Research Article

Liver transplantation in the most severely ill cirrhotic

patients: A multicenter study in acute-on-chronic liver
failure grade 3

Graphical abstract Authors

100
Florent Artru, Alexandre Louvet,
100
83.6% Isaac Ruiz, ..., Georges-Philippe Pageaux,
80 80
Philippe Mathurin, Faouzi Saliba
Survival in %
Survival in %

60 Transplanted with ACLF 3 (n = 73) 60

Non-transplanted controls (n = 119)
No ACLF (n = 292): 90% ( 95% CI: 86.5- 93.4)
40 40 ACLF 1 (n = 119 ): 82.3% (95% CI: 80.6 - 91.8) Correspondence
ACLF 2 (n = 145): 86.2% (95% CI: 75.1- 92.1)
20 20 ACLF 3 (n = 73): 83.6% (95% CI: 75 - 92)
alexandre.louvet@chru-lille.fr
7.9%
0 0
(A. Louvet)
0 50 100 150 200 250 300 350 400 0 50 100 150 200 250 300 350 400
Time (days) Time (days)
Lay summary
Liver transplantation improves survival of
Highlights patients with very severe cirrhosis. These
 Liver transplantation strongly influences survival in patients patients must be carefully monitored and
with cirrhosis and acute-on-chronic liver failure grade 3. managed in a specialized unit. The deci-
sion to transplant a patient must be quick
 However, all ACLF 3 patients develop complications after to avoid a high risk of mortality.
transplantation, especially pulmonary, renal and infectious,
and have a longer hospital stay.

 A rapid decision-making process is needed because of a high-

risk of short-term mortality, taking the ‘‘transplantation
window into account.

http://dx.doi.org/10.1016/j.jhep.2017.06.009
Ó 2017 European Association for the Study of the Liver. Published by Elsevier B.V. All rights reserved. J. Hepatol. 2017, 67, 708–715
Research Article

Liver transplantation in the most severely ill cirrhotic patients:

A multicenter study in acute-on-chronic liver failure grade 3
Florent Artru1, Alexandre Louvet1,⇑, Isaac Ruiz2, Eric Levesque2, Julien Labreuche1,
Jose Ursic-Bedoya3, Guillaume Lassailly1, Sebastien Dharancy1, Emmanuel Boleslawski1,
Gilles Lebuffe1, Eric Kipnis1, Philippe Ichai2, Audrey Coilly2, Eleonora De Martin2,
Teresa Maria Antonini2, Eric Vibert2, Samir Jaber3, Astrid Herrerro3, Didier Samuel2,
Alain Duhamel4, Georges-Philippe Pageaux3, Philippe Mathurin1, Faouzi Saliba2
1
Hôpital Claude Huriez, Services Maladies de l’Appareil Digestif and INSERM Unité 995, CHRU Lille, Lille, France; 2AP-HP Hôpital Paul-Brousse,
Centre Hépato-Biliaire; Univ Paris-Sud, UMR-S 1193, Université Paris-Saclay; Inserm, Unité 1193, Université Paris-Saclay, and DHU Hepatinov,
Villejuif, France; 3Hôpital Saint Eloi, Service d’Hépato-Gastroentérologie et Transplantation Hépatique, Montpellier, France; 4Unité de
Biostatistiques, CHRU de Lille, Lille, France

See Editorial, pages 667–668

Background & Aims: Liver transplantation (LT) for the most
severely ill patients with cirrhosis, with multiple organ dysfunc-
tion (accurately assessed by the acute-on-chronic liver failure
[ACLF] classification) remains controversial. We aimed to report
the results of LT in patients with ACLF grade 3 and to compare
these patients to non-transplanted patients with cirrhosis and
multiple organ dysfunction as well as to patients transplanted
with lower ACLF grade.
Methods: All patients with ACLF-3 transplanted in three liver
intensive care units (ICUs) were retrospectively included. Each
patient with ACLF-3 was matched to a) non-transplanted patients
hospitalized in the ICU with multiple organ dysfunction, or b)
control patients transplanted with each of the lower ACLF grades
(three groups).
Results: Seventy-three patients were included. These severely ill
patients were transplanted following management to stabilize
their condition with a median of nine days after admission
(progression of mean organ failure from 4.03 to 3.67, p = 0.009).
One-year survival of transplanted patients with ACLF-3 was
higher than that of non-transplanted controls: 83.9 vs. 7.9%,
p \0.0001. This high survival rate was not different from that
of matched control patients with no ACLF (90%), ACLF-1 (82.3%)
or ACLF-2 (86.2%). However, a higher rate of complications was
observed (100 vs. 51.2 vs. 76.5 vs. 74.3%, respectively), with a
longer hospital stay. The notion of a ‘‘transplantation window”
is discussed.
Keywords: Intensive care unit; Liver transplantation; Cirrhosis; Acute-on-chronic
liver failure; Multi-organ dysfunction.
Received 21 December 2016; received in revised form 17 May 2017; accepted 10 June
2017; available online 21 June 2017
⇑ Corresponding author. Address: Service Maladies de l’Appareil digestif, Hôpital
Huriez, Rue Polonovski, F-59037 Lille cedex, France. Tel.: +33 3 20 44 55 97; fax:
+33 3 20 44 55 64.
E-mail address: alexandre.louvet@chru-lille.fr (A. Louvet).
Conclusions: LT strongly influences the survival of patients with
cirrhosis and ACLF-3 with a 1-year survival similar to that of
patients with a lower grade of ACLF. A rapid decision-making
process is needed because of the short ‘‘transplantation window”
suggesting that patients with ACLF-3 should be rapidly referred
to a specific liver ICU.
Lay summary: Liver transplantation improves survival of
patients with very severe cirrhosis. These patients must be care-
fully monitored and managed in a specialized unit. The decision
to transplant a patient must be quick to avoid a high risk of
mortality.
Ó 2017 European Association for the Study of the Liver. Published
by Elsevier B.V. All rights reserved.
Introduction
Most countries have adopted the policy of allocating livers to
the most severely ill patients when selecting candidates for
liver transplantation (LT) with decompensated cirrhosis. This
involves providing grafts to patients with the most advanced
liver disease. Indeed, these patients are at the greatest risk of
dying on the waiting list and therefore benefit most from LT,
confirmed by a hazard ratio for mortality that decreases along
with the severity of liver failure.1 End-stage cirrhosis is associ-
ated with the failure of other organs (renal insufficiency,
encephalopathy, coagulopathy etc.). Thus, patients with multi-
ple organ dysfunction may be considered for LT, although this
could be associated with a high risk of perioperative mortal-
ity.2 In the past few years, standardization of care and better
selection of candidates for admission to intensive care units
(ICUs) have improved the short-term prognosis of patients
with end-stage liver disease.3,4 Several specific scoring systems
have been developed to predict such ‘‘futile transplantation” in
patients awaiting LT based on pre-LT variables (organ support,
hospitalization in the ICU, septic shock and the Charlson

