EXPERIMENTAL AND CLINICAL TRANSPLANTATION

Key words : Autoimmune hepatitis, Liver transplantation, Prognosis, Wilson disease

Acute-on-chronic liver failure (ACLF) is acute liver failure (ALF) that develops in patients with preexisting chronic liver disease (CLD) or cirrhosis.¹ Chronic liver disease may already be known or may be diagnosed for the first time in patients presenting to an acute hepatic clinic.² Acute-on-chronic liver failure is a clinical situation that differs from ALF, which develops without any background of chronicity³ or decompensated cirrhosis.^4,5 The Asian Pacific Association for the Study of the Liver (APASL) describes ACLF for adults as an acute hepatic insult manifesting as jaundice (serum bilirubin ≥ 5 mg/dL) and coagulopathy (international normalized ratio [INR] ≥1.5) that is complicated within 4 weeks by clinical ascites and/or hepatic encephalopathy (HE) in a patient with previously diagnosed or undiagnosed CLD or cirrhosis.⁶ Acute-on-chronic liver failure has not yet been described in detail in children. High mortality rates (up to 50%) have been reported in adults without liver transplant (LT),^5,7 but the rate in children is unclear. Although some scoring systems^5,8-13 have claimed to reflect prognosis of ACLF in adult patients, the validity of these scoring systems in pediatric patients is not clearly known.^14,15

The aims of this study were to determine the underlying CLD of ACLF, the acute events causing ACLF, and the prognostic factors leading to LT in children.

The medical records of patients 2 to 18 years of age who met the criteria of ACLF between April 2014 and April 2020 in a tertiary hospital pediatric gastroenterology unit were analyzed retrospectively. The study was reviewed and approved by the Committee of Ethics at the Faculty of Medicine, Atatürk University, Erzurum, Turkey (B.30.2.ATA.0.01.00/176).

In our study, ACLF was similarly defined as the presence of an acute hepatic insult in previously diagnosed or undiagnosed CLD causing jaundice (total serum bilirubin ≥5 mg/dL) and coagulopathy (INR ≥2.0 or more) along with clinical and/or radiological ascites and/or HE within 4 weeks.⁶ Anthropometric measurements, present or previous primary diagnoses, symptoms on admission, physical examination, and laboratory findings and treatments of patients were recorded.

The CLD diagnoses were based on physical examination (presence of hepatomegaly, splenomegaly, ascites, or venous collaterals in the abdominal wall), laboratory tests (deterioration in liver function tests and presence of hypersplenism), radiology texts (heterogeneity of liver parenchyma on ultrasonography), esophagogastroduodenoscopy (fundal and/or esophageal varices), and liver histopathology.⁴ Causes of CLD were determined by detailed tests.^4,16-20 Patients were examined for causes of acute events, such as drug-induced liver injury and acute viral hepatitis (tested using hepatitis A, B, E, C), Epstein-Barr virus, toxoplasmosis, rubella cytomegalovirus, herpes simplex, and parvovirus B19 serological tests.

The West Haven scale²¹ in older children and the modified HE assessment scale²² were used in children under 3 years old to identify HE.

The modified APASL-ACLF Research Consortium (AARC) and Chronic Liver Failure-Sequential Organ Failure Assessment (CLIF-SOFA) scores, as previously created,²³ were calculated. The only difference of these scoring systems from the original AARC¹³ and CLIF-SOFA^9,13 was the use of creatinine reference values according to the age of the children. We calculated scores of all patients during their first admission (referred to as AARC0 and CLIF-SOFA0 scores). In addition, we calculated scores (1) just before the plasmapheresis bridge treatment in patients who developed HE, (2) day before transplant in patients without HE who underwent LT, and (3) on day 5 in the other patients (referred to as AARC1 and CLIF-SOFA1 scores). The cut-off levels of INR, total bilirubin, albumin, and ammonia for predicting LT requirement were also calculated.

Statistical analyses
For statistical analysis, version 24 of the Statistical Package for the Social Sciences (SPSS Inc., Chicago, IL, USA) was used. For categorical data, percentages were calculated. For comparison of groups, chi-square and Mann-Whitney U tests were used. The receiver operating characteristic (ROC) curves and the sensitivity, specificity, and cut-off levels of the scores and some parameters for predicting LT were determined.

