Background
Acute-on-chronic liver failure (ACLF) is a syndrome that develops in patients with an acute decompensation of liver cirrhosis and is characterized by development of organ failure and high short-term mortality [
1]. The diagnostic criteria for organ failure and subsequent ACLF gradation are based on the European Foundation for the study of chronic liver failure (CLIF) organ failure score (CLIF-OF score), a modified version of the Sequential Organ Failure Assessment (SOFA) score [
1,
2]. Depending on the ACLF grade, 28-day mortality ranges from 23.3% in ACLF grade 1 to 75.5% in ACLF grade 3 [
1], and most patients require intensive care and organ support [
3,
4].
In order to prognosticate mortality in patients with ACLF more accurately, the CLIF consortium derived and validated a new score, the CLIF-C ACLF score [
2]. The CLIF-C ACLF score combines CLIF-OF score with patients’ age and white blood cell (WBC) count to generate a composite score of 0–100 in a linear range. Validation in an external prospective cohort showed that this score was significantly more accurate than Child-Pugh score, Model for End-Stage Liver Disease (MELD) score, and MELD with serum sodium score in predicting 28-day mortality in ACLF [
2]. CLIF-C ACLF score predicted short-term mortality 25% better than all listed scores [
2]. The 28-day mortality varied from below 20% in CLIF-C ACLF score < 45 to more than 80% in CLIF-C ACLF score ≥ 65 [
2].
The utility of CLIF-C ACLF score in patients with ACLF grade 3, and specifically CLIF-C ACLF score > 64, has been discussed [
5,
6] because these patients may still have a poor prognosis in spite of maximal treatment efforts and the associated high costs. Validating the CLIF-C ACLF score on the dataset of the CANONIC (EASL-CLIF Acute-on-Chronic Liver Failure in Cirrhosis) study has shown that in a subset of patients with four or more organ failures and/or CLIF-C ACLF score ≥ 65, 3–7 days after ACLF diagnosis, mortality rates were 100%. Single-center experiences in a small subset of such patients with ACLF (
n = 23) presented by Cardoso et al. [
6] supported this notion, albeit that mortality in this cohort was lower at 86% after 90 days [
5]. As a consequence, it has been suggested that intensive care support could be withdrawn in patients with this severity of disease. However, because the available data to support this notion are restricted to the CANONIC cohort and one small, single-center, study, further validation is required before this can be considered for translation into clinical practice.
In this study, we aimed to evaluate the short-term outcome of patients with ACLF and compared the predictive value of the CLIF-C ACLF score against other prognostic scores and clinical variables 48 hours after full intensive care support and regardless of when ACLF was first diagnosed. We also aimed to determine whether the CLIF-C ACLF score could be used to define the futility of ongoing intensive care unit (ICU) support.
Discussion
The data presented in this study suggest the CLIF-C ACLF score is the most accurate in predicting short-term (28-day) mortality for patients with ACLF compared with all other tested prognostic scores for chronic liver disease in patients with ACLF, especially for ACLF grade 3. We identified different thresholds of CLIF-C ACLF score to predict short-term mortality, and in order to maximize specificity around a threshold that would inform very high mortality and thereby question the benefit of ongoing ICU supportive care, further analyses were performed using a CLIF-C ACLF score cutoff ≥ 70. Applying a CLIF-C ACLF score cutoff ≥ 70 had 100% specificity for predicting mortality such that all patients above this threshold died within 28 days after ICU admission, despite maximal treatment efforts, including full organ support as per standard of care in our tertiary center. Maximal supportive treatment was provided up to the point that death was thought to be imminent. Despite the relatively limited number of patients with CLIF-C ACLF ≥ 70 (
n = 15), our data suggest that ongoing intensive care support in these patients, in the absence of liver transplant, may be futile, given no improvement despite full organ support as needed for 48 hours. Our data are in line with previous reports showing a very poor prognosis in similar cohorts of at-risk patients, in whom dynamic assessments of change in CLIF-C ACLF score showed that those with further progression of ACLF grade or failed improvement had high mortality [
5,
6]. The best management options in this scenario, given currently available limited therapies, require further evaluation, including the need for palliative care pathways.