Journal of Hepatology 2017 vol. 67 j 708–715


Comorbidity Index).5–7 These scores require external validation.
The results of LT in this population who are often considered
to be ‘‘too ill to be transplanted”, could have a negative impact
on the efficacy of the allocation system whose aim is to max-
imize the use of a limited supply of organs.8,9 Thus, clinicians
managing patients with cirrhosis and multiple organ dysfunc-
tion must evaluate the benefit of transplanting a patient with
a high risk of dying if LT is not performed, against the risk
of performing LT and allocating a graft to a patient with a high
risk of perioperative death.
Acute-on-chronic liver failure (ACLF) is a clearly defined
entity including acute deterioration of liver function in patients
with cirrhosis, often resulting in the failure of other organs and
a high short-term mortality rate.10 The number of organ fail-
ures defines four grades of ACLF. There is an increasing risk
of spontaneous mortality with each grade, which is the great-
est (e.g. 3.8% at 180 days) in patients with ACLF grade 3 (ACLF-
3).8,11 Recently, the definition of the chronic liver failure con-
sortium (CLIF-C) ACLF score, based on the CLIF-organ failure
(CLIF-OF) score system, has improved the prediction of short-
term outcome compared to other prognostic tools.12 Other
prognostic scores for the evaluation and the management of
organ failures such as the SAPS II or sepsis-related organ fail-
ure assessment (SOFA) score also help identify patients with
the highest risk of mortality during their hospital stay.13 The
aims of our study were: i) to report the results of LT in a
homogeneous population of patients with cirrhosis and ACLF-
3 hospitalized in the ICU; ii) to assess the survival benefit of
LT in these patients compared to non-transplanted controls
with cirrhosis and multiple organ dysfunction; iii) to determine
whether the post-LT outcome of patients with ACLF-3 is differ-
ent from that of patients transplanted with less severe cirrho-
sis; iv) to evaluate the usefulness of available ‘‘futility scores”
in the ACLF-3 population.
Material and methods
Patients
Between January 1st 2008 and December 31st 2014, all liver transplant recipients
in three French transplant centers (Hôpital Huriez, Lille, Hôpital Paul-Brousse,
Villejuif and Hôpital Saint-Eloi, Montpellier) were retrospectively included if they
fulfilled the following inclusion criteria: age [18 years old, LT for cirrhosis with
ACLF-3 based on the ACLF classification.11 All patients were identified through
the ‘‘CRISTAL database” of the French Agency for transplantation (Agence de la
Biomédecine).
Exclusion criteria were multiple organ transplantation (e.g. liver-kidney), LT
for fulminant hepatic failure and primary non-function of the graft. We defined
organ failures using the CLIF-OF score system, which distinguishes six types of
organ dysfunction (i.e. liver, kidney, brain, coagulation, circulatory, respiratory)
with subsequent sub-scores ranging from 1 to 3.12 Prognostic scores were calcu-
lated at admission and/or at transplantation according to the published
formula.11,12,14,15
The following variables were assessed before LT: the etiology of cirrhosis,
time of transplant listing, reason for admission to the ICU, catecholamine use,
renal replacement therapy (RRT), bacterial infection, mechanical ventilation
and its characteristics as well as the Glasgow Coma and West Haven scales
during hospitalization and at LT. Laboratory data at admission, 48–72 hours
post admission and at LT included liver tests, infection parameters (CRP, leuko-
cyte count), creatinine, ICU scores and liver-ICU scores (SOFA, CLIF-SOFA, CLIF-
C ACLF, SAPS II, model for end-stage liver disease [MELD], Child-Pugh-
Turcotte). The following variables were collected at LT: cold ischemia time
and blood transfusions. Post-LT the following elements were evaluated: length
of hospital stay, need for catecholamines, RRT and ventilation, re-intervention
and post-LT complications.
Journal of Hepatology 2017 vol. 67 j 708–715
JOURNAL OF HEPATOLOGY
Case control studies
We performed two case control studies:
a) In the first study, we assessed whether LT improved the survival of
patients admitted to the ICU for ACLF-3. Each patient with ACLF-3 trans-
planted while in the ICU was matched to one, or if possible two controls
with cirrhosis and multiple organ dysfunction who were not transplanted
while in the ICU. These control patients were taken from a prospective
cohort of patients admitted to the ICU in Hôpital Paul-Brousse, Villejuif,
France which was published elsewhere.16,17 Matching criteria were: age
± 5 years, gender, MELD score at admission ± 10 and SOFA score at admis-
sion in ICU ± 5.
b) The goal of the second study was to compare the outcome of patients with
ACLF-3 transplanted while in the ICU with that of patients with less severe
organ dysfunction based on the ACLF classification. For this analysis, each
patient transplanted with ACLF-3 was matched to control patients who
were transplanted without ACLF (each case was matched to four controls),
with ACLF-1 (each case matched to one or two controls if possible) and with
ACLF-2 (each case was matched to one or two controls if possible) with the
following matching criteria: age ± 5 years and gender. These control
patients were retrospectively included in a specific database between
2008 and 2014 including 1,611 patients classified by the ACLF score and
transplanted in the three participating centers. For this second case-
control study, the characteristics, outcome and post-LT complications of
controls were obtained from the CRISTAL database.
Statistical analysis
Quantitative variables were expressed as means (standard deviation) in case of
normal distributions or medians (interquartile range) otherwise. Categorical vari-
ables were expressed as frequencies and percentages. We first described trans-
plant patients with cirrhosis and ACLF-3 (as case analyses). Intra-case
comparisons of the patient main characteristics (all quantitative or ordinal vari-
ables) between admission to ICU and at the time of transplantation were per-
formed using paired Student t tests or Wilcoxon signed rank tests, as
appropriate. The main patient characteristics were compared across the three
participating centers using the analysis of variance or the Kruskall-Wallis test,
in case of a non-normal distribution. The overall survival curves at one year were
estimated using the Kaplan-Meier method, and were compared across the three
participating centers using the log-rank test; the one-year survival rate was cal-
culated with a 95% confidence interval (CI:). We studied the main patient charac-
teristics that were associated with one-year survival on univariate analysis using
the Cox proportional hazards regression models. The log-linearity assumption for
quantitative characteristics was assessed using Martingale residual plots and in
case of deviation, we analyzed the quantitative variable as a binary variable
according to the median cut-off value. We checked the proportional hazards
assumption for each characteristic using Schoenfeld residuals plots. Hazard ratios
(HRs) of mortality were derived from Cox regression models as effect size mea-
sure. No multivariate analysis was performed because of the small sample size.
Univariable Cox proportional hazards models were also used to assess the value
of the existing futility score in predicting one-year survival; the Harrell’s c index
of agreement was also calculated.18 Finally, to accurately evaluate the prognosis
of patients transplanted with ACLF-3, we performed two matched case-control
