Serum interleukin-6 level predicts the prognosis for patients with alcohol-related acute-on-chronic liver failure

Heavy alcohol consumption is the most common etiology of acute-on-chronic liver failure (ACLF) in Japan. In some patients, ACLF is associated with a fatal outcome in less than 6 months. We evaluated the prognosis of patients with alcohol-related ACLF in our cohort and explored the prognostic factors. Forty-six patients with alcoholic liver cirrhosis who fulfilled the Japanese diagnostic criteria for ACLF, including those classified as extended and/or probable, were enrolled in this study. Serum concentrations of inflammatory cytokines (interleukin [IL]-1β, IL-6, IL-8, IL-10, IL-12p70 and TNFα) were measured. We assessed prognosis and identified factors associated with survival. During the median 33-day observation period, 19 patients died, and 3 patients underwent living donor liver transplantation. Cumulative survival rates of patients treated without liver transplantation were 69, 48, 41, and 36% at 1, 3, 6, and 12 months, respectively. Eighteen of the 19 deceased patients died within 6 months after ACLF diagnosis. Serum concentrations of inflammatory cytokines were significantly elevated, and patients who underwent liver transplantation or who died within 6 months after admission had significantly higher serum IL-6 levels than the survival group. Multivariate analysis identified IL-6 > 23.3 pg/mL at admission and model for end-stage liver disease (MELD) score ≥ 25 on day 4 of admission as significant independent factors for mortality within 6 months. Serum IL-6 level and Day-4 MELD were prognostic factors for alcohol-related ACLF. Early liver transplantation is a potential treatment option for patients whose prognosis is expected to be poor.


Introduction
Acute-on-chronic liver failure (ACLF) is a complication that can occur in patients with liver cirrhosis and is characterized by acute deterioration of liver function, organ failure, and a high risk of short-term mortality [1][2][3].Although the variety of definitions makes it difficult to predict the exact proportion of cirrhotic patients who meet the criteria for ACLF, it is estimated that 24-40% of hospitalized cirrhotic patients meet these criteria [4].There is heterogeneity with respect to the underlying causes of chronic liver disease among various international consortiums.It was reported that HBV infection is responsible for the majority of background liver disease in Chinese ACLF patients diagnosed based on APASL criteria [5].On the other hand, alcoholic cirrhosis is the major cause of ACLF in European patients [2].In Japan, most ACLF patients have a background of heavy alcohol consumption [6].
Pathophysiological mechanisms of ACLF include systemic inflammation due to infections, acute hepatic insult with drugs or alcohol, and viral hepatitis [7].Because no specific treatment is available for patients with ACLF, the management of ACLF is mainly etiology-based treatment, treatment for complications, and providing an artificial liver support system.Unfortunately, some patients fail to respond to medical interventions [8], and liver transplantation remains the ultimate therapy for patients with ACLF who face unsuccessful medical treatment [7].A 6-month abstinence from alcohol is generally required for patients with alcoholic liver cirrhosis who are being considered for liver transplantation [9].However, alcoholic hepatitis is a deadly condition in hepatology, and severe patients often develop multiple organ failure, with the 3-month mortality rate remaining at 25-40% [10].Patients whose hepatitis is not responding to medical therapy have a 6-month survival rate of approximately 30%, and most alcoholic hepatitis deaths occur within 2 months [8].Therefore, some patients with ACLF require liver transplantation before completing the 6-month abstinence from alcohol.
Herein, we examined clinical factors affecting the prognosis of patients with alcohol-related ACLF in our cohort and evaluated prognostic factors for alcohol-related ACLF, including measurement of serum cytokine levels.

Patients and study design
This was a retrospective observational study conducted at the Department of Gastroenterology at Hiroshima University Hospital between January 2014 and December 2021.Patients fulfilled the Japanese diagnostic criteria for ACLF, including extended ACLF, probable ACLF, and extended/probable ACLF [11,12] and had a history of heavy alcohol drinking (≥ 60 g/day of pure ethanol for ≥ 5 years).Patients with hepatocellular carcinoma were excluded from the analysis.All patients were offered liver transplant as the treatment option and underwent baseline investigations including hemogram, coagulation profile, renal function test, liver function test, abdominal ultrasound, upper gastrointestinal endoscopy, ascitic fluid analysis, work up for etiology of acute and chronic liver disease, and cultures for evaluation of suspected infection, as required.Standard treatment protocol was followed for all patients as per their clinical requirements.The patients were evaluated daily during hospitalization until the time of discharge.Patients were followed up after discharge for transplantation and prognosis.For the patients who did not present for follow-up, survival status was confirmed through phone contact.Compliance with ethical requirements is described on the title page.

