhosis and ascites and (2) identify the main mechanisms of action of albumin in this setting. Effects of long-term albumin treatment on serum albumin levels and inflammatory cytokines High albumin dose (HAlbD: 1.5 g/kg every week, blue figures) but not low albumin dose (LAlbD: 1 g/kg every 2 weeks: red figures) normalized serum albumin levels and decreased inflammatory cytokines 150 Fernández et al Gastroenterology Vol. 157, No. 1 BACKGROUND & AIMS: We investigated the effect of albumin WHAT YOU NEED TO KNOW treatment (20% solution) on hypoalbuminemia, car- diocirculatory dysfunction, portal hypertension, and systemic BACKGROUND & CONTEXT inflammation in patients with decompensated cirrhosis with Systemic inflammation is an important issue in acute and without bacterial infections. METHODS: We performed a decompensation of cirrhosis, this is thought to play a prospective study to assess the effects of long-term (12 weeks) major role in the pathophysiology of complications, treatment with low doses (1 g/kg body weight every 2 weeks) including multiple organ failure. and high doses (1.5 g/kg every week) of albumin on serum NEW FINDINGS albumin, plasma renin, cardiocirculatory function, portal pres- Long-term high albumin treatment (1.5 g/kg every week sure, and plasma levels of cytokines, collecting data from 18 for 12 weeks) showed significant immunomodulatory patients without bacterial infections (the Pilot-PRECIOSA effects in an exploratory cohort of patients with study). We also assessed the effect of short-term (1 week) decompensated cirrhosis. This finding was validated in a treatment with antibiotics alone vs the combination of albumin short-term (7 days) investigation in patients with plus antibiotics (1.5 g/kg on day 1 and 1 g/kg on day 3) on infections. plasma levels of cytokines in biobanked samples from 78 pa- LIMITATIONS tients with bacterial infections included in a randomized CLINICAL LIVER controlled trial (INFECIR-2 study). RESULTS: Circulatory Low number of patients included in the investigation dysfunction and systemic inflammation were extremely unsta- cohort. ble in many patients included in the Pilot-PRECIOSA study; IMPACT these patients had intense and reversible peaks in plasma Albumin treatment improves markers of systemic levels of renin and interleukin 6. Long-term high-dose albumin, inflammation in patients with decompensated cirrhosis. but not low-dose albumin, was associated with normalization of This effect may underlie the beneficial effects of albumin serum level of albumin, improved stability of the circulation therapy on clinical outcomes in this setting. and left ventricular function, and reduced plasma levels of cy- tokines (interleukin 6, granulocyte colony-stimulating factor, ANSWER study recently showed that long-term (18 months) interleukin 1 receptor antagonist, and vascular endothelial growth factor) without significant changes in portal pressure. prophylactic administration of albumin (40 g every week) to The immune-modulatory effects of albumin observed in the patients with prior history of ascites is highly effective in Pilot-PRECIOSA study were confirmed in the INFECIR-2 study. preventing follow-up development of new episodes of as- In this study, patients given albumin had significant reductions cites, refractory ascites, HRS, hepatic encephalopathy, and in plasma levels of cytokines. CONCLUSIONS: In an analysis of bacterial infections, reducing hospital admissions and data from 2 trials (Pilot-PRECIOSA study and INFECIR-2 study), improving survival.4 This successful research activity on the we found that albumin treatment reduced systemic inflamma- therapeutic use of albumin in cirrhosis contrasts sharply tion and cardiocirculatory dysfunction in patients with with the low number of investigations performed on its decompensated cirrhosis. These effects might be responsible mechanisms of action.1 for the beneficial effects of albumin therapy on outcomes of This article reports the results of the Pilot-PRECIOSA patients with decompensated cirrhosis. ClinicalTrials.gov, study, which aimed to identify an albumin dosage that Numbers: NCT00968695 and NCT03451292. normalizes serum albumin concentration and investigate the effects of the administration of this albumin dosage for 12 weeks on hypoalbuminemia, cardiocirculatory hemody- Keywords: Liver-Related Complications; Immune Response; namics, effective blood volume, portal pressure, and sys- Splanchnic Hemodynamics; Interventional Trials. temic inflammation (as estimated by the plasma levels of T he first studies supporting the use of albumin treatment in cirrhosis were performed in the 1980s and consisted of several randomized controlled trials * Authors share co-first authorship. Abbreviations used in this paper: ACLF, acute-on-chronic liver failure; (RCTs) showing that paracentesis was a rapid, effective, and ANP, atrial natriuretic peptide; BNP, brain natriuretic peptide; EASL-CLIF Consortium, European Association for the Study of the Liver–Chronic safe therapy of ascites if performed with intravenous albu- Liver Failure Consortium; ELISA, enzyme-linked immunosorbent assay; min administration (8 g/L of ascitic fluid removed).1 Sort HAlbD, high albumin dose; HRS, hepatorenal syndrome; IL, interleukin; INFECIR, albumin administration in the prevention of hepatorenal syn- et al2 subsequently showed that treatment of spontaneous drome and death in patients with CIRrhosis, bacterial INFEctions other bacterial peritonitis (SBP) with antibiotics plus albumin (1.5 than SBP and high risk of hospital mortality; LAlbD, low albumin dose; LV, g/kg body weight at infection diagnosis and 1 g/kg on day left ventricle; PRA, plasma renin activity; PRECIOSA, PREvention of mor- tality with long-term administration of human albumin in subjects with 3) was associated with a 60% reduction in the prevalence of decompensated CIrrhOSis and Ascites; PRC, plasma renin concentration; type 1 hepatorenal syndrome (HRS), a special form of acute- RCT, randomized controlled trial; SBP, spontaneous bacterial peritonitis. on-chronic liver failure (ACLF), and in hospital mortality. Most current article Ortega et al3 later showed that the simultaneous adminis- © 2019 by the AGA Institute. Published by Elsevier Inc. This is an open tration of terlipressin and albumin (20–40 g/d for 7–14 access article under the CC BY-NC-ND license (http://creativecommons. org/licenses/by-nc-nd/4.0/). days) normalized serum creatinine concentration in 0016-5085 approximately 50% of patients with HRS) Finally, the https://doi.org/10.1053/j.gastro.2019.03.021 July 2019 Pleiotropic Effects of Albumin Therapy 151 interleukin [IL] 6) in 18 patients with decompensated mg of furosemide). PRA was used to assess sequential changes cirrhosis. in effective arterial blood volume. The exclusion criteria are Recent investigations have suggested that systemic detailed in the Supplementary Materials. inflammation plays a major role in the pathogenesis of acute A total of 135 patients were evaluated; 72 were eligible, and decompensation and ACLF in cirrhosis.5 The observation of among these, 39 were excluded based on the exclusion criteria. a marked suppression of the plasma levels of IL-6 during Of the 33 remaining patients, 12 were excluded for data anal- albumin treatment in the Pilot-PRECIOSA STUDY, which ysis due to (1) lack of abnormal PRA value (<2 ng/mL$h) at suggests an immunomodulatory effect of albumin treat- enrollment (2 patients), (2) development of complications ment, prompted us to perform additional investigations to requiring treatment that interfered with the interpretations of the results (intensive care, liver transplantation, and insertion confirm this feature. These investigations consisted of the of a transjugular intrahepatic portosystemic