studies as mentioned above. All control groups were matched to the ACLF-3 case
group using the optimal matching methods without replacement, with no knowl-
edge of their survival. The 1-year survival of ACLF-3 case was compared with each
control group by Cox regression models using a robust sandwich covariance
matrix to account for the matched set.19
Comparisons of the burden of morbidity for each of the ACLF control groups
with the group of ACLF-3 cases were performed using linear mixed models for
quantitative variables and mixed logistic regression models for binary variables,
to account for the matched set. Statistical testing was done at the two-tailed a
level of 0.05. Data were analyzed using the SAS software package, release 9.4
(SAS Institute, Cary, NC).
For further details regarding the materials used, please refer to the Supple-
mentary material and the CTAT table.
Results
Between January 1st 2008 and December 31st 2014, 1,684
patients were transplanted in the three centers. Seventy-three
patients fulfilled the inclusion criteria (Fig. 1). This population
709
Research Article
Patients transplanted between
Jan 1st 2008 and Dec 31st 2014
N = 1,684
Combined transplantation, N = 115
Liver-kidney, N = 110
Liver-heart, N = 4
Liver-pancreas, N = 1
Fulminant hepatic failure, N = 92
Primary non-function, N = 46
Absence of cirrhosis, N = 128
Patients transplanted for cirrhosis
N = 1,303
No ACLF, N = 916
ACLF 1, N = 133
ACLF 2, N = 181
Patients transplanted with ACLF 3
N = 73
Fig. 1. Flowchart illustrating the number of patients identified for this study.
represented 4.3% of all liver transplant recipients during this per-
iod in the three centers. The median age was 56.8 years old
(48.6–62) and 51 (69.9%) were males. The causes of liver disease
were alcohol n = 39 (53.4%), hepatitis B virus/hepatitis C virus
n = 20 (27.4%), autoimmune/cholestatic n = 11 (15.1%) and others
n = 3 (4.1%). The diagnosis of cirrhosis was made at admission in
the ICU in 20 patients (27.4%).
Pre-LT data
Most patients had no severe liver dysfunction three months
before admission to the ICU, as shown by a median MELD score
of 12 (6–19) at that time. Only 15 (20.5%) patients were on the
waiting list for more than one month before admission.
Patients were managed in the liver ICU under the supervision
of liver intensive care specialists using homogeneous procedures
in the three centers: all patients requiring catecholamines were
treated with noradrenalin, all patients requiring RRT were treated
with continuous central veno-venous hemofiltration or hemodi-
afiltration, mechanical ventilation was indicated in case of acute
respiratory distress syndrome (ARDS) or severe encephalopathy.
Liver specific management (e.g. treatment of spontaneous bacte-
rial peritonitis, management of gastrointestinal bleeding related
to portal hypertension, treatment of hepatorenal syndrome,
etc.) was based on international guidelines.
The reasons for admission and the details of organ dysfunc-
tion are given in Table 1. The median pre-LT ICU stay was nine
days (5–14). During the ICU stay, there was a slight improvement
in the number of organ failures defined by the CLIF-OF score sys-
tem (4.03 at admission vs. 3.67 at LT, p = 0.009). This slight
decrease in organ failures was mainly related to an improvement
in respiratory failure (Table 1), whereas the other organ failures
remained stable except for renal dysfunction. Thirty-six patients
(49.3%) with ACLF-3 were admitted with septic shock and were
transplanted once infection had been controlled after a median
of seven days. Another 20 patients developed severe sepsis or
septic shock after admission and were transplanted after a med-
ian of 4.5 days. All infected patients were transplanted after sep-
sis had been controlled for at least 24 h. No patients were
receiving a noradrenalin dose of more than 3 mg/h at transplan-
tation. The median time between placement on the waiting list
and LT was eight days (3–24). Based on our analysis of all patients
710 Journal of Hepatology 2017 vol. 67 j 708–715
transplanted with ACLF-3, our aim was to define the absolute
clinical contraindications for LT in our three centers. We noted
that none of the patients were transplanted with the following
conditions: active gastrointestinal bleeding, control of sepsis for
less than 24 h, hemodynamic instability requiring dose of nora-
drenalin [3 mg/h, severe ARDS defined as a PaO2/FiO2 (P/F) ratio
\150.20
At LT, 46 (63.9%) patients were receiving mechanical
ventilation with a mean P/F ratio of 268 (78), 45 (61.6%) received
a median dose of noradrenalin of 0.5 mg/h (0–1) and 34 (47.2%)
RRT. Mean albumin was 32.3 (5.8) g/L, median international
normalized ratio 3.4 (3–4.8), median creatinine 1.1 (0.7–2.1)
mg/dl, mean bilirubin 26.3 (13) mg/dl and median platelet count
48 G/L (29.5–66). There was no difference in patient severity
assessed by the MELD, SOFA, CLIF-OF and CLIF-C ACLF scores in
the three centers at admission and at transplantation
(Table S1). We found no difference in the delay between
admission into the ICU and transplantation: 11 vs. 10 vs. 7.5
(p = 0.7). This suggests that the time to decision-making was
similar in the three centers.
Surgical procedure and post-LT data of patients transplanted with
ACLF-3
Mean cold ischemia time was 463 min (93) and recipients
required a median of eight (5–10) packed red blood cell transfu-
sions. Catecholamines were withdrawn on post-LT day two (1–6),
mechanical ventilation on day five (2–27) and RRT on day six
(3–31). All patients developed post-LT complications. The main
complications were hepatic vascular (i.e. hematoma, vascular
anastomosis complications, dissection and thrombosis) (27.4%)
and biliary tract complications (27.4%). Other complications were
pulmonary (65.8%), neurological (61.6%) and cardiovascular
(38.4%). Acquired infections were common: 80.8% of patients
developed bacterial infections, 35.6% viral infections and 15.1%
fungal infections. One year after transplantation, 11 patients
had died and survival was 83.6% (95% CI: 75–92). There was no
difference in 1-year survival in the populations in the three cen-
ters: 94.4% (95% CI: 84–100) vs. 82.3% (95% CI: 64–100) vs. 79%
(95% CI: 66–92), p = 0.33. Only age and SAPS II scores were asso-
ciated with 1-year mortality on univariate analysis among the 16
studied factors at transplantation (Table 2). Organ failure
according to the CLIF-organ failure score system, CLIF-OF and
CLIF-C ACLF at transplantation were not significantly associated
with 1-year mortality after transplantation (Table 2). The median
post-LT ICU stay was 18 days (10.0–33.5) and the median total
hospital stay was 51 days (37.0–79.8). Length of hospitalization
after transplantation was not different among the centers,
however patients in the Villejuif and Montpellier centers
spent more days in the ICU due to local practices (Table S1).
The burden of morbidity complications in patients transplanted
with ACLF-3 showed that 15% of patients had a creatinine
level [2.0 mg/dl at long-term follow-up after LT (and only two
of them required chronic RRT), 19 patients required treatment
of biliary complications, including surgery (n = 12), endoscopic
procedures (n = 5) and both endoscopic and surgical procedures
(n = 2). Twenty patients (27.4%) underwent relaparotomy
within 30 days and another 12 (16.4%) after 30 days. Five patients
(6.8%) died from infection within the first year after
transplantation.
 JOURNAL OF HEPATOLOGY
Table 1. Main characteristics of patients with ACLF 3 at admission to ICU and at liver transplantation.