Cytokine assay
The serum concentrations of inflammatory cytokines interleukin (IL)-1β, IL-6, IL-8, IL-10, IL-12p70 and TNFα were measured using a human inflammation kit (Cytometric Bead Array; BD Biosciences, San Jose, CA, USA) by FACS Verse™ flow cytometer according to the manufacturer's instructions.Cytokine measurements were performed using cryopreserved serum.Individual cytokine concentrations were computed using FCAP Array™ v3.0 Software (BD Biosciences).The minimum and maximum detection limits were 10 and 2500 pg/mL, respectively.Serum samples obtained from 23 healthy donors and 24 patients with alcoholic liver cirrhosis (LC) with Child-Pugh Grade A or B were also evaluated as controls.Characteristics of the patients with Child-Pugh Grade A or B are shown in supplementary table 1.

Statistical analysis
Comparisons were performed by t-test or the non-parametric Mann-Whitney U test.Overall survival was calculated from the date of the first diagnosis to the date of death, transplant surgery, or the most recent contact with the patient.Multivariate comparisons were made using the Cox proportional hazards model.The stepwise selection method was used for the selection of variables for the final regression model.Univariate and multivariate analyses were performed using EZR.The predictive accuracy of mortality was measured by estimating the area under the receiver operating characteristic (ROC) curve; ROC analysis was performed to determine the cut-off value for each item that best predicted mortality.A two-sided p value < 0.05 was considered as statistically significant.

Patient background characteristics
Forty-six alcohol-related ACLF patients were enrolled in this retrospective cohort study.The characteristics of the patients on admission are summarized in Table 1.The numbers of cases that met all of the diagnostic criteria for ACLF, extended cases, probable cases, and extended/ probable cases were 13, 4, 24, and 5, respectively.The median age of the patients was 48 (26-70) years, and 31 (67%) patients were male.The median model for end-stage liver disease (MELD) score on admission was 25 .ACLF was caused by alcohol abuse in 35 patients (76%), gastrointestinal bleeding in 7 patients (15%), and other causes in 4 patients (9%).Median total bilirubin was 9.5 (1.8-4.5)mg/dL, and median prothrombin-INR (PT-INR) was 1.90 (1.19-8.24).Treatment for disseminated intravascular coagulation (DIC) included supplemental fluids and vitamin K2 administration, and artificial liver support was performed in 7 patients.The median observation period after diagnosis of ACLF was 33 (3-2638) days.

Overall survival in alcohol-related ACLF patients
Of the 46 patients, 19 died and 3 underwent living donor liver transplantation during the observation period.The remaining 24 patients recovered following medical treatment.Cumulative survival rates in the patients with alcohol-related ACLF without liver transplantation were 69, 48, 41, and 36% at 1, 3, 6, and 12 months, respectively (Fig. 1).Causes of death for the 19 deceased patients were liver failure in 9 cases, hemorrhage in 5 cases (2 large psoas hemorrhages, one abdominal wall hemorrhage, one subdural hematoma, and one alveolar hemorrhage), pancreatitis in 2 cases, and sepsis in 3 cases, respectively.There was no significant difference in mortality between the causes of ACLF.Notably, 18 of the 19 deceased patients died within 6 months after ACLF diagnosis.Histological analysis of the liver was performed in 9 of the 19 deceased patients, revealing hepatocyte swelling, necrosis, and macrovesicular droplets consistent with the findings of alcoholic hepatitis in all patients.