Patients
Twenty-nine pediatric patients with ACLF were included in the study. Nineteen patients (65.51%) were female, and the mean age was 10.97 years.

Presentation symptoms
On admission, patients presented with jaundice in 22 cases (75.86%), abdominal pain in 20 cases (68.96%), malaise in 19 cases (65.52%), abdominal distension in 18 cases (62.06%), edema in 10 cases (34.48%), and nausea and/or vomiting in 8 cases (27.58%) (Table 1).

Underlying chronic liver disease
Patients were evaluated in terms of the cause of CLDs. The most common underlying primary CLDs were autoimmune hepatitis (AIH) in 15 cases (51.72%) (12 type 1 and 3 type 2 AIH) and Wilson disease (WD) in 8 cases (27.58%). Other underlying primary CLDs were chronic hepatitis with unknown etiology in 2 cases, homozygous alpha-1 antitrypsin deficiency in 1 case, glycogen storage disease type III in 1 case, and progressive familial intrahepatic cholestasis type 3 (PFIC3) in 2 cases (Table 1). Although 4 patients with AIH and 2 patients with Wilson disease and patients with alpha-1 antitrypsin deficiency, glycogen storage disease type III, cryptogenic cirrhosis, and PFIC3 (41.4%) had been diagnosed previously, the other 17 patients (58.6%) were diagnosed for the first time.

Liver histopathology
Hepatectomy or liver biopsy samples were evaluated histopathologically. Unlike ALF, macroscopic appearances were abnormal in hepatectomy materials from those who underwent orthotopic LT. Massive confluent necrosis and fibrosis with mild to moderate inflammation (neutrophil and eosinophil) and evidence of regeneration were observed. The remaining patients underwent liver biopsy. Liver biopsies of 2 patients with AIH and 3 patients with WD were done after improvement of INR with treatment. Rare and patchy hepatocellular necrosis was seen in 8 patients. Lymphoplasmacytic cell infiltration in AIH and micro- and macrovesicular steatosis in WD were prominent features. All patients had stage 3 to 5 fibrosis with bridging according to the classification from Ishak and associates.²⁴

Acute events
The most common acute events were AIH flare-up in 14 cases (48.28%) and WD flare-up in 8 cases (27.58%). These patients had been diagnosed for the first time. Acute events were nonadherence to treatment in 3 patients with AIH and in 2 patients with WD. In 1 patient with AIH, an acute event was due to drug-induced liver injury (due to ibuprofen) (3.44%). Other acute events were Epstein-Barr virus infection, hepatitis A virus (HAV) infection, and cytomegalovirus infection (13.80%). In 2 patients with PFIC3 and chronic hepatitis, an etiological cause of acute events was not found (6.90%) (Table 2).

Treatment of autoimmune hepatitis and Wilson disease
Patients with flare-up of AIH (14 patients) were treated with prednisolone after diagnosis, with 4 of these patients developing HE. Five of the 8 patients with flare-up of WD received rescue therapy with chelation, and the remaining 3 patients who developed HE underwent LT.

Characteristics of patients who underwent liver transplant
Seven patients underwent successful LT (24.14%) (Table 2). Three patients had deceased donors, and 4 had living donors (2 from sisters, 1 from mother, and 1 from father). At mean follow-up duration of 28.8 months (range, 4-62 months), all patients were alive. In patients who underwent LT, HE developed on days 5, 5, 6, 9, 9, 15, and 16 of follow-up. Plasmapheresis was performed in these patients. The ages of patients who did and did not undergo LT were not significantly different (12.88 vs 12.04 years old; P = .34) (Table 2).

Among those who did and did not undergo LT, alanine aminotransferase (ALT), aspartate aminotransferase (AST), total bilirubin, and albumin levels were not different. The plasma lactate levels were higher in patients who underwent LT versus those who did not (1.84 vs 1.21 mmol/L; P = .01). The INR levels of patients who underwent LT were higher than in patients who did not undergo LT (3.56 vs 2.38; P < .01). Although the ALT-to-AST ratio was <1 in all patients who underwent LT, 6 of the 22 patients who did not undergo LT also had an ALT-to-AST ratio of < 1 (P < .01) (Table 3). The positive and negative predictive values for ALT-to-AST ratio of < 1 to predict LT were 53.8% and 100%, respectively.