At admission to ICU At transplantation p value

Primary single reason for admission in ICU, N (%) n.a. n.a.
Septic shock 36 (49.3)
Gastrointestinal bleeding 12 (16.4)
Neurological failure 4 (5.5)
Respiratory failure 7 (9.6)
Renal failure 10 (13.7)
Other cause 4 (5.5)
Coma Glasgow scale, median (IQR) 8 (4–13) 10 (5–13) 0.21
MELD score, median (IQR) 36 (30–40) 40 (32–40) 0.08
Child-Pugh, score median (IQR) 13 (12–14) 13 (13–14) 0.71
SOFA score, mean (SD) 15.8 (3.7) 15.2 (2.7) 0.15
SAPS II, mean (SD) 66.8 (18.3) 58.1 (13.0) \0.0001
CLIF SOFA, mean (SD) 17.4 (3.8) 16.5 (3.5) 0.09
CLIF-OF, mean (SD) 15.04 (1.90) 14.76 (1.59) 0.16
CLIF-C ACLF, mean (SD) 65.8 (8.7) 63.5 (7.5) 0.003
Liver OF* (1/2/3), in % 1.4/14.5/84.1 1.4/15.7/82.9 0.74
Kidney OF* (1/2/3), in % 41.4/24.3/34.3 38.3/8.8/52.9 0.05
Brain OF* (1/2/3), in % 9.9/19.7/70.4 5.6/28.2/66.2 1.00
Coagulation OF* (1/2/3), in % 2.9/13/84.1 5.7/4.3/90 0.62
Circulatory OF* (1/2/3), in % 0/38.9/61.1 0/38.4/61.6 0.82
Respiratory OF* (1/2/3), in % 19.7/36.6/43.7 37.1/47.1/15.8 \0.0001
Number of organ failuresy, mean (SD) 4.03 (1.24) 3.67 (1.06) 0.009
No analysis was made for the primary single reason. For the other tests comparison between admission to ICU and at time of transplantation were performed using paired
Student t tests or Wilcoxon signed rank tests, as appropriate.
n.a.: Not applicable; OF, organ failure.
*
Based on the CLIF-organ failure score system12.
y
Defined in the CLIF-organ failure score system12: subscore = 3 for all organ/systems except for kidney failure (subscore = 2 or 3).

Table 2. Univariate predictors of 1-year mortality from all causes.

Characteristics Hazard ratio 95% CI p value

Sex, women vs. men 1.19 0.35–3.95 0.78
Age P57 years old* 0.19 0.04–0.86 0.03
ICU length of stay at LTy 1.16 0.52–2.55 0.72
SOFA at day of LT 1.05 0.84–1.31 0.66
CLIF-SOFA at day of LT 0.97 0.81–1.15 0.73
SAPS II at day of LT 1.05 1.01–1.10 0.02
CLIF-OF at day of LT 1.23 0.81–1.87 0.32
CLIF C ACLF LT 1.00 0.93–1.07 0.99
Number of organ failures (CLIF-OF) 1.16 0.63–2.12 0.64
Liver failure on day of LT (CLIF-OF) 0.66 0.2–2.2 0.5
Kidney failure on day of LT (CLIF-OF) 1.08 0.56–2.1 0.82
Brain failure on day of LT (CLIF-OF) 1.53 0.48–4.9 0.47
à
Coagulation failure on day of LT (CLIF-OF)
Circulatory failure on day of LT (CLIF-OF) 0.87 0.28–2.75 0.82
Respiratory failure on day of LT (CLIF-OF) 1.57 0.69–3.59 0.28
MELD at LT 1.01 0.91–1.11 0.88
Univariate analysis uses the Cox proportional hazards regression models.
*
Median cut-off value.
y
Calculated after log-transformation of data.
à
Impossible to test because too few patients had subscores of 1 and 2.

