Liver transplantation for acute‐on‐chronic liver failure predicts post‐transplant mortality and impaired long‐term quality of life

Among patients with cirrhosis, candidate selection and timing of liver transplantation (LT) remain problematic. Acute‐on‐chronic liver failure (ACLF) is a severe complication of cirrhosis with excessive short‐term mortality rates under conservative therapeutic measures. The role of LT in the management of ACLF is uncertain.


| INTRODUC TI ON
Advanced chronic liver disease is a substantial global health burden with growing incidence in recent decades. 1 While liver transplantation (LT) has been widely established as a life-saving medical intervention for end-stage acute or chronic liver diseases in many countries, the disparity between organ demand and available donor liver grafts generates complex problems including the selection of suitable transplant candidates, patient prioritization on the waiting list and timing of surgery. 2 Acute-on-chronic liver failure (ACLF) is a frequent complication of chronic liver disease characterized by single-or multi-organ failure and dismal short-term prognosis. 3 Therapeutic measures aim at the treatment of ACLF triggers as well as medical support of failing organs, yet a high number of patients fails to stabilize under conservative treatment, placing them at high risk of death from progressive multi-organ failure. The role of LT in the management of ACLF patients is a matter of substantial debate. In the ACLF-defining CANONIC and NACSELD studies, only a small minority of patients underwent LT. 4,5 Subsequent studies yielded conflicting results regarding outcomes of ACLF patients undergoing LT, [6][7][8][9] and selection criteria for patients with regard to acceptable post-transplant outcomes are not rigorously studied. We recently reported on the positive impact of clinical stabilization of ACLF patients prior to LT on short-term post-transplant survival. 10 Our findings were generally confirmed in a large patient cohort from the UNOS database, 11 however, effects of pretransplant ACLF on aspects of post-transplant medical care that exceed sheer survival are incompletely understood. 12 Here, we assessed the effects of ACLF on long-term post-transplant survival, graft function, patient morbidity and quality of life (QoL). Our evaluation may aid in the prediction of context-dependent risks and the estimation of realistically attainable goals of LT in the setting of ACLF. Our data further call attention to a strikingly poor QoL in a large fraction of ACLF transplant recipients, necessitating the evaluation of adjunctive measures before and after LT in order to maximize the benefit obtained from each available donor graft.

| Patients
In this retrospective study, we included all adult (age ≥ 18 years at the time of surgery) patients with cirrhosis who underwent their first LT at our institution between January 1st, 2009 and December 31st, 2014. The study was approved by the Aerztekammer Hamburg (permit no. PV6061). No organs from executed prisoners were used.
Data were obtained from electronic medical records including the 3-month pretransplant period. We included the following recipient data: age, gender, age-factored Charlson comorbidity index (CCI), underlying liver disease, presence of hepatocellular carcinoma, hospitalization in an intensive care unit before and after LT, ACLF-relevant organ dysfunction and disease scores (MELD, ACLF grade, CLIFc ACLFs, CLIFc OFs). MELD score was determined via either standard MELD formula ('labMELD') or, where applicable, in coordination with Eurotransplant for MELD standard exception points ('seMELD'). 13 Donor and graft criteria included donor age and BMI, warm and cold ischemia time and the degree of macrovesicular steatosis of the graft liver. For the QoL analysis, we requested via mail all surviving patients to complete a composite survey comprised of EQ-5D-3L, 13 PHQ-4 14 and WHO-QOL-BREF 15 questionnaires.

| Diagnostic criteria
Cirrhosis was histologically verified in explanted organs. LT recipients without underlying cirrhosis and re-transplantations were excluded from the study, resulting in 250 patients eligible for analysis. Organ failures and disease scores were determined according to EASL diagnostic criteria. 4