shunt) (3 patients), measurement of a large panel of inflammatory mediators in and (3) discontinuation of albumin treatment (7 patients). biobanking material from the Pilot-PRECIOSA study and Three of the remaining 21 patients died within the study from the INFECIR-2 study, a RCT aimed at comparing the period, and 3 did not give informed consent for car- efficacy of antibiotics alone vs albumin plus antibiotics in diocirculatory and hepatic hemodynamic assessment. The patients with decompensated cirrhosis and bacterial infec- clinical characteristics at enrollment and the main complica- tion unrelated to SBP.6 CLINICAL LIVER tions and causes of death during the study period are presented in Supplementary Table 1. Chronogram. Day 0. Samples were obtained for stan- Methods dard laboratory tests, serum albumin concentration, PRA (as The Pilot-PRECIOSA study and the INFECIR-2 study were a marker of effective blood volume), plasma concentrations approved by the corresponding ethics committees of each of atrial natriuretic peptide (ANP) and brain natriuretic hospital involved. The informed consent forms of the 2 studies peptide (BNP) (markers of central blood volume expansion), included the potential use of biobanking material for measuring and IL-6, followed by the hepatic and cardiocirculatory he- serum albumin levels and plasma renin and cytokine modynamic study. The methods for these studies have been concentrations. previously described (Supplementary Table 2).7 Immediately afterward, patients received the first albumin dose, and they were followed up for 20 weeks. The Pilot-PRECIOSA Study Weeks 1–12. The first 10 patients received an albumin The Pilot-PRECIOSA study (IG0802, registered at dose of 1g/kg body weight every 2 weeks for 12 weeks (a total ClinicalTrials.gov as NCT00968695) is a proof-of-concept, of 7 albumin treatments). PRA was measured every 2 weeks open-label, multicenter, nonrandomized (single-group), pro- before each albumin dose in the first 5 patients and ad hoc spective, phase 4 safety and dosage exploratory investigation weekly in the remaining 5 patients. Plasma IL-6 and serum sponsored by Grifols with the aim of getting preliminary albumin concentrations were measured every 2 weeks. An information to design a currently ongoing multicenter, ran- interim analysis in these first 10 patients showed that this dose domized, controlled therapeutic trial to assess the efficacy of of albumin was insufficient to normalize serum albumin con- long-term (1 year) albumin treatment in the prevention of centration throughout the last 10 weeks of the study period in ACLF and mortality in decompensated cirrhosis (PRECIOSA most patients (normal serum albumin concentration, 34–47 g/ study, ClinicalTrials.gov number NCT03451292). L). Accordingly, albumin dosage was increased to 1.5 g/kg body Investigators of the European Association for the Study of the weight every week in the remaining patients. Therefore, this Liver–Chronic Liver Failure (EASL-CLIF) Consortium from 3 second group of patients received a higher albumin dosage per hospitals (Hospital Clinic and Hospital de Sant Pau in Barcelona treatment and more albumin treatments (13) within the same and Hospital Ramón y Cajal in Madrid, Spain) participated in the time period (on day 0 and then every week for 12 weeks). design and implementation of the study, which started in July Samples for PRA were taken ad hoc weekly during treatment. 2009 and was completed in April 2014. These hospitals use the Samples for serum concentration of albumin and plasma levels same methodology for cardiocirculatory and hepatic hemody- of IL-6 were obtained every 2 weeks. For the description of the namic studies and have extensive experience in cooperative results, the group of patients who received albumin at a dose of hemodynamic, pathophysiological, and therapeutic studies. 1g/kg every 2 weeks was defined as the low albumin dosage Nonstandard laboratory measurements (hormones and bio- (LAlbD) group and that receiving albumin at a dose of 1.5 g/kg markers estimating systemic inflammation) were centralized at every week as the high albumin dosage (HAlbD) group. the Hospital Clinic. The results of the Pilot-PRECIOSA study were Week 14. Two weeks after the last albumin dose, the submitted to embargo until the start of the PRECIOSA study. cardiopulmonary and hepatic hemodynamic study was Inclusion and Exclusion Criteria and Patients repeated. Evaluated. The study enrolled noninfected patients with Post Hoc Measurements of Cytokines, Chemo- decompensated cirrhosis and severe circulatory dysfunction as kines, and Other Inflammatory Markers. The post hoc defined by the presence of ascites, renal dysfunction (serum assessment of the effects of albumin treatment on systemic creatinine  1.2 mg/dL or blood urea nitrogen  25 mg/dL or inflammation was performed by assessing a large panel of in- dilutional hyponatremia [serum sodium  130 mEq/L]), high flammatory mediators and biomarkers, including 24 cytokines, 10 levels of plasma renin activity (PRA) ( 2 ng/mL$h), and need chemokines, 4 growth factors, and 6 markers of endothelial for diuretic treatment to prevent ascites recurrence (at least dysfunction (2), coagulation/platelet dysfunction (2), and mono- 200 mg of spironolactone or 100 mg of spironolactone and 40 cyte activation (2), in biobanking material (September 2018). 152 Fernández et al Gastroenterology Vol. 157, No. 1 The INFECIR-2 Study serum albumin, PRA, and plasma IL-6. We averaged all the The INFECIR-2 study is an EASL-CLIF Consortium available values within the last 10 weeks of treatment to obtain investigator-promoted, phase 4, randomized, open-label, par- a single on-treatment value for comparison with the corre- allel, multicenter trial promoted by the Fundació Clínic (Hos- sponding baseline value. pital Clínic, University of Barcelona, Spain). It began in Results are presented as median and interquartile range. For September 2014 and was finished in December 2016 univariate analysis, Mann-Whitney test and Wilcoxon signed rank (ClinicalTrials.gov number NCT02034279). The inclusion and test were used for non-normally distributed variables. In all sta- exclusion criteria are detailed in the Supplementary Materials. tistical analyses, significance was set at P < .05. Analyses were The study aimed to assess the efficacy of short-term albumin performed with SAS, version 9.4 (SAS Institute, Cary, NC) statistical treatment in the prevention of ACLF and hospital mortality in packages. Graphs were performed with GraphPad Prism, version 136 patients with decompensated cirrhosis and acute bacterial 5.00 (GraphPad Software, San Diego, CA). infections unrelated to SBP. Eighteen patients were considered to have been erroneously included. Therefore, 118 patients were randomly assigned to receive either antibiotics alone Results (antibiotics-alone group, n ¼ 57), or antibiotics plus 2 albumin Baseline Clinical Characteristics of the Patients doses (ie, 1.5 g/kg at inclusion [day 1] and 1 g/kg on day 3) CLINICAL LIVER (albumin-plus-antibiotics group, n ¼ 61). Plasma samples for Included in the Pilot-PRECIOSA Study biobanking were obtained on day 1, before the administration All of the 18 patients included were admitted to the of the first albumin dose; on day 3, before the second albumin hospital for the treatment of ascites; 3 had diabetes mellitus, dose; and/or on day 7 in 48 patients from the albumin-plus- 1 had hepatocellular carcinoma, and 2 had minimal hepatic antibiotics group and 47 patients from the antibiotics-alone encephalopathy. Other characteristics at enrollment are group. “On-treatment” values of plasma cytokine levels given presented in Supplementary Table 1. in the article represent the average of those obtained on days 3 and 7 (in