Futility scores
We evaluated whether three of the available futility scores would
have been useful in our cohort. The median for Preallocation
Survival Outcomes Following Liver Transplantation (P-SOFT) score
was 30 (27–33) and would have predicted a 1-year survival \75%,5
the median for Balance of Risk (BAR) score was 19
(16–21) and would have predicted a 1-year survival of approxi-
mately 60%6 and the median University of California, Los Angeles
(UCLA) Futility Risk score was 20 (18–23) and would have predicted
a 1-year survival of 89%.7 Thus, in our cohort, the UCLA score seemed
to have the best clinical value while the P-SOFT and BAR scores
overestimated mortality by around 10 and 25%, respectively. In
addition, none of these scores was associated with 1-year mortal-
ity, with an HR (95% CI:) of 0.98 (0.86–1.12) for P-SOFT score, 0.95
(0.80–1.12) for the BAR score, and 1.06 (0.91–1.23), for the UCLA
score. Their capacity to distinguish between patients who would
survive or die was poor, with a Harrell’s c-index (95% CI:) of 0.55
(0.39–0.71) for the P-SOFT score, 0.54 (0.37–0.70) for the BAR
score, and 0.56 (0.36–0.75) for the UCLA score.

Journal of Hepatology 2017 vol. 67 j 708–715 711

Research Article
Case control study 100

To assess the impact of LT in the most severely ill patients with

80
cirrhosis in the first case-control study, we matched the 73 cases
transplanted with ACLF-3 to 119 non-transplanted controls with

Survival in %
cirrhosis and multiple organ dysfunction, hospitalized in the ICU. 60
The 1-year survival rate was much higher in transplanted
No ACLF (n = 292): 90% ( 95% CI: 86.5- 93.4)
patients than in controls: 83.6% (95% CI: 75–92) vs. 7.9% (95%
40 ACLF 1 (n = 119 ): 82.3% (95% CI: 80.6 - 91.8)
CI: 2.9–12.5) with an HR for mortality of 0.07 (95% CI: 0.03– ACLF 2 (n = 145): 86.2% (95% CI: 75.1- 92.1)
0.12, p \0.0001) (Fig. 2). ACLF 3 (n = 73): 83.6% (95% CI: 75 - 92)

To assess the impact of the grade of ACLF on outcome after LT, 20

in the second case control study we matched the 73 patients
transplanted with ACLF-3 to controls, with cirrhosis, transplanted 0
without ACLF (n = 292), with ACLF-1 (n = 119) and with ACLF-2 0 50 100 150 200 250 300 350 400
(n = 145). Like in ACLF-3 cases, we found a high 1-year survival Time (days)
No. at risk
rate in controls with cirrhosis who were transplanted: 90% (95% No ACLF 292 287 277 270 267 264 262 261
CI: 86.5–93.4) in patients with no ACLF, 82.3% (95% CI: 80.6– ACLF 1 119 110 105 104 100 99 99 97
ACLF 2 145 135 131 127 126 126 125 123
91.8) in ACLF-1 and 86.2% (95% CI: 75.1–92.1) in ACLF-2 ACLF 3 73 69 67 65 62 62 62 62
(Fig. 3). None of the transplanted control groups differed signifi-
Fig. 3. One-year survival of transplanted patients according to the ACLF
cantly from transplanted patients with ACLF-3 with an HR of 1-
grade. None of the transplanted control groups differed significantly from
year mortality of 1.75 (95% CI: 0.82–3.71, p = 0.15) for patients transplanted patients with ACLF-3 with an HR of 1-year mortality of 1.75% (95%
with no ACLF, 0.93 (95% CI: 0.44–1.90, p = 0.83) for ACLF-1 and CI: 0.82–3.71; p = 0.15) for patients with no ACLF, 0.93 (95% CI: 0.44–1.9; p = 0.83)
1.19 (95% CI: 0.61–2.31, p = 0.60) for ACLF-2. To further evaluate for ACLF-1 and 1.19 (95% CI: 0.65–2.31; p = 0.6) for ACLF-2. The 1-year survival of
the burden of morbidity in patients transplanted with ACLF-3, we ACLF-3 case was compared with each control group by Cox regression models
using a robust sandwich covariance matrix to account for the matched set.
also compared these patients to the matched controls (Table 3).
Patients transplanted with ACLF-3 had a longer duration of
post-LT hospital stay in ICUs and total hospital stay compared
Discussion
to controls. All patients (100%) with ACLF-3 developed complica-
tions within the first year, compared to 74.3%, 76.5% and 51.2% of
Results on the post-LT outcome of ICU transplanted patients are
patients with ACLF-2, -1 and no ACLF respectively. Transplant
still controversial because of the heterogeneity of transplant
patients with ACLF-3 had a particularly high rate of neurological,
patient populations in published studies. For example, one large
pulmonary, renal and infectious complications compared to other
study showed that hospitalization status (ICU vs. medical ward)
groups. The rejection rate of patients with ACLF-3 within 1-year
had a negative influence on post-LT outcome.2 Conversely, a
was similar to control patients.
recent controlled single center series reported that the 1-year
survival in patients transplanted while in the ICU was no differ-
ent from that of patients transplanted outside the ICU.21 Several
factors may play a role in these discrepancies in outcome. For
example, disease severity, cause of admission, the existence of a
100 specific liver ICU, the duration of the ICU stay, or the experience
of the center. The present case-control study confirms that LT
83.6% (95% CI: 75-92)
strongly influences the natural history of critically ill patients
80
with cirrhosis, confirmed by the significant improvement in sur-
vival in patients transplanted in the ICU for ACLF-3, compared to
Survival in %