Serum cytokine levels in patients with alcohol-related ACLF
Because inflammatory cytokines are thought to be associated with the development of ACLF [13][14][15], we measured serum levels of inflammatory cytokines at admission.Serum IL-1β, IL-6, IL-8, IL-10, IL-12p70 and TNFα levels were significantly higher in patients with alcohol-related ACLF compared to healthy controls or patients with alcoholic LC (Fig. 2).
To investigate whether serum cytokine levels could serve as a prognostic factor, we divided the patients into two groups, a group containing patients who survived more than 6 months after admission (survival, n = 25) and a group containing patients who underwent liver transplantation or who terminated in death within 6 months (transplant or death, n = 21) and compared serum cytokine levels.In the latter group, 9 out of 21 patients (42.9%) were fulfilled, higher than those in survival group (supplementary table 2).Although the percentage of ACLF grade 3 was 14.3% in the transplant or death group, no patient with ACLF grade 3 was included in the survival group.There was no difference between the two groups in the cause of ACLF.When serum cytokine levels at admission were compared between the two groups, the levels of IL-1β, IL-8, IL-10, IL-12p70 and TNF-α were similar, but serum IL-6 levels were significantly higher in patients in the transplant or death group compared to the survival group (105.6 [8.7-2754.8]ng/dL vs. 30.0[4.5-218.3],respectively, p = 0.04) (Fig. 3).

Factors contributing to mortality in patients with alcohol-related ACLF
We next analyzed the factors associated with mortality within 6 months for patients with alcoholic-related ACLF.In our experience, patients with chronic nutritional disorders and dehydration due to alcohol disorder often immediately improve by rehydration, nutritional management, and supplementation, and clinical markers such as serum creatinine and PT-INR improve.Therefore, blood test findings and MELD scores on day 4 as well as on admission were included in the analysis.Univariate analysis showed that the transplant or death group was significantly older, had lower cholinesterase and albumin levels on admission, and had  2).Because overlapping prognostic factors (MELD score, total bilirubin, PT-INR and serum creatinine) were identified in the univariate analysis, multivariable analysis was carried out with the inclusion of MELD score alone to prevent multicollinearity.Multivariate analysis revealed that IL-6 > 23.3 pg/mL (HR 5.70, p < 0.01), and Day-4 MELD score > 25 (HR 2.72, p = 0.04) were significant independent factors contributing to mortality within 6 months in patients with alcoholic-related ACLF (Table 3).