patients with 2 measurements) or those obtained on day 3 or 7 in patients with only a single measurement. Both Effect of Long-Term Albumin Treatment and Its groups were similar regarding patient characteristics (except Dosage on Serum Albumin Concentration (Pilot- for the combined prevalence of ACLF and kidney dysfunction at PRECIOSA Study) baseline, which was higher in the albumin arm), type of infec- Thirteen out of the 18 patients included who completed the tion, and antibiotic therapy. The results of the INFECIR-2 study sequential measurement of plasma albumin concentration had have recently been reported.6 baseline hypoalbuminemia (serum albumin concentration < The current study used biobanking aliquots from the 34 g/L). The effect of albumin treatment on serum albumin INFECIR-2 study for measurement of the serum concentration concentration was related to 2 factors. The first factor was al- of albumin, the plasma concentration of renin (PRC), and the bumin dosage. Although patients of the LAlbD group with plasma concentrations of the same panel of cytokines, chemo- baseline hypoalbuminemia (n ¼ 7) exhibited increases in kines, growth factors, and other inflammatory markers studied serum levels of albumin during treatment, only 1 achieved in the Pilot-PRECIOSA study. Measurements were performed at normalized the serum albumin concentration, that is, had an baseline and during treatment among 40 patients from the increase in albumin level to a value 34 g/L in all measure- antibiotics-alone group and 38 patients from the albumin-plus- ments. In contrast, all patients of the HAlbD group with baseline antibiotics group. The prespecified criteria to select these 78 hypoalbuminemia (n ¼ 6) achieved on-treatment normalized patients were (1) availability of biobanking samples, (2) infection receiving appropriate empirical antibiotic treatment serum albumin concentration (P < .001) Figure 1A). The me- (3) absence of severe complications within the first week of dian increases in serum albumin among patients receiving treatment that could affect the interpretation of the results, and HAlbD or LAlbD are detailed in Table 1 for all patients and in (4) completion of 1 week of follow-up. Figure 1B specifically for patients with baseline hypo- albuminemia. Four of the 5 patients with normal baseline serum albumin concentration (3 from the LAlbD group and 2 Laboratory Methods from the HAlbD group) showed relatively stable serum albumin Hormones and IL-6 were measured by radioimmunoassay concentration (always within the normal limits of 34–47 g/L) (PRA), chemiluminescent immunoassay (PRC), immunoassay throughout the study (Figure 1C). The fifth patient exhibited an (ANP and BNP), and enzyme-linked immunosorbent assay on-treatment increase in serum albumin, but this was also (ELISA) (IL-6). Measurement of the panel of cytokines, che- within normal limits. mokines, and other inflammatory mediators in patients from The second factor influencing the effect of albumin the Pilot-PRECIOSA and INFECIR-2 studies were performed by treatment was hypoalbuminemia grade at baseline. There using 2 multiplex immunoassays based on Luminex multi- was a significant inverse correlation between the baseline analyte profiling technology (Luminex Corp, Austin, TX). The serum albumin concentration and the median change in plasma levels of sCD163 and sMR/sCD206 were determined by serum albumin during treatment in both the HAlbD and the ELISA. Methods are detailed in the Supplementary Materials. LAlbD groups (Figure 1D): the lower the baseline albumin concentration, the higher the median increase in the serum Statistical Methods albumin levels achieved during treatment. The response to In the Pilot-PRECIOSA study, for a given patient receiving albumin treatment at each level of serum albumin concen- albumin treatment, there were several available results for tration was higher in patients receiving HAlbD. July 2019 Pleiotropic Effects of Albumin Therapy 153 Effect of Long-Term Albumin Treatment and Its Effect of Long-Term Albumin Treatment and Its Dosage on Plasma Renin Activity (Pilot- Dosage on a Large Panel of Plasma Cytokines PRECIOSA Study) (Pilot-PRECIOSA Study) Long-term albumin treatment was not associated with sig- The finding that elevated baseline plasma IL-6 levels, as nificant suppression in PRA in patients receiving HAlbD or determined by ELISA, can be reduced by albumin therapy in LAlbD, suggesting a minor effect on the effective blood volume a dose-dependent manner prompted us to investigate the (Table 1). Figure 1E and F shows the individual time-course effects of this treatment on the plasma levels of a large changes of PRA in patients receiving LAlbD and HAlbD, number of cytokines (n ¼ 24) in biobanking samples ob- respectively. An intriguing observation was the extreme insta- tained at baseline and at week 6 of albumin treatment in 10 bility of effective blood volume, as indicated by the development patients from the LAlbD group and 5 patients from the of acute, high, and transient positive peaks of PRA (increase in HAlbD group. In addition, we measured the plasma levels of PRA >100% to levels over 10 ng/mL$h) in a significant number the 24 cytokines in 25 healthy donors recruited at the of patients. Peaks were observed more frequently in the LAlbD Hospital Clínic Blood Bank. group (6 patients, 60%) than in the HAlbD group (1 patient, Among the 24 cytokines measured, 11 were not detectable 12.5%) (P ¼ .04), suggesting that although albumin treatment in any patient/healthy individual. Baseline values of all but 2 CLINICAL LIVER was not effective in improving mean effective blood volume, it of the remaining 13 cytokines included in the panel were was capable of stabilizing circulatory function. significantly higher among patients with decompensated cirrhosis than among healthy individuals, confirming the ex- Effect of Long-Term Albumin Treatment and Its istence of full-blown systemic inflammation in decom- pensated cirrhosis (Table 2). In the next tables, only changes Dosage on Plasma Interleukin 6 Levels (Pilot- in relevant cytokines are presented. Patients receiving LAlbD PRECIOSA Study) experienced only a small reduction or moderate increase To explore the possibility that albumin treatment can affect during treatment in the plasma levels of these cytokines, a systemic inflammation, we sequentially measured the plasma feature that contrasts sharply with the marked suppression of levels of IL-6 on day 0 and every 2 weeks after day 0 in the Pilot- most cytokines in patients receiving HAlbD (Table 3 and PRECIOSA study. IL-6 is a paradigmatic proinflammatory Figure 2C and D). These results strongly suggest that long- cytokine whose plasma levels are increased in most patients term albumin treatment, if given at high dosage, has a sig- with cirrhosis and systemic inflammation.5 Nine patients from nificant immunomodulatory effect in decompensated the LAlbD group and 7 from the HAlbD group had measurable cirrhosis, reducing the degree of systemic inflammation. levels of IL-6 at baseline and during treatment. The effect of albumin treatment on systemic inflammation in each patient could then be estimated as the absolute or percent change of IL- Effects of Long-Term Albumin Treatment on 6 between baseline value and on-treatment value (Table 1). The Systemic and Splanchnic Hemodynamics, median baseline value for plasma IL-6 levels in the 16 patients Natriuretic Peptides, and Liver and Renal was well above the normal range, consistent with the existence Function (Pilot-PRECIOSA study) of systemic inflammation in this group of patients. We arbi- Treatment with HAlbD, but not with LAlbD, was associated trarily defined a patient as having developed significant with a significant increase in cardiac index, systolic