60 Transplanted with ACLF 3 (n = 73) controls who were not. The survival of patients transplanted with
Non-transplanted controls (n = 119)
ACLF-3 was similar to patients transplanted with less severe liver
40 disease and lower grades of organ failure. However, all ACLF-3
patients developed complications, especially pulmonary, renal
20 and infectious, compared to 50–75% of patients with no ACLF,
7.9% (95% CI: 2.9 -12.5) or ACLF-1 and -2. This emphasizes the need for special manage-
0 ment when transplanting ACLF-3 patients, with repeated system-
0 50 100 150 200 250 300 350 400 atic screening for infection and careful monitoring of renal and
No. at risk Time (days)
respiratory parameters. This increased risk of complications
Transplanted 73 69 67 65 62 62 62 62
with ACLF 3 was also associated with a longer stay in the ICU and in hospital
Non-transplanted 119 19 17 14 13 9 9 7 in general. Death occurs rapidly in patients with ACLF-3 who are
controls
not transplanted.22 Thus, the selection process for these severely
Fig. 2. One-year survival of patients transplanted with ACLF-3 and of non- ill patients must be short and a complete standard evaluation (i.e.
transplanted matched controls with cirrhosis and multiple organ dysfunc- for candidates for elective LT) cannot be performed. In our study,
tion. The 1-year survival rate was much higher in the transplanted patients than the median delay between admission to the ICU and transplanta-
in controls: 83.6% (95% CI: 75–92) vs. 7.9% (95% CI: 2.9–12.5) with an HR for
mortality of 0.07 (95% CI: 0.03–0.12; p \0.0001). The 1-year survival of ACLF-3
tion was nine days and was similar in the three centers. A rapid
case was compared with the control group by Cox regression models using a evaluation explains in part this short median time to transplanta-
robust sandwich covariance matrix to account for the matched set. tion. During management in the ICU patients’ conditions

712 Journal of Hepatology 2017 vol. 67 j 708–715

 JOURNAL OF HEPATOLOGY
Table 3. Morbidity burden of transplanted patients according to ACLF grade.

ALCF-3 ACLF-2 ACLF-1 No ACLF

(n = 73) (n = 145) (n = 119) (n = 292)
MELD at LT 38 (33–40) 35 (31–40)** 29 (26–33)** 16 (10–20)**
Hospitalization status at LT, (%)
Home 0 15.9 27.7 85.2
Ward 0 43.4 53.8 12
ICU 100 40.7 18.5 2.8
Surgery and post-LT data
Cold ischemia time (minutes), mean (SD) 463 (93) 488 (119) 494 (118) 496 (143)
Transfusions PRBC (units) 8 (5–10) 7 (5–13) 7 (4–9)* 4 (2–7)**
Post-LT ICU stay (days) 18 (10–33) 10 (5–16)** 10 (5–16)** 7 (5–12)**
Post-LT hospital stay (days) 51 (37–80) 31 (24–44)** 31 (22–49)** 25 (19–34)**
Complications within 1 year, n (%) 73 (100.0) 107 (74.3)** 91 (76.5)** 149 (51.2)**
Rejection 9 (12.3) 16 (11.0) 19 (16.0) 19 (6.5)
Hepatic vascular complications 20 (27.4) 34 (23.4) 29 (24.4) 54 (18.5)
Biliary complications 20 (27.4) 35 (24.1) 36 (30.3) 53 (18.2)
Neurological complications 45 (61.6) 39 (26.9)** 41 (34.4)** 46 (15.8)**
Cardiovascular complications 28 (38.4) 34 (23.5)* 34 (28.6) 35 (12.0)**
Pulmonary complications 48 (65.8) 35 (24.1)** 35 (29.4)** 41 (14.0)**
Renal complications 57 (78.1) 55 (37.9)** 44 (37.0)** 60 (20.6)**
Bacterial complications 59 (80.8) 61 (42.1)** 53 (44.5)** 59 (20.2)**
Viral complications 26 (35.6) 24 (16.6)* 24 (20.2)* 10 (3.4)**
Fungal complications 11 (15.1) 9 (6.2)* 4 (3.4)* 5 (1.7)**
Comparisons of the burden of morbidity for each of the ACLF control groups with the group of ACLF-3 cases were performed using linear mixed models for quantitative
variables and mixed logistic regression models for binary variables, to account for the matched set.
Values are median (IQR) unless otherwise as indicated.
Morbidity patterns were compared between ACLF-3 cases and all the other groups using mixed linear or logistic regression models (after log-transformation of data for
transfusions PRBC, post-LT ICU stay and hospital length stays).
*
p for comparison with ACLF-3 cases \0.05.
**
p for comparison with ACLF-3 cases \0.0001.