The cut-off levels of INR, total bilirubin, albumin, and ammonia for predicting LT requirement were calculated. The cut-off level of INR was 3.04 with 100% sensitivity and 79.6% specificity (area under the ROC curve of 0.89; 95% confidence interval [95% CI], 0.76-1.00; P < .01). The other parameters had no predictive value on outcome alone. The area under the ROC values for total bilirubin, albumin, and ammonia levels were 0.65 (95% CI, 0.39-0.91; P = .26), 0.28 (95% CI, 0.06-0.50; P = .11), and 0.62 (95% CI, 0.35-0.89; P = .35), respectively (Figure 1).

Scoring systems
The mean AARC0 scores of patients who underwent LT and patients who did not receive LT were 8.71 and 6.18 (P < .01), and the mean CLIF-SOFA0 scores were 6.80 and 6.09 (P = .028), respectively. The corresponding mean AARC1 scores were 10.57 and 5.63 (P < .01), respectively, and the corresponding mean CLIF-SOFA1 scores were 9.57 and 5.68 (P < .01), respectively (Table 4). The cut-off scores with maximum sensitivity and specificity for predicting LT requirement were 7.5, 8.5, 6.5, and 7.5 for AARC0, AARC1, CLIF-SOFA0, and CLIF-SOFA1, respectively (Table 5). The PaO₂/fraction of inspired oxygen ratio was in the range of 300 to 400 in the 5 patients who developed HE and in the range of 200 to 300 in the other 2 patients who developed HE. Circulation parameter in the CLIF-SOFA scoring system had no effect on scoring, since inotropes were not required in any patients, including those with HE.

The definition, treatment, and follow-up of ALF and CLD have been well-defined in children and adults.^3,4 After the APASL defined ACLF criteria in a guideline²⁵ and its later revisions,^6,26 new terminology has been included in the literature, and this topic has become the focus of research. Our literature search found few studies on children with ACLF.^{14,15,23,27-29} The most common underlying CLDs were reported as WD, AIH, and indeterminate.^14,15,27-29 Alam and colleagues²⁷ reported the most common acute events as WD and AIH flare-up. Jagadisan and associates¹⁴ reported that the most common acute events were hepatitis E virus and HAV infection, whereas Lal and associates15 reported the most common acute events as HAV (41.9%), undetermined (22.6%), hepatitis E virus (9.7%), and AIH flare-up (9.7%). These 3 studies were conducted in India. In a study in Italy, acute events were reported as AIH, WD, and urea cycle disorder flare-up.²⁸ In our study, the most frequent causes for an acute event were WD flare-up and AIH flare-up. All patients who underwent LT due to WD had Coombs-negative hemolytic anemia.

Adult studies have reported ACLF mortality in excess of 50%.^5,7 Jagadisan and associates¹⁴ reported 3-month mortality of 59% in their study of 17 children with ACLF. Lal and colleagues²³ reported the 28-day mortality and LT rates as 25% and 8.3%, respectively. Alam and colleagues²⁷ reported a 3-month mortality rate of 30.4% and an LT rate of 8.9%. Lal and associates¹⁵ calculated mortality rate (in 28.5 days) as 19.4% and stated the causes of death as multiorgan failure and liver failure. However, LT was not available in these centers. In the study from Italy,²⁸ 59% of patients survived without LT. The hospital where we conducted our study is also a pediatric transplant center. None of the patients with ACLF died in our study, and the LT rate was 24.14%. The high mortality rates in adult patients may be due to multiorgan failure and more frequent associated comorbid diseases.⁶ In the centers that reported pediatric ACLF data,^14,15,23,27 the high mortality rates may have been because of the inability to perform LT. Excessive hepatic reserves in children have also been suggested as a reason for lower mortality rates than in adults.³⁰ In the Pediatric Acute Liver Failure Study Group,²² rates of LT were reported as 20% to 33% in patients diagnosed with non-acetaminophen-induced ALF. In our center, LT was performed in 21% of patients who were followed up for non-acetaminophen-induced ALF during the same period (unpublished data). Our results show that ACLF has a poor prognosis that is similar to ALF but lower than in adult patients with ACLF.