The time point at which patients’ prognosis is assessed seems to be key. Our data suggest that mortality was relatively low (approximately 35–40%) within the first week after intensive treatment unit (ITU) admission, but beyond this, all remaining patients died quickly (within 2 weeks). This might imply that patients with CLIF-C ACLF ≥ 70 may have limited reserve and regenerative capacity, even if receiving full intervention support for the initial precipitating event. Moreover, the short survival period is an argument that either palliative care or, if eligible, liver transplant [
10‐
12] should be discussed early after assessing the response to intensive care therapy for 48 hours, because the time until death and the window for intervention is very short thereafter.
Liver transplant in so-called high-MELD patients is highly debated because it is associated with significant posttransplant morbidity [
13,
14]. Importantly, in our center, transplant selection aims at > 90% one-year survival, which necessitates the exclusion of urgently listing patients with decompensation or ACLF. However, there are data to suggest that overall survival can be in excess of 80% and comparable to patients transplanted without ACLF [
12], which is also substantiated in other studies, including studies of living donor liver recipients [
10,
15,
16]. By contrast, a retrospective study by Levesque et al. showed in a subgroup of 30 patients with ACLF grade 3 a 12-month survival rate of 43% after cadaveric liver transplant [
11]. These studies clearly highlight that although it is worthwhile discussing liver transplant in ACLF grade 3, this must be tempered by assessment of factors that may indicate worse outcome after liver transplant, such as infections, age, and presence of hepatocellular carcinoma, as proposed by Levesque et al. [
11]. In addition, patients through debilitation of their advanced liver disease and a continued severe inflammatory state, as seen with ACLF, would be expected to be frail and may not be rescued by liver transplant [
17].
To date, interventions such as extracorporeal liver support, such as the trials with MARS (molecular adsorbent recirculation system) [
18,
19], have failed to show any clear survival benefit in ACLF 3. When undertaking consideration for such interventions in such an advanced disease cohort, appropriate resource allocation and effectiveness of the intervention must remain major considerations for implementation. Until there are new interventions with proven efficacy, futility of ongoing intensive care support should be discussed early, also taking into consideration that cirrhosis and ACLF represent an increasing health and socioeconomic burden [
20]. Such early decision-making processes help facilitate an appropriate and adequate palliative care option in a cohort in whom mortality is high, despite maximal intensive treatment support.
It is important to note that a CLIF-C ACLF score ≥ 70 was associated with distinct clinical features. Notably, the SIRS score and WBC count, which are reflective of an inflammatory response, were significantly higher in those with ACLF score ≥ 70, albeit that infections as specific precipitating events were not overrepresented and patients received antibiotic treatment as part of the standard procedure. This in line with the assertion that increasing disease severity in ACLF is accompanied by a systemic inflammatory response. Claria et al. [
21] and others have shown that proinflammatory cytokines increase throughout the different severity grades of ACLF and that such inflammation is associated with higher mortality [
22‐
24]. This may imply that strategies to lower inflammation and thereby risk of new infection, such as gut decontamination, may improve outcomes, but further clinical trials of such interventions are needed. Moreover, it remains to be seen whether these strategies are cost-effective in such sick patients.
There are some limitations of this study that need consideration. First, this study is a retrospective analysis of prospectively gathered data, which may be regarded as a weakness because some potential contributory factors that might influence outcome may not have been assessed at the time of enrollment. Second, a further potential limitation of this study is that the response to supportive therapy in the ICU was evaluated at 48 hours and not beyond. The previously reported outcomes in ACLF grade 3 patients in the CANONIC study by Gustot et al. showed that assessment of CLIF-C ACLF score between days 3 and 7 and a change in score determined longer-term outcome. This supports the idea of repeated assessments to define futility in such patients, in whom a fixed time of assessment may sometimes be difficult [
5].