volume, immunomodulatory response to albumin treatment when the and left ventricular (LV) stroke work index, indicating an in- on-treatment IL-6 level decreased by more than 20% below the crease in LV function (Table 4). There were no changes in most baseline level. An outstanding finding of the current study was parameters estimating cardiac preload, including atrial pres- that the majority of patients receiving HAlbD (6 of 7 patients, sure, pulmonary capillary wedged pressure, and plasma con- 85.7%) but only 1 of 9 patients receiving LAlbD (11%) (P ¼ .003 centrations of ANP and BNP. There was, however, a significant for between-group comparison) had a reduction of plasma IL-6 increase in mean pulmonary artery pressure in patients of >20%, suggesting that long-term treatment with HAlbD, but receiving HAlbD, although it might be related to improvement not with LAlbD, induces significant immunomodulatory effects in right ventricular function. All patients had severe portal in patients with decompensated cirrhosis. Consistent with these hypertension at enrollment. HAlbD and LAlbD treatment, findings, we found that the median reduction from baseline for however, was not associated with significant changes in he- IL-6 was significantly greater among patients receiving HAlbD patic venous pressure gradient, a sensitive marker of portal than among those receiving LAlbD, whether reduction was pressure. There were also no major changes in other relevant expressed by percentage or absolute values (Table 1). standard laboratory parameters in either group. A second important finding was that systemic inflam- mation was unstable in a significant number of patients (1 of 7 receiving HAlbD and 4 of 9 receiving LAlbD), with acute, Effect of Short-Term Albumin Treatment on high, and reversible peaks of the plasma IL-6 (ie, increases Serum Albumin and Plasma Levels of Renin and by at least 100% to levels over 100 pg/mL) during albumin of a Large Panel of Inflammatory Cytokines in treatment (Figure 2A). The remaining 11 patients showed Patients With Infections (INFECIR-2 Study) small changes (mainly patients receiving LAlbD) or marked Next, we asked whether albumin therapy could have a reductions (mainly patients receiving HAlbD) of on- reducing effect on plasma cytokine levels in patients with treatment IL-6 (Figure 2B). bacterial infections included in the INFECIR-2 study. 154 Fernández et al Gastroenterology Vol. 157, No. 1 Bacterial infections are known to result in an enhancement and among those assigned to receive albumin plus antibi- of the systemic inflammation already present in patients otics (Table 5). with decompensated cirrhosis.5,7 This explains why baseline Treatment with antibiotics alone was not associated levels of tumor necrosis factor a, IL-4, IL-6, and IL-10 were with significant changes in most cytokines. Only 1 patient significantly higher among patients included in the INFECIR- showed a significant suppression (tumor necrosis factor a) 2 study than among those included in the Pilot-PRECIOSA during treatment. In contrast, patients treated with albumin study (Table 2). As expected for a randomized trial, in the plus antibiotics had, during treatment, a significant decrease INFECIR-2 study, the baseline plasma cytokine levels were or a clear trend for a reduction in most cytokines (Table 5), similar among patients assigned to receive antibiotics alone suggesting that albumin associated with antibiotics was CLINICAL LIVER July 2019 Pleiotropic Effects of Albumin Therapy 155 Table 1.Serum Albumin, PRA, and Plasma Levels of IL-6 at Baseline and During Albumin Treatment in 18 Patients With Decompensated Cirrhosis Unrelated to Bacterial Infection Who Were Enrolled in the Pilot-PRECIOSA Study and Divided Into 2 Groups Depending on Whether They Received LAlbD or HAlbD Treatment Patient Group HAlbD group LAlbD group P Parameter (n ¼ 8) (n ¼ 10) valuea Serum albumin concentration Baseline, g/L 27.6 (22.7 to 34.0) 26.5 (24.8 to 40.3) .83 On-treatment average value, g/L 39.2 (38.7 to 43.0)b 33.3 (31.8 to 37.9)c .004 Absolute change, g/L 12.7 (8.5 to 16.6)b 5.7 (–1.8 to 8.0)c .01 Percent change, % 48.7 (26.7 to 71.3)b 20.2 (–4.1 to 32.5)c .04 Plasma renin activity Baseline, ng/mL$h 5.5 (3.6 to 7.9) 7.9 (3.8 to 12.3) .41 On-treatment average value, ng/mL$h 4.9 (3.9 to 5.8) 6.9 (3.8 to 11.3) .17 Absolute change, ng/mL 0.2 (–4.2 to 1.3) –0.4 (–5.5 to 5.8) .64 CLINICAL LIVER Percent change, % 2.0 (–44.4 to 36.2) –6.7 (–45.3 to 146.2) .69 Plasma IL-6 concentration Baseline, pg/mL 123.5 (51.5 to 151.5) 41.5 (25.8 to 75.0) .02 On-treatment average value, pg/mL 62.5 (24.5 to 93.6)b 57.5 (30.0 to 79.2) .76 Absolute change, pg/mL –53.0 (–108.0 to –18.0)b –3.2 (–11.1 to 30.0) .04 Percent change, % –56.0 (–68.8 to –24.2)b –7.6 (–15.7 to 79.7) .04 NOTE. Values are reported as median (interquartile range). For each variable in each patient, the average value during treatment was obtained by using all on-treatment values of this variable available. a P value for the between-group comparison. Values in bold indicate P < .05. b P < .05 for the within-group comparison with baseline values. c P ¼ .05 for the within-group comparison with baseline values. more effective than antibiotics alone in attenuating baseline (–1.2 [–26.5 to 129.6] mIU/mL). In contrast, albumin-plus- systemic inflammation in patients with bacterial infections. antibiotics treatment significantly increased serum albu- In the INFECIR-2 study, baseline values for serum al- min concentration (7.0 [4.0–10.0] g/L, P < .0001) and bumin concentration were similar between patients of the suppressed PRC (–40. 5 [–272.9 to –4.5] mIU/mL; P ¼ .002). antibiotics-alone group (26 [20–30] g/L) and those of the albumin-plus-antibiotics group (25 [19–30] g/L) (P ¼ .91). Effect of Short-Term and Long-Term Albumin The baseline activity of the renin-angiotensin system, esti- Treatment on Chemokines; Growth Factors; and mated by PRC, was greater among patients in the albumin- plus-antibiotics group than among those in the antibiotics- Biomarkers of Macrophage Activation, alone group, although the difference was not statistically Endothelial Dysfunction, and Coagulation/ significant (241.6 [46.3–903.0] mIU/mL and 125.0 [34.0– Platelet Function (Pilot-PRECIOSA and INFECIR- 398.6] mIU/mL, respectively; P ¼ .25). Antibiotics alone 2 Studies) were not associated with significant changes from baseline To have a comprehensive view of the effects of albumin for serum albumin concentration (0 [–20 to 1.0] g/L) or PRC treatment in both the Pilot-PRECIOSA and the INFECIR-2 = Figure 1. Changes in serum albumin concentration and PRA induced by treatment with HAlbD (blue color in all panels) and LAlbD (red color) in the 18 patients included in the Pilot-PRECIOSA Study. (A) Individual changes in serum albumin con- centration among the 13 patients with baseline hypoalbuminemia (serum albumin concentration < 34 g/L). The horizontal lines indicate the upper and lower normal limits of serum albumin. All 6 patients with hypoalbuminemia treated with HAlbD developed a rapid increase (within 2 weeks) in serum albumin concentration up to normal levels, which persisted during the remaining 10 weeks of the study. In contrast, although all 7 patients with baseline hypoalbuminemia treated with LAlbD had increased serum levels of albumin during treatment, only 1 normalized the serum albumin concentration. (B–D) Two factors influenced the response to albumin treatment. (B) The first factor was the albumin dosage: among the 13 patients with baseline hypoalbuminemia, the individual absolute median increase in serum albumin was almost double in patients receiving HAlbD vs those receiving LAlbD. (C) The second factor was the feedback mechanism by which baseline serum albumin concentration influences the hepatic synthesis of albumin. In most patients without hypoalbuminemia, the inhibition of hepatic synthesis of albumin prevented the increase in the serum concentration of albumin to abnormal levels during albumin treatment. (D) This feedback mechanism was also reflected by the close inverse correlation between the baseline serum albumin concentration and the mean increase in serum albumin during treatment. The lower the baseline levels of serum albumin, the higher the absolute mean increase in the serum concentration of albumin in both the LAlbD and HAlbD groups. (E) Circulatory dysfunction was extremely unstable during albumin treatment in patients receiving LAlbD, with high peaks of PRA in 6 patients. (F) Circulatory instability was significantly improved in patients receiving HAlbD, with only 1 patient presenting 1 peak of PRA throughout treatment. 