Cardiodynamic state is associated with systemic inflammation and fatal acute‐on‐chronic liver failure

Acute‐on‐chronic liver failure (ACLF) is characterized by high short‐term mortality and systemic inflammation (SI). Recently, different cardiodynamic states were shown to independently predict outcomes in cirrhosis. The relationship between cardiodynamic states, SI, and portal hypertension and their impact on ACLF development remains unclear. The aim of this study was therefore to evaluate the interplay of cardiodynamic state and SI on fatal ACLF development in cirrhosis.


| INTRODUC TI ON
In the natural history of cirrhosis, acute decompensations (AD) are frequent and often lead to development of acute-on-chronic liver failure (ACLF), 1,2 a syndrome characterized by organ failure and high short-term mortality. 2,3 The pathophysiological pathways leading to this syndrome are only partly understood but systemic inflammation (SI) seems to play a crucial role. 4 SI can occur in the setting of chronic inflammation via translocated proinflammatory signals from the intestinal lumen to the systemic circulation. [5][6][7][8] A full-blown burst of proinflammatory mediators has been found in patients with ACLF. 9 Portal hypertension is a prerequisite for AD and can be quantified by measurement of the hepatic venous pressure gradient (HVPG). 10,11 In patients with clinically significant portal hypertension (CSPH), SI measured by C-reactive protein (CRP) and hypodynamic [defined by cardiac index (CI) <3.2 L/min/m 2 ] and hyperdynamic (defined by CI >4.2 L/min/m 2 ) cardiodynamic states have been independently associated with an increased risk of first ascites development and mortality. 12 In addition, SI as assessed by IL-6 and IL-8 plays a crucial role in AD. 4,9 In outpatients with cirrhosis, arterial hypotension, as an element of cardiocirculatory dysfunction, has also been identified as an independent risk factor for ACLF development. 13 To date, neither the interplay between cardiodynamic states or markers of SI, other than CRP, nor their relation with ACLF development have been studied. 14 The aim of this study was therefore to evaluate the risk of fatal ACLF development in different cardiodynamic states and their association with SI. We hypothesize that a hypo-as well a hyperdynamic circulatory state represents a risk factor for the development of ACLF.

| Patients and data collection
In this retrospective analysis of a prospectively observed patient cohort from a single centre study, 224 outpatients with cirrhosis were inclusion and records of non-fatal ACLF were therefore not primarily recorded. 3 Blood samples were collected at the time of the enrolment and patients were followed until October 2017. Primary endpoint was fatal ACLF defined as ACLF leading to death. The ACLF definition according to the CLIF consortium was established in 2013 and the ACLF diagnosis was therefore established retrospectively. 3 Causes of death that were recorded as liver failure, sepsis and multiorgan failure were classified as fatal ACLF. Non-specified shock, cardiovascular disease, malignancy and unknown causes were classified as non-ACLF deaths.
Biochemical blood analyses were performed using standard tests.
Written informed consent was obtained from included patients; the study was approved by the local ethics committee and Data Protection Agency (J-No.2008-41-2020 and HVH-2011-02).

| Assessment of circulating levels of biomarkers
After collection, blood samples were centrifuged at 4ºC and serum samples were stored at −80ºC. Serum concentrations of interleukins (ILs) IL-6, IL-8 and the soluble IL-33 receptor (sIL-33R) were assessed. The analyses were performed with DuoSet ® Elisa kits (R&D Systems) according to the manufacturer's instructions (Table S1), at the Department of Internal Medicine I, Bonn, Germany. The ILs and sIL-33R levels were quantified in undiluted serum samples. On all ELISA plates, two patient serum samples were used as controls accessing variability within plates. CRP as a common marker of SI was not part of the routine analysis in this study.
Undetectable levels were assigned a value equal to lower limit of detection, 4,15 whereas for values higher than the upper limit of detection concentrations were extrapolated using GraphPad Prism version 5.00 (GraphPad Prism Software) according to the plotted standard curve.

| Assessment of hemodynamic parameters
All patients underwent a hemodynamic investigation in the morning after an overnight fast and at least 1-hour rest in the supine position under local analgesia. None of the patients received diuretics or beta blockers in the 24 hours preceding investigations.
Catheterization of the hepatic veins and right atrium was performed as previously described. 16

| Statistical analysis
Data are presented as median and ranges or absolute frequency and percentage, if not otherwise specified.
Mann-Whitney, Kruskal-Wallis and Chi-square tests were used for unpaired comparisons. Receiver operating characteristic (ROC) analysis was used to calculate cut-off values. Kaplan-Meier curves and log-rank tests were used to analyse rates of fatal ACLF development and all-cause mortality. Spearman's correlation was used to evaluate the relationship between inflammatory markers and hemodynamic parameters. Univariate Cox regression was used to identify predictors of ACLF. Parameters with a P value < .05 (P-in) and P-out The significance level for all tests was set at P-value < .05.
Statistical analyses were performed using SPSS V25 (IBM SPSS Statistics for Macintosh, Version 25.0: IBM Corp). Principal components analysis was performed using R (R core team, Version 3.6.0).