Discussion
Alcohol-related ACLF often results in multiple organ failure, leading to high mortality.In this study, the 28-day and 90-day mortality rates in alcohol-related ACLF patients were 33 and 52%, respectively, consistent with previous studies [2,3].No specific treatment except alcohol abstinence is available for patients with alcohol-related ACLF.Although corticosteroids, pentoxifylline and N-Acetylcysteine are reported as treatment options for ACLF, the efficacy of these treatments is limited [7].Liver transplantation is an effective treatment for ACLF patients [16] and considerably improved the prognosis of patients with severe alcoholic hepatitis who were not responding to medical therapy [17].In Japan, a 6-month period of alcohol abstinence is recommended prior to living donor liver transplantation for patients with alcoholic liver cirrhosis because alcohol abstinence might result in an improvement of liver function.However, 95% of the deceased patients with alcohol-related ACLF had died within 6 months of diagnosis in this study.In addition, a previous study also revealed high near-term (less than six months) mortality among patients with severe alcoholic hepatitis [18].Interestingly, Mathurin et al. reported that early liver transplantation could improve survival in patients with a first episode of severe alcoholic hepatitis who do not respond to medical therapies [19].In our institution, the indication for living donor liver transplantation for alcoholic cirrhosis is generally based on the 6-month rule.However, for patients with severe alcoholic hepatitis who do not respond to medical therapy, liver transplantation is indicated if there has been no episode of alcohol dependence as well as family cooperation, determination to abstain from alcohol, and approval by the hospital's Indication Evaluation Committee.Various markers have been developed to predict the prognosis of ACLF.White blood cell count is an independent factor contributing to mortality that has been incorporated into the CLIF-C ACLF score [1].The neutrophil/ lymphocyte ratio [20,21] and CRP reduction [22] are also associated with mortality in ACLF patients.A sub-analysis of the CANONIC study showed that ACLF grades from day 3 to day 7 after ACLF diagnosis were more accurate predictors of short-and medium-term prognosis than initial grades [23].A multicenter study in Japan showed age and organ failure as prognostic factors in ACLF patients [6].Consistent with this multicenter study, older patients with ACLF showed an unfavorable prognosis compared to younger patients.In our cohort, not MELD score at admission, but MELD score at day 4 of admission was identified as a prognostic factor within 6 months for alcoholrelated ACLF, suggesting that early response to intensive care could determine the prognosis of the patients because multiple pathologies, such as infection, dehydration, and nutritional disorders were related to the pathogenesis of alcohol-related ACLF.Additionally, Day-4 MELD score was useful to predict the prognosis in alcohol-related ACLF patients as well as Day-4 CLIF-C ACLF score (Supplementary Fig. 1).
In the present study, serum concentrations of inflammatory cytokines such as IL-1β, IL-8, IL-10, IL-12p70 and TNF-α were significantly elevated in alcohol-related ACLF patients.Among these inflammatory cytokines, higher IL-6 concentration was particularly associated with poor prognosis.Some studies have shown that inducing systemic inflammation is a feature of ACLF and is more prominent in alcohol-related ACLF [4,24].In a sub-analysis of the CANONIC study, elevation of serum IL-8 and IL-6 levels was associated with heavy alcohol drinking and bacterial infection in ACLF patients, respectively [13].In the liver, IL-6 is an important inducer of the acute phase response and infection defenses [25] and is crucial for hepatocyte homeostasis.It is also involved in liver regeneration as well as in metabolic functions of the liver [26].Recently, several reports have shown that serum and ascitic IL-6 levels were elevated in ACLF patients [14,15,27,28].Dynamics of serum IL-6 levels were analyzed in three patients in our cohort.Serum IL-6 levels were elevated at the time of ACLF development and decreased with recovery from the disease in all three patients (Supplementary Fig. 2).Although measurement of serum cytokines is not readily available for clinical use and serum IL-6 level was less accurate compared to Day-4 MELD score and Day-4 CLIF-C ACLF score (Supplementary Fig. 1), the combination of serum IL-6 with Day-4 MELD score enables a more accurate prediction for the prognosis of the patients with alcohol-related ACLF.
The present study has several limitations.First, this study was a retrospective analysis from a single facility and includes not only patients diagnosed as ACLF fulfilled but also extended, probable, and extended/probable cases.Further detailed analysis is needed to elucidate prognosis and predictive factors for mortality in a prospective multicenter study using a larger number of patients diagnosed as fulfilled ACLF.Second, although serum cytokine concentrations were associated with the prognosis of alcohol-related ACLF, only six cytokines were analyzed in this study.Increasing the number of cases within multiple centers and including more detailed cytokine measurements will be necessary in the future to evaluate the potential of serum IL-6 as a predictive marker for prognosis in alcohol-related ACLF patients.
In conclusion, serum IL-1β, IL-8, IL-10, IL-12p70 and TNF-α levels were elevated in patients with alcohol-related ACLF.Among these inflammatory cytokines, higher IL-6 concentration was particularly associated with poor prognosis.Serum IL-6 concentration and MELD score 4 days after admission were prognostic factors for patients with alcoholrelated ACLF.Early liver transplantation may be a treatment option for patients whose prognosis is expected to be poor.

Fig. 3
Fig. 3 Comparison of serum inflammatory cytokine levels.Patients with alcohol-related ACLF were divided into two groups by outcome; the patients who survived more than 6 months (survival, n = 25) and patients who underwent liver transplantation or who died

Table 1
Patient characteristics on admission D/T ratio direct bilirubin/total bilirubin, eGFR estimate glomerular filtration rate MELD model for end-stage liver disease, ACLF acute-on-chronic Fig. 1 Cumulative overall survival rates in 46 patients with alcoholrelated ACLF

Table 2
Comparison of clinical and laboratory data with respect to prognosis LC liver cirrhosis, D/T ratio direct bilirubin/total bilirubin, MELD model for end-stage liver disease Comparisons were performed using the Fisher exact test or the Mann-Whitney U test

Table 3
Factors contributing to mortality within 6 months