156 Fernández et al Gastroenterology Vol. 157, No. 1 CLINICAL LIVER Figure 2. Changes in IL-6 and other cytokines induced by treatment with HAlbD (blue color in all panels) and LAlbD (red color) in the 15 patients included in the Pilot-PRECIOSA study with sequential cytokines measurements. (A) The degree of systemic inflammation, as estimated by repeated measurements of plasma IL-6 in baseline conditions and during treatment, was extremely unstable in 4 patients receiving LAlbD and in 1 receiving HAlbD. (B) In the remaining patients, there was a marked suppression of the circulating plasma levels of IL-6 (mainly in patients receiving HAlbD) or no change to minor changes (mainly in patients receiving LAlbD). (C, D) Data derived from the assessment of a large panel of inflammatory cytokines at baseline and at week 6 showed that plasma levels of IL-6, vascular endothelial growth factor, G-CSF, and IL-1 receptor antagonist had a median reduction from baseline (interquartile range, %) that was significantly greater among patients treated with HAlbD than among those receiving LAlbD. G-CSF, granulocyte colony-stimulating factor; IL1-ra, IL-1 receptor antagonist. studies, we assessed a broad variety of soluble factors, endogenous and exogenous ligands.1 The ability of albumin including chemokines, growth factors, and markers of to bind proinflammatory molecules such as pathogen- macrophage activation and endothelial and coagulation/ associated molecular patterns (e.g., the Gram-negative bac- platelet dysfunction. As shown in Supplementary Tables 3 teria byproduct lipopolysaccharide),9 prostaglandins,10 ni- and 4, in both studies, albumin treatment was associated tric oxide,11 and reactive oxygen and nitrogen species12 with minor or no changes in most of these factors, sug- could be of great importance in the context of cirrhosis, gesting that it exerts its immunomodulatory effect mainly by because these molecules are involved in the pathogenesis of influencing production and/or release of specific cytokines. the systemic inflammation and circulatory and organ dysfunction/failure that characterize decompensated cirrhosis and ACLF.13 Because the occurrence of these non- Discussion osmotic effects of albumin therapy in cirrhosis was elusive, Current albumin dosage in cirrhosis is based on empir- there was an urgent need to address this question, which ical assumptions and on the concept that albumin mainly gave rise to the present study. acts as a plasma volume expander.8 Of note, albumin ther- The current article describes 5 important and, to our apy can have many other important biological effects, knowledge, previously unreported observations on the because it is able to bind to and inactivate a wide range of pathophysiology and albumin treatment of decompensated July 2019 Pleiotropic Effects of Albumin Therapy 157 Table 2.Baseline Plasma Levels of Cytokines Among Healthy Individuals (HS), Patients From the Pilot-PRECIOSA (P-PR) Study, and Patients From the INFECIR-2 (INF-2) Study P valuea HS P-PR Study INF-2 Study Cytokine (n ¼ 25) (n ¼ 15) (n ¼ 78) HS vs p-PR HS vs INF-2 p-PR vs INF-2 TNF-a Median level (IQR), pg/mL 12.3 (11.5–16.9) 21.8 (16.0–30.6) 32.0 (21.9–49.8) .001 .0001 .04 Missing variable, n (%) 0 (0) 0 (0) 0 (0) G-CSF Median level (IQR), pg/mL 3.6 (2.4–5.7) 20.0 (8.8–156.9) 74.4 (32.0–155.5) .0008 .0001 .11 Missing variable, n (%) 0 (0) 1 (7) 17 (22) IL-1ra Median level (IQR), pg/mL 7.1 (3.8–13.2) 13.1 (9.1–32.8) 29.6 (8.3–76.5) .02 .0001 .26 Missing variable, n (%) 0 (0) 0 (0) 0 (0) IL-6 Median level (IQR), pg/mL 0.9 (0.9–0.9) 10.5 (8.0–25.1) 37.1(22.6–107.6) .0001 .0001 .0001 CLINICAL LIVER Missing variable, n (%) 0 (0) 0 (0) 0 (0) IL-10 Median level (IQR), pg/mL 1.1 (1.1–2.4) 2.7 (0.8–10.8) 10.9 (6.7–19.0) .20 .0001 .02 Missing variable, n (%) 0 (0) 3 (20) 13 (17) IL-17A Median level (IQR), pg/mL 0.7 (0.7–3.3) 17.7 (2.3–32.4) 3.2 (1.4–7.2) .0002 .002 .05 Missing variable, n (%) 0 (0) 3 (20) 0 (0) IFNg Median level (IQR), pg/mL 3.0 (2.2–4.7) 6.7 (2.1–35.8) 6.8 (1.5–18.4) .11 .04 .49 Missing variable, n (%) 0 (0) 0 (0) 0 (0) VEGF Median level (IQR), pg/mL 24.4 (14.7–45.2) 59.0 (26.3–230.7) 61.0 (32.1–183.0) .02 .002 .77 Missing variable, n (%) 0 (0) 1 (7) 32 (41) G-CSF, granulocyte colony-stimulating factor; IFN, interferon; IL-1ra, IL-1 receptor antagonist; IQR, interquartile range; TNF, tumor necrosis factor; VEGF, vascular endothelial growth factor. a Bold type indicates P < .05. Italic type indicates P ¼ .05. cirrhosis. The first is that the long-term albumin dosage with HAlbD in patients without bacterial infection also oc- required to normalize serum albumin concentration is much curs after short-term (1 week) HAlbD treatment in patients higher than that used in the therapeutic RCTs so far per- with bacterial infections. For these objectives, we leveraged formed.4,14 Second, circulatory dysfunction is not a steady the availability of biobanking material from the Pilot- state or a slowly progressive process, as it has been tradi- PRECIOSA and INFECIR-2 studies. tionally considered, but rather an extremely unstable con- The initial albumin dose evaluated in the Pilot-PRECIOSA dition. Third, systemic inflammation in cirrhosis is also study (1 g/kg every 2 weeks for 12 weeks) was based on unstable, with acute episodes of bursts of circulating cyto- that used in the pioneer RCT by Gentilini et al14 (25 g per kines in the absence of any identifiable precipitating event. week) that explored the effect of long-term of albumin Fourth, HAlbD, but not LAlbD treatment, is associated with treatment on the response to diuretics in patients with significant improvement in LV function in decompensated cirrhosis ascites and in 2 RCTs that explored the long-term cirrhosis, which is currently considered an important effect of albumin administration on the natural course of mechanism of systemic circulatory dysfunction.15 Finally, decompensated cirrhosis (the ANSWER study, 40 g of al- and most importantly, the sequential assessment of the bumin every week,4 and the MATCH study, 40 g of albumin plasma levels of IL-6 during albumin treatment showed, for every 2 weeks16). The results of the current study indicate the first time to our knowledge, that long-term albumin that a dose of 1g/kg, which is higher than the MATCH study treatment at high dosage has immunomodulatory effects in dose and only slightly lower than the ANSWER study dose, decompensated cirrhosis. The transcendence of this last was clearly insufficient to normalize serum albumin con- finding was the reason to complete the study with 2 addi- centration in 7 of the 8 patients with hypoalbuminemia tional investigations. The first was aimed at assessing included in the LAlbD group. In contrast, our second albu- whether the suppressive effect of albumin on IL-6 observed min dosage (1.5 g/kg per week) rapidly normalized serum in the patients included in the Pilot-PRECIOSA