Lay summary
Acute-on-chronic liver failure (ACLF) is a severe complication of cirrhosis with high mortality under conservative therapy. In this long-term follow-up analysis of patients with and without ACLF undergoing liver transplantation (LT), we report that ACLF predicts high post-transplant mortality, predominantly inferred by lethal infections both early and late after LT. Long-term ACLF transplant survivors have graft functions and comorbidities comparable to non-ACLF transplant recipients, but report aggravated problems in the domains of physical and psychological health.
Survival curves were compared via Mantel-Cox log-rank test. Metric variables were compared using t-test (two groups) or ANOVA (multiple groups), and dichotomous variables were compared using chi-squared analysis. Survey data were analysed according to the respective manuals [13][14][15] including the calculation of sum scores and imputation of missing values. For data management and analyses, we used SPSS 21 (SPSS, Inc, Chicago, IL), GraphPad Prism 8 (Graphpad Software, La Jolla, CA) and Microsoft Excel 2016. All P values reported are twosided, and P < .05 was considered statistically significant.

| Patient characteristics
Patient characteristics are summarized in Table 1. Of 250 cirrhosis patients undergoing LT, 98 patients (39.2%) fulfilled the EASL diagnostic ACLF criteria in the 3-month period preceding transplantation. Of those, 24 patients had ACLF grade 1, 45 patients had ACLF grade 2 and 29 patients had ACLF grade 3. Disease dynamics under conservative therapy effectuated changes in ACLF grade prior to LT as previously described. 10 Overall, non-ACLF and ACLF transplant recipients did not significantly differ in terms of age, gender, cirrhosis aetiology or comorbidity burden. Hepatocellular carcinoma (HCC) was significantly more prevalent in non-ACLF LT recipients, in whom HCC constituted the indication for LT in almost 50% of patients. At the time of LT, lab-MELD scores were significantly higher in the ACLF group compared to the non-ACLF group (31.5 points for ACLF vs 12.5 points for non-ACLF patients, P < .0001). Apart from organ functions relevant for MELD score, the prevalence of circulatory, respiratory and/or central nervous system failure was substantially higher in ACLF than in non-ACLF transplant recipients at LT. Liver grafts of transplant recipients did not differ with regard to donor age, donor BMI, graft macrovesicular steatosis or cold and warm ischemia time between groups (Table S1).

| ACLF prior to LT affects posttransplant mortality
Median follow-up duration was 8.7 (IQR 7.0-9.9) years. The presence of ACLF prior to LT impaired long-term post-transplant patient   Table S2). Of all scoring systems evaluated, seMELD was least predictive of post-transplant survival, resulting in an AUC of 0.634 ( Figure S2D). Within the groups of ACLF-and non-ACLF LT recipients, we observed significant overlap in terms of baseline patient characteristics between short-term and long-term survivors. Among non-ACLF LT recipients, patients with shorter survival tended to be older and had more comorbidities compared to long-term survivors. This association, however, was not observed among ACLF-LT recipients (Table S3).
Lethal recurrence of hepatocellular carcinoma was the leading cause of death in non-ACLF LT recipients, accounting for 15 deaths (10% of patients) ( Figure S3A). In contrast, infectious complications accounted for more than two thirds of deaths among ACLF LT recipients, and outnumbered all other causes of death both in the early and late phases after transplantation ( Figure 1B). This observation was seemingly independent of initial immunosuppression, consisting of CNI in combination with corticosteroids, to which mTOR inhibitors or mycophenolate mofetil, respectively, were added in the majority of patients from both groups prior to discharge from the hospital. More patients in the non-ACLF group, however, were changed to everolimus-containing immunosuppression after the early post-transplant phase, predominantly to prevent HCC recurrence ( Figure S3B-C).
In total, only 10/152 (6.6%) patients from the non-ACLF LT group died from infections, compared to 33 (33.7%) patients in the ACLF-LT group (P < .001) (Table S4) of pathogens were non-MDR bacteria in both groups, and we detected higher frequencies of 3-and 4-MDRGN bacteria in ACLF LT sepsis decedents despite comparable baseline MDRO prevalence in both groups; this difference, however, did not reach statistical significance because of the low number of lethal infections in the non-ACLF LT cohort ( Figure 1C).

| Comparable graft functions in ACLF-and non-ACLF LT survivors
We next assessed indicators of liver graft function in ACLF and non-ACLF patients who survived the early phase following LT. Parameters was present in significantly more ACLF LT recipients after 12 months (P = .03). The prevalence of CKD ≥ grade 3 increased to 73% in the non-ACLF LT group compared to 82% in the ACLF LT group (P = n.s.) after 5 years. End-stage renal insufficiency with haemodialysis was documented only in a small subset of cases ( Figure 2D).