improved slightly with a decrease in the number of organ failures
from 4.03 to 3.67, with particular improvements in pulmonary
condition. Organ failures were stabilized or controlled in all
patients at transplantation. Although objective criteria were not
used to define eligibility for LT, patient characteristics suggest
that subjective criteria included controlled sepsis, as well as sta-
bilization or improvement of hemodynamic status and respira-
tory parameters (i.e. absence of active gastrointestinal bleeding,
uncontrolled sepsis \24 h, hemodynamic instability requiring
[3 mg/h noradrenalin and severe ARDS defined by a P/F ratio
\150.20 Potential LT recipients were mainly identified during
the ICU stay and this short period to obtain stabilization/
improvement of organ failure and control of sepsis or gastroin-
testinal bleeding could be used to define the ‘‘transplantation
window”. Although standardization of the management of
patients with cirrhosis has improved in the ICU, there is still no
consensus on ‘‘if” and ‘‘when” these patients should be trans-
planted. We hope that the present study will provide new data
to help the selection process, which is usually center-
dependent. Most patients admitted to the ICU are not selected
for LT. For example, 17 patients were transplanted with ACLF-3
in the Paul Brousse liver ICU in six years, while nearly 98 patients
with cirrhosis are admitted to the unit each year according to
prospective studies.16,17 This suggests that 2.9% of all admitted
patients with cirrhosis are transplanted each year while in the
ICU. A prospective study is needed to evaluate the selection pro-
cess in patients with cirrhosis admitted to the ICU who are poten-
tial candidates for LT.
Although the patients were enrolled from three centers, the
study population was homogenous. Management in the ICU
was based on the number of organ failures and the presence of
a project for LT in all three centers. There was no difference in
patient severity among the centers or in 1-year survival after
transplantation. The selection process must be short, ‘‘the earlier
the better”,8 because of the risk of complications and the worsen-
ing of organ failures resulting in death and limiting the chances of
a patient being transplanted. The time from admission to ICU to
transplantation was similar in the three centers, showing that a
rapid decision-making process was applied in all cases. One of
the limitations of our study is that ideally, patients transplanted
with ALCF-3 should have been compared to non-transplanted
patients who were admitted to the ICU from all three centers.
However, Lille and Montpellier did not have control databases
so the matching process only used the Villejuif database. Another
limitation is the lack of power in the evaluation of predictive fac-
tors of mortality after transplantation, making it impossible to
perform multivariate analysis.
The good results in survival in our most severely ill patients
(ACLF-3) are in contrast with past studies showing increased
mortality in patients transplanted with severe cirrhosis.23–25
There are several reasons for this discrepancy: two studies were
performed before use of the allocation system based on the MELD
score, which enables a shorter waiting time in patients with sev-
ere decompensated cirrhosis.24,25 Furthermore, in the past few
years the prognosis of patients with severe liver decompensation
has improved in the ICU3 and several studies have been published
to help make clinical decisions.4,26,27 During the same period,
post-transplantation management has also significantly
improved (e.g. in terms of immunosuppressive regimens, the
treatment of infection and mechanical ventilation, etc.).28 Results
of a third study23 suggest that the ‘‘transplantation window”
plays an important role. Indeed, during the stay in the ICU in that
study, the SOFA score worsened from 13 to 19, suggesting that
the patient’s clinical condition was not stabilized before

Journal of Hepatology 2017 vol. 67 j 708–715 713

Research Article
transplantation. No worsening of the SOFA score was observed in
our study (SOFA score at admission to the ICU: 15.8 vs. 15.2 at
transplantation). However, the length of stay in ICU in the study
by Umgelter was 14 days,23 compared to nine in our patients.
Data on the length of stay in the ICU are not available in the Saab
and Jacob studies. Again, these results suggest that liver trans-
plantation should be discussed early in patients with ACLF-3 in
the ICU and patients should be transplanted as soon as their clin-
ical status is stabilized.
The notion of ‘‘futile LT” has been discussed in the last dec-
ade and defined as survival \3 months or in-hospital post-LT
mortality. In a context of worldwide organ scarcities, numerous
studies have tried to identify the predictive factors of early
post-LT mortality to avoid futile LT.5–7 Some of these predictive
factors are related to a population of patients with cirrhosis
hospitalized in the ICU (i.e. MELD score, hospitalization in
ICU, and life support pre-transplant) as well as associated
comorbidities (e.g. ‘‘cardiac risk”7 and Charlson Comorbidity
Index [CCI]). Among the three scores tested in our study, the
BAR and P-SOFT seemed to overestimate the risk of death com-
pared to survival in the original papers. Based on these two
scores, many patients would not have been considered for LT
(‘‘futile LT”). Conversely, the UCLA score, based on comorbidity
scores (cardiac risk and CCI) seemed to be helpful in selecting
candidates for LT because the predicted survival rate was close
to that observed in the present study. Nevertheless, the global
performance of the three scores based on the Harrell’s c-index
was poor, and because none of these scores was associated
with survival, we do not feel that they are clinically useful
in our population. One possible explanation could be that they
do not take the ‘‘transplantation window” into account.
In conclusion, LT can improve the poor prognosis of the most
severely ill patients with cirrhosis (ACLF-3), enabling them to
achieve survival rates similar to transplanted patients with a
lower ACLF grade. These good survival results are associated with
an increased post-LT hospital stay and high rate of complications.
However, to obtain these good results a short decision-making
process is needed because of the short ‘‘transplantation window”
suggesting that patients with ACLF-3 must be rapidly referred to
a specific liver ICU.
Financial support
The authors declare that no financial support was provided for
this manuscript.
Conflict of interest
The authors who have taken part in this study declared that they
do not have anything to disclose regarding funding or conflict of
interest with respect to this manuscript.
Please refer to the accompanying ICMJE disclosure forms for
further details.
Authors’ contributions
Design of the study: FA, AL, FS. Acquisition of data: FA, AL, IR, EL,
JL, JUB, GL, SD, EB, GL, EK, PI, AC, EDM, TMA, EV, SJ, AH, DS, GPP,
714 Journal of Hepatology 2017 vol. 67 j 708–715
PM, FS. Statistical analysis: FA, AL, JL, AD. Drafting of the manu-
script: FA, AL, FS.
Acknowledgments
We particularly thank Dr Anne Bignon (Lille), Dr Guillaume Millet
(Lille), Dr Marc Boudon (Villejuif), Dr Stephanie Faure (Montpel-
lier) and Dr Mickael Bismuth (Montpellier) for their involvement
in patients’ daily management.
Supplementary data
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.jhep.2017.06.
009.
References
[1] Merion RM, Schaubel DE, Dykstra DM, Freeman RB, Port FK, Wolfe RA. The
survival benefit of liver transplantation. Am J Transplant 2005;5:307–313.
[2] Bittermann T, Makar G, Goldberg DS. Early post-transplant survival:
Interaction of MELD score and hospitalization status. J Hepatol
2015;63:601–608.
[3] McPhail MJW, Shawcross DL, Abeles RD, Chang A, Patel V, Lee G-H, et al.
Increased survival for patients with cirrhosis and organ failure in liver
intensive care and validation of the chronic liver failure-sequential organ
failure scoring system. Clin Gastroenterol Hepatol 2015;13:1353–1360, e8.
[4] Nadim MK, Durand F, Kellum JA, Levitsky J, O’Leary JG, Karvellas CJ, et al.
Management of the critically ill patient with cirrhosis: A multidisciplinary
perspective. J Hepatol 2016;64:717–735.
[5] Rana A, Hardy MA, Halazun KJ, Woodland DC, Ratner LE, Samstein B, et al.
Survival outcomes following liver transplantation (SOFT) score: a novel
method to predict patient survival following liver transplantation. Am J
Transplant 2008;8:2537–2546.
[6] Dutkowski P, Oberkofler CE, Slankamenac K, Puhan MA, Schadde E,
Müllhaupt B, et al. Are there better guidelines for allocation in liver
transplantation? A novel score targeting justice and utility in the model for
end-stage liver disease era. Ann Surg 2011;254:745–753, [Discussion 753].
[7] Petrowsky H, Rana A, Kaldas FM, Sharma A, Hong JC, Agopian VG, et al. Liver
transplantation in highest acuity recipients: identifying factors to avoid
futility. Ann Surg 2014;259:1186–1194.
[8] Putignano A, Gustot T. New concepts in acute-on-chronic liver failure:
implications for liver transplantation. Liver Transplant 2016. http://dx.doi.
org/10.1002/Lt.24654.
[9] Karvellas CJ, Lescot T, Goldberg P, Sharpe MD, Ronco JJ, Renner EL, et al. Liver
transplantation in the critically ill: a multicenter Canadian retrospective
cohort study. Crit Care Lond Engl 2013;17:R28.
[10] Jalan R, Gines P, Olson JC, Mookerjee RP, Moreau R, Garcia-Tsao G, et al.
Acute-on chronic liver failure. J Hepatol 2012;57:1336–1348.
[11] Moreau R, Jalan R, Gines P, Pavesi M, Angeli P, Cordoba J, et al. Acute-on-
chronic liver failure is a distinct syndrome that develops in patients with
acute decompensation of cirrhosis. Gastroenterology 2013;144:1426–1437,
e1–e9.
[12] Jalan R, Saliba F, Pavesi M, Amoros A, Moreau R, Ginès P, et al. Development
and validation of a prognostic score to predict mortality in patients with
acute-on-chronic liver failure. J Hepatol 2014;61:1038–1047.
[13] Saliba F, Ichai P, Levesque E, Samuel D. Cirrhotic patients in the ICU:
prognostic markers and outcome. Curr Opin Crit Care 2013;19:154–160.
[14] Vincent JL, Moreno R, Takala J, Willatts S, De Mendonça A, Bruining H, et al.
The SOFA (Sepsis-related Organ Failure Assessment) score to describe organ
dysfunction/failure. On behalf of the Working Group on Sepsis-Related
Problems of the European Society of Intensive Care Medicine. Intensive Care
Med 1996;22:707–710.
[15] Le Gall JR, Lemeshow S, Saulnier F. A new simplified acute physiology score
(SAPS II) based on a European/North American multicenter study. JAMA
1993;270:2957–2963.