study also albumin concentration in all patients with hypoalbuminemia extended to other cytokines and inflammatory molecules. included in the HAlbD group. The second was aimed at investigating whether the immu- The time-course changes of serum albumin concentra- nomodulatory effect observed during long-term treatment tion during albumin treatment suggest that the homeostatic 158 Fernández et al Gastroenterology Vol. 157, No. 1 G-CSF, granulocyte colony-stimulating factor; IFN, interferon; IL-1ra, IL-1 receptor antagonist; IQR, interquartile range; TNF, tumor necrosis factor; VEGF, vascular baseline baseline feedback mechanism by which hepatic albumin synthesis for between-group Absolute Percent change change from is regulated by the serum albumin concentration17 is fully .19 .03 .04 .01 .43 .14 .12 .01 comparisona P value operative in patients with advanced cirrhosis. Normali- zation of serum albumin concentration in patients with Table 3.Plasma Levels of Cytokines at Baseline and at The 6th Week of Treatment in Patients Receiving Either LAlbD or HAlbD in the Pilot PRECIOSA Study from hypoalbuminemia receiving HAlbD occurred very rapidly .12 .05 .03 .01 .35 .07 .12 .03 (within 2 weeks) after the onset of albumin treatment, but once normalized, it remained within normal limits –53.2) –13.8) –46.8) –51.7) –15.2) –16.8) throughout the study despite the weekly administration of 21.3) 7.5) median (IQR) change from albumin at a concentration of 20 g/dL (5 times higher baseline, Percent to to to to to to to to than the normal serum albumin concentration). This rapid (–16.1 (–79.5 (–82.9 (–67.3 (–66.2 (–82.4 (–65.6 (–91.4 and intense initial increase in serum albumin concentra- tion was probably the consequence of the combination of –15.1 –63.1 –70.3 –50.1 –24.4 –59.6 –51.7 –75.2 increased albumin synthesis by the liver secondary to hypoalbuminemia and the effect of the exogenous albumin (6.1 to 315.5) –63.1 (–79.5 to –53.2) administrations. In contrast, after normalization of serum (10.5 to 28.4) –9.2 (–14.2 to –5.0) –9.2 (–16.4 to –1.5) –4.5 (–14.8 to –0.2) (23.5 to 50.1) –8.4 (–17.6 to –4.0) –3.4 (–14.9 to 0.7) –4.0 (–7.0 to –2.8) median (IQR) (18.4 to 53.6) –4.9 (–9.3 to 0.7) change from CLINICAL LIVER albumin, the inhibitory effect of normo-albuminemia on baseline, Absolute albumin synthesis precluded any further increase in HAlbD group serum albumin concentration, despite continuous albumin (n ¼ 5) treatment. The homeostatic feedback mechanism of serum albumin would also explain why albumin treatment did not increase serum albumin concentration in patients without hypoalbuminemia. For additional explanatory cytokine level, (6.7 to 29.3) (2.2 to 27.6) (2.9 to 19.9) (1.7 to 22.6) median (IQR) details, see Figure 1 legend. Baseline pg/mL, The most relevant finding of our study was the observation that both long-term and short-term albumin treatment, if given at high dosage, are associated with 8.5 5.6 8.7 30.9 47.3 10.7 15.4 26.3 significant immunomodulatory effects in decompensated Undetectable levels, n (%) cirrhosis. Three lines of evidence supported this conclu- sion. The first derived from the sequential measurement (20) (20) (0) (0) (0) (0) (0) (0) of IL-6 during albumin treatment in patients included in 0 1 0 0 1 0 0 0 the pilot-PRECIOSA study. The median reduction from baseline of plasma IL-6 levels was significantly greater (–14.5 to 60.3) (–15.2 to 53.6) (–11.7 to 83.9) (–42.8 to 27.0) (–46.1 to 40.8) (–3.5 to 15.9) (–7.7 to 32.0) among patients receiving HAlbD than among those (0.0 to 29.2) median (IQR) change from receiving LAlbD. This finding is important considering baseline, Percent that IL-6 has broad effects on immune and nonimmune cells and often displays hormone-like characteristics that can affect homeostatic processes.18 The second line of –3.1 5.6 11.0 20.5 28.8 44.1 12.4 11.7 evidence derived from the analysis of the effect of albumin 4.9 (–1.4 to 13.8) 2.8 (–1.2 to 18.9) treatment on cytokines other than IL-6 in biobanking Bold type indicates P < .05. Italic type indicates P ¼ .05. (56.1 to 230.7) 26.5 (0.0 to 50.8) 1.8 (–0.7 to 3.5) 0.8 (–2.5 to 7.4) 0.3 (–0.7 to 1.3) 0.65 (–0.9 to 9.8) 0.2 (–0.6 to 2.2) median (IQR) change from material from the Pilot-PRECIOSA study. This investiga- baseline, Absolute tion confirmed the observations of the first investigation. Treatment with HAlbD, but not with LAlbD, was associ- LAlbD group (n ¼ 10) ated with significant decreases in plasma IL-6 during treatment. Moreover, it showed that this effect also involved other keystone cytokines (eg, granulocyte (13.6 to 28.1) (10.2 to 35.3) (8.8 to 156.9) cytokine level, (6.5 to 24.6) (0.6 to 10.8) (1.4 to 33.7) (2.6 to 49.0) median (IQR) colony-stimulating factor), confirming that long-term Baseline pg/mL, therapy with HAlbD, but not with LAlbD, induces a sig- nificant and extensive immunomodulatory effect in decompensated cirrhosis. Finally, the third line of evi- 8.9 1.8 5.6 198.0 20.3 20.0 13.1 24.7 dence was obtained from the analysis of biobanking endothelial growth factor. plasma samples from the INFECIR-2 study. Treatment Undetectable Cytokine levels, n (%) with albumin plus antibiotics was associated with a rapid, significant, and widespread suppression of the circulating (20) (30) (10) (0) (0) (0) (0) (0) levels of cytokines, an effect not observed with antibiotics 0 0 0 0 2 3 0 1 alone. It was interesting to observe that the immuno- modulatory effect of albumin in the Pilot-PRECIOSA and INFECIR-2 studies was related mainly to the inhibitory G-CSF IL-17A TNF-a IL-1ra VEGF IL-10 effect of albumin on cytokine production but not to an IFNg IL-6 effect on other inflammatory molecules. a July 2019 Pleiotropic Effects of Albumin Therapy 159 Table 4.Effects of LAlbD and HAlbD Administration on Cardiovascular and Splanchnic Hemodynamics, Cardiac Peptides and Standard Liver and Renal Function Parameters LAlbD (n ¼ 10) HAlbD (n ¼ 8) P P Baseline Week 14 value Baseline Week 14 value Systemic hemodynamics n¼7 n ¼8 RAP, mm Hg 6 (4–8) 4 (4–10) 0.87 8 (4–8) 9 (6–9) .26 MPAP, mm Hg 16 (15–17) 15 (12–20) 0.55 15 (11–18) 18 (15–25) .02 PCWP, mm Hg 10 (9–11) 8 (7–13) 0.74 11 (8–15) 12 (10–14) .11 Cardiac index, L/min/m2 3.9 (1.8–4.6) 3.8 (2.3–5.3) 0.09 4.2 (3.0–5.0) 5.3 (3.1–6.8) .04 Heart rate, bpm 61 (59–82) 75(62–86) 0.21 69 (59–91) 68 (62–76) .21 Systolic volume, mL 120 (40–127) 90 (52–135) 0.74 125 (85–145) 165 (110–190) .04 LV stroke work index, g$m/m2 48 (24–64) 44 (27–68) 0.74 54 (49–69) 82 (51–97) .04 SVRI, dyn$s/cm5/m2 1158 (1042–3840) 1182 (944–2762) 0.18 1257 (952–1693) 1072 (728–1183) .09 MAP, mm Hg 78 (63–88) 77 (75–85) 0.61 78 (74–85) 77 (66–84) .48 Cardiac peptides n¼9 n ¼6 CLINICAL LIVER ANP, fmol/mL 58 (23–84) 53 (37–64) 0.59 41 (13–87) 65 (29–155) .14 BNP, pg/mL 82 (25–221) 37 (34–126) 0.45 41 (32–69) 49 (18–128) .46 Splanchnic hemodynamics n¼7 n ¼6 FHVP, mm Hg 15 (9–17) 12 (5–17) 0.45 11 (8–14) 9 (5–10) .12 WHVP, mm Hg 35 (25–38) 30 (26–39) 0.67 32 (29–36) 28 (25–31) .21 HVPG, mm Hg 19 (15–20) 21 (14–22) 0.34 19 (17–27) 22 (18–25) .89 Liver and renal function n ¼ 10 n ¼8 AST, UI/L 58 (27–78) 53 (26–69) 0.15 45 (40–112) 34 (31–55) .02 ALT, UI/L 27 (19–46) 25 (16–34) 0.11 35 (28–56) 24 (21–39) .13 Serum creatinine, mg/dL 1.3 (1.0–1.4) 1.0 (0.9–1.3) 0.06 0.9 (0.8–1.3) 0.9 (0.7–1.2) .24 BUN, m/dL 26 (16–47) 22 (16–32) 0.16 20 (15–31) 24 (17–36) .25 Serum sodium, mEq/L 132 (126–136) 133 (129–137) 0.67 130 (129–135) 132 (131–134) .87 Serum albumin, g/L 27 (25–40) 35 (31–40) 0.06 27 (22–35) 40 (35–41) .03 Serum bilirubin, mg/dL 1.8 (1.0–2.1) 1.8 (1.1–2.8) 0.41 3.7 (1.9–13.0) 1.9 (1.6–16.8) .40 INR 1.3 (1.1–1.6) 1.5 (1.1–1.7) 0.29 1.4 (1.3–2.3) 1.4 (1.3–2.5) .46 Child-Pugh score, points 8 (6–10) 8 (7–9) 0.37 9 (8–11) 7 (6–8) .02 MELD score, points 14 (11–17) 14 (9–17) 0.59 16 (13–26) 16 (13–27) .25 ALT, alanine aminotransferase; ANP, atrial natriuretic peptide; AST, aspartate aminotransferase; BNP, brain natriuretic peptide; BUN, blood urea nitrogen; FHVP, free hepatic venous pressure; HVPG, hepatic venous pressure gradient; LV, left ventricular; MAP, mean arterial pressure; MELD, model for end stage liver disease; MPAP, mean pulmonary artery pressure; PCWP, pulmonary capillary wedge pressure; RAP, right atrial pressure; SVRI, systemic vascular resistance index; WHVP, wedge hepatic venous pressure. NOTE. Values are reported as median (interquartile range). Normal value ranges are as follow: right atrial pressure, 2–10 mm Hg; mean pulmonary arterial pressure, 10–25 mm Hg; pulmonary capillary pressure, 6–14 mm Hg; cardiac index, 2.5–4 L/min/ m2; SV, 60–100 mL; LV stroke work index, 45–75 g$m/m2/beat; SVRI, 1970–2390 dyn$sec/cm5/m2; MAP, 80–95 mm Hg; ejection fraction, >50%; ANP, 9–24 fmol/mL; BNP, 4–37 pg/mL; HVPG, 1–5 mm Hg. a Bold type indicates P < .05. An intriguing finding of our study was the observation of 1 of circulatory dysfunction and systemic inflammation in or 2 acute, intense, and spontaneously reversible peaks of decompensated cirrhosis. Although the current study is, to PRA and plasma IL-6 during albumin treatment in many pa- our knowledge, the first to show these abnormalities, the tients included in the Pilot-PRECIOSA study. There are rea- existence of such episodes of acute circulatory impairment sons to suggest that the prevalence and frequency of these and systemic inflammation had already been anticipated by peaks in the current study are not representative of their the systemic inflammation hypothesis as an explanation for actual prevalence and frequency in patients with decom- the 40% prevalence of ACLF in patients without any identi- pensated cirrhosis. First, we monitored PRA and plasma IL-6 fiable exogenous precipitating event of the syndrome.13,19 The only once every week or every 2 weeks during the study proposed mechanism of such changes by the systemic period. However, according to our data, the duration of these inflammation hypothesis is the existence of transient bursts of peaks may range from less than 1 to 2 or more weeks. translocation of viable bacteria or bacterial products from the Therefore, we could have lost a significant number of peaks in intestinal lumen to the systemic circulation. Therefore, the our patients. Interestingly, the prevalence of PRA and IL-6 potential futility of single measurements of renin and cyto- peaks was lower in patients receiving HAlbD than in those kines as surrogate markers of effective blood volume and receiving LAlbD, suggesting that treatment with HAlbD may systemic inflammation in patients with decompensated prevent the occurrence of these acute episodes of aggravation cirrhosis has to be considered in the design of future studies. CLINICAL LIVER 160 Fernández et al Table 5.Baseline Plasma Levels of Cytokines and Their Changes During the First Week of Treatment, in Patients From the INFECIR-2 Study Who Were Randomly Assigned to Receive Either Antibiotics Alone or Albumin Plus Antibiotics Antibiotics Alone Albumin plus (n ¼ 40) antibiotics (n ¼ 38) P value P value for change for change from baselinea from baselinea Absolute Baseline level, Absolute change Percent change P for P for Baseline level, change Percent change P for P for Undetectable pg/mL, from baseline, from baseline, absolute percent Undetectable pg/mL, from baseline, from baseline, absolute percent Cytokine levels, n (%) median (IQR) median (IQR) median (IQR) change change levels, n (%) median (IQR) median (IQR) median (IQR) change change TNF-a 0 (0) 37.9 (23.3 to 50.0) –2.8 (–12.6 to 3.2) –8.2 (–28.9 to 11.9) .05 .01 0 (0) 31.1 (21.2 to 45.3) –3.4 (–14.7 to 3.1) –16.2 (–40.5 to 12.8) .01 .04 G-CSF 9 (23) 74.4 (19.2 to 185.5) –3.4 (–55.3 to 11.5) –21.2 (–91.4 to 21.5) .33 .85 8 (21) 73.5 (33.6 to 115.0) –41.5 (–65.1 to 8.4) –58.6 (–88.8 to 30.2) .01 .01 IL-1ra 0 (0) 29.6 (8.3 to 71.5) –0.6 (–28.0 to 9.0) –5.9 (–51.5 to 23.6) .37 .31 0 (0) 29.9 (8.3 to 76.5) –0.5 (–34 to 2.1) –3.5 (–73.0 to 8.3) .05 .28 IL-6 0 (0) 37.7 (18.3 to 94.7) –7.0 (–19.9 to 20.7) –14.8 (–43.5 to 66.4) .53 .27 0 (0) 36.9 (23.9 to 158.9) –7.7 (–33.1 to 0.3) –23.0 (–55.0 to 3.8) .003 .005 IL-10 8 (20) 10.7 (6.3 to 20.9) –0.2 (–6.7 to 4.4) –1.8 (–56.7 to 50.1) .74 .99 5 (13) 11.0 (6.7 to 15.1) –1.5 (–7.6 to 2.8) –15.6 (–53.5 to 63.0) .03 .03 IL-17A 0 (0) 3.7 (1.2 to 8.2) 0.1 (–1.7 to 1.3) 2.5 (–38.6 to 81.1) .92 .99 0 (0) 2.7 (1.6 to 7.2) –0.5 (–2.7 to 0.4) –15.4 (–52.4 to 29.9) .05 .09 IFNg 0 (0) 4.8 (1.2 to 15.6) –0.2 (–6.2 to 1.3) –7.9 (–51.7 to 37.1) .17 .52 0 (0) 8.4 (2.0 to 19.7) –0.5 (–8.6 to 4.3) –24.8 (–63.3 to 52.9) .48 .24 VEGF 16 (40) 65.4 (45.6 to 204.0) –18.5 (–41.4 to 18.2) –16.4 (–45.3 to 29.4) .30 .31 16 (42) 50.6 (24.1 to 183.0) –13.9 (–49.0 to 11.9) –24.7 (–36.9 to 32.5) .10 .29 NOTE. Changes during the first week of albumin treatment were assessed between day 3 and day 7 after inclusion. There were no significant between-group differences in cytokine levels at baseline. G-CSF, granulocyte colony-stimulating factor; IFN, interferon; IL-1ra, IL-1 receptor antagonist; IQR, interquartile range; TNF, tumor necrosis factor; VEGF, vascular endothelial growth factor. a Bold type indicates P < .05. Italic type indicates P ¼ .05. Gastroenterology Vol. 157, No. 1 July 2019 Pleiotropic Effects of Albumin Therapy 161 Although circulatory dysfunction in cirrhosis has been Supplementary Material traditionally attributed to splanchnic arterial vasodilation, Note: To access the supplementary material accompanying there is now evidence that impairment in LV function also this article, visit the online version of Gastroenterology at plays a major role. In fact, the cardiac index in cirrhosis falls www.gastrojournal.org, and at https://doi.org/10.1053/ progressively from compensated cirrhosis to decom- j.gastro.2019.03.021. pensated cirrhosis and HRS.15 Our data show that normal- ization of serum albumin concentration with long-term HAlbD treatment in noninfected patients with decom- References pensated cirrhosis does not induce significant changes in 1. Arroyo V, García-Martínez R, Salvatella X. Human serum central blood volume and portal pressure. However, it was albumin, systemic inflammation, and cirrhosis. J Hepatol associated with a significant improvement in LV function. 2014;61:396–407. These observations are important for 2 reasons. First, they 2. Sort P, Navasa M, Arroyo V, et al. Effect of intravenous explain why treatment with HAlbD is generally not associ- albumin on renal impairment and mortality in patients ated with variceal bleeding or pulmonary edema in with cirrhosis and spontaneous bacterial peritonitis. decompensated cirrhosis without bacterial infections. Sec- N Engl J Med 1999;341:403–409. 