| ACLF status prior to LT does not affect long-term burden of common post-transplant comorbidities
LT recipients are at considerable risk of extrahepatic morbidity and progressive loss of function of various organ systems. 20

TA B L E 2 Comorbidities 5 years after liver transplantation in non-ACLF and ACLF LT survivors
unidimensional visual analogue scale regarding their present health status (P < .05), with less than half of ACLF-LT patients reaching 70% of their 'optimal' health condition ( Figure 3B). In the second part of the EQ-5D-3L, patients scored each of the five domains on a threelevel scale, from which an index value was normalized to a VAS value set. 22 ACLF LT survivors reached a significantly lower index value (P = .01), and particularly reported aggravated problems in the domains of self-care, the ability to perform usual activities and anxiety/ depression ( Figure 3C). The latter was confirmed in the PHQ-4 score ( Figure 3D), where ACLF LT recipients reached significantly higher scores than non-ACLF LT recipients for anxiety and depression (both  Figure 3E). All results largely correlated with increasing MELD scores when stratifying patients according to labMELD at LT ( Figure S4). Importantly, we did not observe pretransplant alcohol use disorder (AUD), a condition frequently associated with psychological comorbidity 23,24 as a negative predictor of post-LT QoL or anxiety/depression (Table S5), suggesting a minor impact of the aetiology of cirrhosis on post-transplant psychological health.
Of particular note, ACLF-LT was not homogenously linked to poor post-LT health and QoL in all patients. Rather, a subset of ACLF-LT patients reached highly satisfactory scores, while the fraction of patients with unsatisfactory self-reported health and QoL was significantly larger among ACLF-LT survey responders. Specifically, 68.9% of non-ACLF LT patients reached a sum score of ≥0.75 on the EQ-5D-3L index, compared to merely 38.5% of ACLF LT responders (P < .05). We thus aimed to identify pre-LT factors associated with impaired QoL. In our analysis, age, gender, comorbidity burden and even the presence of HCC at transplantation were evenly distributed among patients in the highest and the lowest QoL tercile (Table 3). In contrast, the fraction of patients with ACLF prior to LT and the duration of ICU hospitalization following LT were significantly higher in the group with lowest QoL, indicating a significant and measurable impact of pretransplant disease severity and the immediate posttransplant course on long-term well-being of LT survivors. Acute-on-chronic liver failure has recently been characterized as a frequent and severe complication of cirrhosis with high shortterm mortality rates. 3,4 For the majority of ACLF patients, conservative medical therapy alone cannot achieve sufficient and sustained stabilization, 30 and LT has been placed at the centre of therapeutic considerations for these patients with an otherwise dismal prognosis. 31 A large fraction of ACLF patients exhibit contraindications for LT such as severe comorbidities, active alcoholism or ongoing infection. 4,5 More recent analyses, however, have indicated that in select ACLF patients, LT is feasible with comparably good results, 6,8,12 particularly in patients who experience clinical stabilization prior to surgery. 10 in ACLF patients is the unmet goal of an important ongoing scientific debate.

| D ISCUSS I ON
Here, we aimed to investigate multiple dimensions of LT outcome in a large cohort of ACLF and non-ACLF transplant recipients, and extend current knowledge on post-LT dynamics beyond sheer patient survival. We found that ACLF in the pretransplant period Importantly, we did not detect significant differences between ACLF and non-ACLF transplant survivors in terms of graft function, indicating that a large subset of patients from both groups can be successfully transplanted with satisfactory outcomes. Only a small subset of patients underwent combined kidney/liver or kidney-af-  (Table S6).
Importantly, we did not observe immediate effects of underlying liver disease aetiology on post-LT QOL. Particularly anxiety and depression frequently co-occur with AUD, 23,24 yet we did not observe differences in long-term QOL between AUD and non-AUD

ACK N OWLED G EM ENTS
Open Access funding enabled and organized by ProjektDEAL.

CO N FLI C T S O F I NTE R E S T
All authors declare they have nothing to disclose.

AUTH O R CO NTR I B UTI O N S
LG-data acquisition, analysis and interpretation, drafting of man-