[16] Levesque E, Hoti E, Azoulay D, Ichaï P, Habouchi H, Castaing D, et al.
Prospective evaluation of the prognostic scores for cirrhotic patients
admitted to an intensive care unit. J Hepatol 2012;56:95–102.
[17] Levesque E, Saliba F, Ichaï P, Samuel D. Outcome of patients with cirrhosis
requiring mechanical ventilation in ICU. J Hepatol 2014;60:570–578.
[18] Harrell FE, Lee KL, Mark DB. Multivariable prognostic models: issues in
developing models, evaluating assumptions and adequacy, and measuring
and reducing errors. Stat Med 1996;15:361–387.
[19] Lee EW, Wei LJ, Amato DA, Leurgans S. Cox-type regression analysis for large
numbers of small groups of correlated failure time observations. In: Goel PK,
Klein JP, editors. Surviv. Anal. State Art. Netherlands: Springer; 1992. p.
237–247.
[20] Guérin C, Reignier J, Richard J-C, Beuret P, Gacouin A, Boulain T, et al. Prone
positioning in severe acute respiratory distress syndrome. N Engl J Med
2013;368:2159–2168.
[21] Knaak J, McVey M, Bazerbachi F, Goldaracena N, Spetzler V, Selzner N, et al.
Liver transplantation in patients with end-stage liver disease requiring
intensive care unit admission and intubation. Liver Transplant
2015;21:761–767.
Journal of Hepatology 2017 vol. 67 j 708–715
JOURNAL OF HEPATOLOGY
[22] Gustot T, Fernandez J, Garcia E, Morando F, Caraceni P, Alessandria C, et al.
Clinical Course of acute-on-chronic liver failure syndrome and effects on
prognosis. Hepatology 2015;62:243–252.
[23] Umgelter A, Lange K, Kornberg A, Büchler P, Friess H, Schmid RM. Orthotopic
liver transplantation in critically ill cirrhotic patients with multi-organ
failure: a single-center experience. Transplant Proc 2011;43:3762–3768.
[24] Saab S, Wang V, Ibrahim AB, Durazo F, Han S, Farmer DG, et al. MELD score
predicts 1-year patient survival post-orthotopic liver transplantation. Liver
Transplant 2003;9:473–476.
[25] Jacob M, Copley LP, Lewsey JD, Gimson A, Toogood GJ, Rela M, et al.
Pretransplant MELD score and post liver transplantation survival in the UK
and Ireland. Liver Transplant 2004;10:903–907.
[26] Cárdenas A, Ginès P. Management of patients with cirrhosis awaiting liver
transplantation. Gut 2011;60:412–421.
[27] Ginès P, Fernández J, Durand F, Saliba F. Management of critically-ill
cirrhotic patients. J Hepatol 2012;56:S13–S24.
[28] Razonable RR, Findlay JY, O’Riordan A, Burroughs SG, Ghobrial RM, Agarwal
B, et al. Critical care issues in patients after liver transplantation. Liver
Transplant 2011;17:511–527.
715