3. Ortega R, Ginès P, Uriz J, et al. Terlipressin therapy with CLINICAL LIVER ond, because systemic inflammation induces direct delete- rious effect on heart function, our study supports the and without albumin for patients with hepatorenal syn- concept that the beneficial effect of albumin treatment in the drome: results of a prospective non-randomized study. management of organ dysfunction/failure in cirrhosis may Hepatology 2002;36:941–948. be mediated, at least in part, by its immunomodulatory ef- 4. Caraceni P, Riggio O, Angeli P, et al. Long-term albumin fect. In fact, this has also been observed in rats with carbon administration in decompensated cirrhosis (ANSWER): an tetrachloride-induced cirrhosis, which develop evidence of open-label randomised trial. Lancet 2018;391:2417–2429. systemic inflammation and inflammation in the cardiac tis- 5. Clària J, Stauber RE, Coenraad MJ, et al. Systemic sue associated with severe impairment of LV contractibility, inflammation in decompensated cirrhosis: characteriza- which reverses after albumin treatment.20 tion and role in acute-on-chronic liver failure. Hepatology 2016;64:1249–1264. One of the strengths of our study is the use of multiple plasma samples, the prospective collection of which was 6. Fernández J, Angeli P, Trebicka J, et al. Albumin treat- ment in patients with cirrhosis and infections unrelated to prespecified in the context of 2 well-designed multicenter SBP: the INFECIR-2 randomized trial. Clin Gastroenterol controlled trials, 1 of which was randomized. A limitation of Hepatol 2019. our study was the relatively low number of patients 7. Fernández J, Navasa M, Garcia-Pagan JC, et al. Effect of included in the Pilot-PRECIOSA study. However, the most intravenous albumin on systemic and hepatic hemody- important finding of this investigation, the significant namics and vasoactive neurohormonal systems in pa- immunomodulatory effect of albumin treatment in patients tients with cirrhosis and spontaneous bacterial with advanced cirrhosis, was confirmed by assessing the peritonitis. J Hepatol 2004;41:384–390. effect of albumin treatment on a large panel of inflammatory 8. European Association for the Study of the Liver. cytokines, both in patients included in the Pilot-PRECIOSA EASL clinical practice guidelines for the management study and in a relatively large number of patients of decompensated cirrhosis. J Hepatol 2018;69:406–460. included in the INFECIR-2 study, thus offering solid addi- 9. Gioannini TL, Zhang D, Teghanemt A, et al. An essential tional arguments supporting our conclusions. role for albumin in the interaction of endotoxin with In summary, the current study allowed us to uncover lipopolysaccharide-binding protein and sCD14 and resul- important findings related to the efficacy of albumin treat- tant cell activation. J Biol Chem 2002;277:47818–47825. ment in cirrhosis. The most outstanding were that high 10. O’Brien AJ, Fullerton JN, Massey KA, et al. Immunosup- doses of albumin, but not low doses of albumin, in patients pression in acutely decompensated cirrhosis is mediated with decompensated cirrhosis have significant immuno- by prostaglandin E2. Nat Med 2014;20:518–523. modulatory effects, prevent a phenomenon shown by the 11. Stamler JS, Jaraki O, Osborne J, et al. Nitric oxide cir- present study that consists of bursts of circulatory culates in mammalian plasma primarily as a S-nitroso dysfunction, improve LV function, and correct serum albu- adduct of serum albumin. Proc Natl Acad Sci U S A min levels without inducing albumin overdose, probably 1992;89:7674–7677. because of the preservation of negative feedback mecha- 12. Anraku M, Chuang VT, Maruyama T, et al. Redox prop- nisms controlling albumin synthesis, even in advanced liver erties of serum albumin. Biochim Byophys Acta 2013; disease. Because albumin is capable of binding to and 1830:5465–5472. inactivating many inflammatory promoters, such as 13. Bernardi M, Moreau R, Angeli P, et al. Mechanisms of pathogen-associated molecular patterns, bioactive lipid decompensation and organ failure in cirrhosis: from pe- metabolites, reactive oxygen species, and nitric oxide, the ripheral arterial vasodilation to systemic inflammation immunomodulatory effects of albumin could be related to hypothesis. J Hepatol 2015;63:1272–1284. this scavenging function. However, this explanation may be 14. Gentilini P, Casini-Raggi V, Di Fiore G, et al. Albumin too simplistic, and further investigations are clearly needed improves the response to diuretics in patients with to understand the anti-inflammatory effect of albumin cirrhosis and ascites: results of a randomized, controlled treatment in cirrhosis. trial. J Hepatol 1999;30:639–645. 162 Fernández et al Gastroenterology Vol. 157, No. 1 15. Ruiz-del-Arbol L, Monescillo A, Arocena C, et al. Circu- Acknowledgments Author contributions: Javier Fernández, Miriam Castro, Juan Acevedo, Luis latory function and hepatorenal syndrome in cirrhosis. Ruiz-del-Arbol, Cándido Villanueva, Aníbal Silva, Manuel Morales-Ruiz, and Hepatology 2005;42:439–447. Vincente Arroyo participated in the design and execution of the pilot 16. Solà E, Solé C, Simón-Talero M, et al. Midodrine PRECIOSA study (IG0802); Javier Fernández, Verónica Prado, Mireya Arteaga, Jonel Trebicka, Paolo Angeli, Manuela Merli, Carlo Alessandria, and albumin for prevention of complications in patients Niels Kristian Aagaard, German Soriano, François Durand, Alexander Gerbes, with cirrhosis awaiting liver transplantation. A random- Thierry Gustot, Tania M. Welzel, Francesco Salerno, Rafael Bañares, Victor Vargas, Agustin Albillos, and Vicente Arroyo participated in the INFECIR-2 ized placebo-controlled trial. J Hepatol 2018;69:1250– study. Javier Fernández, Joan Clària, AA, Ferrán Aguilar, Marco Pavesi, and 1259. Vicente Arroyo participated in data analysis and interpretation. Javier 17. Pietrangelo A, Panduro A, Chowdhury JR, et al. Albumin Fernández, Joan Clària, Alex Amoros, Ferrán Aguilar, Juan Acevedo, Luis Ruiz-del-Árbol, Cándido Villanueva, Verónica Prado, Jonel Trebicka, Paolo gene expression is down-regulated by albumin or macro- Angeli, Manuela Merli, Carlo Alessandria, Niels Kristian Aagaard, German molecule infusion in the rat. J Clin Invest 1992;89:1755–1760. Soriano, François Durand, Alexander Gerbes, Thierry Gustot, Tania M. Welzel, Francesco Salerno, Rafael Bañares, Victor Vargas, Agustin Albillos, 18. Hunter CA, Jones SA. IL-6 as a keystone cytokine in Rajiv Jalan, Mauro Bernardi, Richard Moreau, and Vicente Arroyo health and disease. Nat Immunol 2015;16:448–457. participated in the writing group. Laura Nuñez, Miriam Castro, Mireia Torres, Raquel Horrillo, and Antonio Páez from Grifols participated in the design of 19. Moreau R, Jalan R, Gines P, et al. Acute-on-chronic liver the pilot PRECIOSA study (IG0802). failure is a distinct syndrome that develops in patients with acute decompensation of cirrhosis. Gastroenter- Conflicts of interest CLINICAL LIVER These authors disclose the following: Javier Fernández has received research ology 2013;144:1426–1437. support from Grifols. Mauro Bernardi has received speaker honorarium and act 20. Bortoluzzi A, Ceolotto G, Gola E, et al. Positive cardiac as a consultant for Grifols. Laura Nuñez, Montserrat Costa, Mireia Torres, Raquel Horrillo, and Antonio Páez are full-time employees of Grifols and inotropic effect of albumin infusion in rodents with have no other competing interests to declare. The remaining authors cirrhosis and ascites: molecular mechanisms. Hepatol- disclose no conflicts. ogy 2013;57:266–276. Funding The study was supported by European Foundation for the Study of Chronic Liver Failure (EF-Clif), a nonprofit private organization aimed to stimulate research in cirrhosis. EF-Cliff has 2 main activities: the European Association Author names in bold designate shared co-first authorship. for the Study of the Liver (EASL) Chair, which is mainly devoted to clinical research through the CLIF-Consortium, and the Grifols Chair, which has Received November 19, 2018. Accepted March 14, 2019. recently been developed to promote translational research. EF-Clif receives unrestricted donations from Cellex Foundation, Grifols, and the European Reprint requests Union (coordinator center, partner, and contributor in several projects of the Address requests for reprints to: Javier Fernández, MD, PhD, EF-Clif, EASL- European Union Horizon 2020 research program). The funders had no CLIF Consortium and Grifols Chair, Travessera de Gràcia 11, 7th floor, influence on data analysis, decision to publish, or preparation of the 08021, Barcelona, Spain. e-mail: javier.fernandez@efclif.com. manuscript. Jonel Trebicka is an EF-Clif-Cellex Visiting Professor.