It has been suggested that some prognostic scoring systems can be used to estimate ACLF outcomes in adult cases. However, although no ideal prognostic scoring system has been tested prospectively, data from retrospective studies have shown that the sensitivity and specificity of the CLIF-SOFA and AARC scores are high.^6,9,13 In their study of pediatric patients, Lal and associates²³ reported that CLIF-SOFA and AARC scores were superior for predicting poor outcome (those who underwent LT and those who died) compared with the Pediatric End-Stage Liver Disease, Child-Pugh, and the Pediatric Risk of Mortality III scores. In another study,15 maximum sensitivity and specificity were achieved at a cut-off level of 6.5 for the SOFA score to predict mortality. Alam and associates²⁷ found maximum sensitivity (100%) and specificity (64.7%) in CLIF-SOFA cut-off level of 8.5 to predict mortality. In our study, the aim was to evaluate the dynamic changes in the patients and determine the cut-off level that can predict the outcome most accurately, thus allowing a prediction of the LT requirement as early as possible, to prepare the patient for LT and to reduce mortality. For this purpose, scores were calculated at 2 different times. In our study, the second scores (AARC1 and CLIF-SOFA1) were calculated from day 5 data, since HE developed at the earliest on day 5 and clinical recovery was evident from day 5 in patients who did not undergo LT. The AARC0 score predicted LT requirement at a cut-off level of 7.5 with high sensitivity and specificity, whereas the AARC1 score predicted LT requirement with a cut-off level of 8.5 with high specificity and sensitivity. The CLIF-SOFA1 score was sensitive and specific to predict LT, but the specificity was still lower than AARC1. In CLIF-SOFA0 and CLIF-SOFA1 scoring systems, total scores were affected by total bilirubin, INR, HE, and respiratory changes in those who developed HE. None of our patients developed multiorgan failure. Therefore, no points were obtained from multiorgan failure findings (respiratory, circulatory, and renal failure) in the CLIF-SOFA scoring system. Respiratory failure was caused by changes in consciousness in HE rather than by multiorgan failure in our cases. However, parameters in AARC scoring system are total bilirubin, HE grade, INR, plasma lactate level, and serum creatinine levels.²³ To obtain 2 points from the total bilirubin level in the AARC scoring system, total bilirubin must be at least 15 mg/dL, which is a high threshold for children with ACLF. In our study, there was only 1 patient who underwent LT who had total bilirubin level ≥ 15 mg/dL. No patients had elevated serum creatinine levels. Therefore, we suggest that CLIF-SOFA and AARC scoring systems need to be modified so that they can be used in children. Nevertheless, in our study, the AARC1 and CLIF-SOFA1 scores had high LT-predictive specificity and sensitivity. However, these second scores still do not allow enough time to prepare for LT or for referral to a center where LT is performed. If existing scores are used and the patient is followed in a center where LT is not performed, it would be more accurate to act according to the cut-off level of 7.5 obtained with AARC0.

Pediatric ALF is defined as biochemical evidence of acute liver injury and coagulopathy (INR of ≥2.0 in patients with or without HE) not corrected by vitamin K without evidence of CLD. However, HE must be present if prothrombin time is 15 to 19.9 seconds.³ It is stated that, in APASL recommendations, cut-off INR for defining ACLF can be increased to 2 regardless of the presence or absence of clinical HE in pediatric patients.⁶ Therefore, in this study, we determined the cut-off of INR of ACLF as 2. Our cut-off INR of 3.04 predicted LT requirements with maximum sensitivity and specificity. Although ALT and AST levels were not predictors of LT, the negative predictive value of having ALT/AST ratio of < 1 was high (100%). Therefore, we suggest that patients with ALT/AST ratio of < 1 should be carefully monitored.

The limitation of our study is that it was retrospective and was performed with a limited number of cases in a single center. With the current ACLF definition, the presence of jaundice is required. However, some pediatric patients with liver failure do not show features of jaundice.³ Although this issue is mentioned in the APASL guideline, no new suggestions have been made.⁶

We found AARC to be the most sensitive and specific scoring system to predict LT. Although an AARC score of ≥8.5 indicated completely a LT, we suggest that, if the AARC score is ≥7.5 or INR is ≥3.04 and the patient is being followed up in a nontransplant center, it would be appropriate to refer to a LT center.

Volume: 19 Issue: 7 July 2021

FULL TEXT