| General characteristics of patients
In total, 208 patients (73% male, 77% alcoholic cirrhosis and median age 60 years) were included in the final analyses (Table 1). Median Child-Pugh, MELD and CLIF-C AD scores were 7, 11 and 50 points. At inclusion, none of the patients with alcoholic cirrhosis had acute alcoholic liver injury according to NIAAA criteria (at least 3 of the following: active alcoholism of 6 months >40 g/d for female or >60 g/d for male, bilirubin >3 mg/dL, AST >50IE, AST/alanine transaminase (ALT) None of the patients underwent liver transplantation or transjugular intrahepatic portosystemic shunt placement.

| Characteristics of patients stratified by cardiodynamic state
Based on cut-offs of CI as previously applied, 12  ACLF was the cause of death in 68%, 66% and 45% of hyperdynamic, hypodynamic and normodynamic patients respectively ( Table 3). The cumulative probability of fatal ACLF in the three groups was 35%, 25% and 14% (P = .006, Figure 1A). After 1 year, the numbers were 20%, 8% and 3% (P < .01) respectively. The allcause mortality at the end of follow-up and after 1 year was also higher in the hyperdynamic group and lowest in the normodynamic group ( Figure 1B, Table 1).

Most common known trigger of fatal ACLF was infection in
20%. There were no differences between ACLF triggers within cardiodynamic groups but a trend showing higher rates of infection-triggered ACLF in patients with hyperdynamic circulation (Table S3).
In other settings, normodynamic circulation has been defined as CI of 2.5-4.2 L/min/m 2 . 17 We therefore stratified our patients  Figure S1).

| Inflammatory markers in different cardiodynamic states
In the assessment of the SI, 18 Table 2). The hyperdynamic group showed the highest rate of detection of inflammatory biomarkers, but no difference in detection rate was seen between the hypo-and normodynamic groups (  Figure 2C). Across all cardiodynamic states patients, with ascites showed higher rates of detectable SI markers compared with patients without ascites ( Figure 2D).
Looking at the concentrations of the inflammatory markers, in the hyperdynamic patients, levels of IL-8 and sIL-33R were higher compared with normodynamic patients, but the difference was not statistically significant. In hyperdynamic patients, levels of IL-8 were significantly higher compared with hypodynamic patients (Table 2).

| SI predicts development of fatal ACLF
Among the univariate analyses, predictors of fatal ACLF included age, MELD, Child-Pugh and CLIF-C AD scores, Hb, bilirubin, albumin, MAP, hyperdynamic state, SVRI, HVPG ≥16 mmHg (median), ascites, sIL-33R and IL-6 ( Table 4)  Moreover, the risk of developing of fatal ACLF seems related to both the type of cardiodynamic state and degree of SI. 12 We found a strong circuital correlation among plasma ILs, HVPG and CO. Moreover, we showed that patients with a CI >4.2 L/min/ m 2 had significantly higher rate of detectable circulating IL-6/-8 in comparison to patients with lower CI at baseline. This adds further granularity to previous findings that extreme cardiodynamic states and SI as determined by circulating CRP are independent predictors of outcomes in compensated patients and in those with ascites. 12 Indeed, CRP did not capture the differences in the degree of SI among cardiodynamic states. 12 This underlines the need for more specific markers of SI in this setting. Detectable levels of IL-6, which has been selected as an independent predictor of fatal ACLF development in our cohort, may be an appropriate candidate for such a purpose.
It has been shown that patients with decompensated cirrhosis are those with the most pronounced circulatory dysfunction (hyperdynamic/hypodynamic circulation). 30  showing high rates of undetectable levels in the healthy cohort (eg IL-6 healthy 87.5%). 4 Of note, the cirrhotic inpatients without ACLF had a much higher mean MELD of 17 compared with our outpatient cohort (median MELD 11). In our cohort of cirrhotic outpatients, we found undetectable rates of IL-6 of 55%-71% depending on cardiodynamic state. These detection rates fall right between the healthy group and cirrhotic inpatients of Clària et al suggesting an association of cytokine detection rates with progression of liver disease. The detection rates in our cohort therefore seem expected and reasonable.
Moreover, this simple and natural cut-off (detectable vs undetectable) would allow for semi-quantitative testing to identify patients at risk.
Our study has several limitations. First, no data on cardiovascular diseases and cirrhotic cardiomyopathy were available. In addition, no data on alcohol consumption in follow-up, episodes of AD and non-fatal ACLF were collected and no data of trigger and degree of ACLF were available, which is a major limitation.

| CON CLUS ION
This study shows that in patients with cirrhosis the risk of developing fatal ACLF is independently associated with the degree of SI, which is associated with cardiodynamic state. Accordingly, further stratification may help in the identification of novel therapeutic targets.

ACK N OWLED G EM ENTS
We thank Gudrun Hack for his excellent technical assistance and Sabine Dentler for critical reading.

CO N FLI C T O F I NTE R E S T
No conflict of interest exists.