Gamma-glutamyl transpeptidase is a useful predictor in evaluating the prognosis in pediatric acute liver failure

Background and objective: Pediatric acute liver failure (PALF) progresses rapidly and has a poor prognosis. Therefore, simple, sensitive and specific clinical indicators are needed. Gamma-glutamyl transpeptidase (GGT) plays a role in predicting the prognosis in infantile cholestatic liver diseases. However, its role in predicting the prognosis in PALF remains unclear. Methods: In present study, children with PALF were divided into a normal GGT group and a high GGT group using the GGT level of 50 U/L as the demarcation line. Age, sex, serum total bilirubin, direct bilirubin, albumin, total bile acid, international normalized ratio (INR) and pediatric end-stage liver disease (PELD) score were compared between the 2 groups. In addition, GGT level was subjected to receiver operating characteristic (ROC) curve analysis, and the area under the curve and the optimal diagnostic cutoff value were calculated. Results: A total of 41 children with PALF were enrolled in the study. INR, PELD score and mortality rate were significantly higher in the normal GGT group in comparison to the high GGT group. GGT level had area under the ROC curve of 0.8194 (95% CI : 0.680-0.959); the optimal diagnostic cutoff values were 60 U/L. At the cutoff value, the sensitivity and specificity of GGT level in predicting the prognosis in PALF were 86.36% and 73.68% respectively. Conclusion: GGT exhibited high sensitivity and specificity in predicting the prognosis in PALF. It can be used as one useful prognostic indicator of PALF.


Background
Pediatric acute liver failure (PALF) is a critical condition. Patients with PALF may rapidly develop massive necrosis of hepatocytes, significant abnormalities in blood coagulation function, cholestasis, and a substantial reduction in serum albumin (Alb) in a short period of time. The patients may even suffer serious complications such as hepatic encephalopathy, multiple organ dysfunction and disseminated intravascular coagulation. The PALF mortality rate is rather high [1,2]. With the development and widespread application of liver transplantation technology, the PALF survival rate has reached 60-80% [3]. To distribute donor livers equitably and accurately, the pediatric end-stage liver disease (PELD) scoring system has been utilized clinically since 2002. The scoring system 3 consists of 5 indices: age, total bilirubin (TB), albumin (Alb), international normalized ratio (INR) and growth retardation. Liver transplantation is prioritized based on the PELD score [4], which has achieved satisfactory outcomes in evaluating the condition and prognosis of patients with pediatric chronic liver diseases and acute liver failure [5][6][7][8][9]. In recent years, studies have found that the PELD scoring system has some limitations under certain circumstances. For example, children under 2 years of age with biliary atresia have a higher mortality rate during the waiting period for liver transplantation. Moreover, although children with PALF complicated by portal hypertension or hepatopulmonary syndrome may have low PELD scores, they still suffer a high risk of death [10,11].
Therefore, the short-term mortality rate of children with PALF is underestimated [12,13]. Some studies suggest that bilirubin, Alb, INR, alanine aminotransferase (ALT), blood ammonia and alphafetoprotein (AFP) can be used as prognostic indicators [7,14,15]. The role of gamma-glutamyl transpeptidase (GGT) in the evaluation of the prognosis in PALF remains unclear. The present study retrospectively analyzed the clinical data of children with acute liver failure who were hospitalized and treated in our hospital between 2012 and 2018 and explored the role of GGT in predicting the prognosis in PALF.

Study subjects
The study subjects were children with PALF who were treated in a tertiary hospital between August 2010 and August 2018. All children aged 1 month to 14 years met the following criteria: (1) the child had no clear previous history of chronic liver disease; (2) biochemical markers indicated liver function damage; (3) the child had a bleeding time (due to liver injury) of ≥ 15 s or had an INR of ≥ 1.5, which was accompanied by hepatic encephalopathy, or the child had an INR of ≥ 2.0, regardless of whether she/he suffered concurrent hepatic encephalopathy; and (4) the coagulation abnormalities described above could not be corrected with vitamin K1.
After hospital admission, the children with PALF were given timely treatments, including supplementation with fat-soluble vitamins, promotion of bile acid metabolism, reduction in blood ammonia production, nutritional support, vital sign monitoring, and complication prevention. In addition, sever patients were given plasma exchange therapy if they had a serum TB level of >

Patient grouping
The children were divided into a normal GGT group and a high GGT group based on the criterion that the upper normal limit (UNL) of GGT was 50 U/L.

Prognosis 5
The children were followed up until 3 weeks after the peak INR was reached. Native liver survival was considered a good prognosis, whereas death or accepting liver transplantation was considered a poor prognosis.

Ethical approval of the study
The study protocol was performed in compliance with the Declaration of Helsinki. Informed consent was obtained from the guardian of every participant.

Statistical methods
Analysis was performed using stata10.0 statistical software. Nonnormally distributed data are expressed using the median (interquartile range). Intergroup comparisons of the measurement data was performed using the Mann-Whitney U test (nonnormal distribution), whereas the count data were analyzed using Fisher's exact probability test. The Spearman rank test was employed to examine correlations. GGT was subjected to receiver operating characteristic (ROC) curve analysis, and the areas under the ROC curves and the optimal diagnostic cutoff value were calculated. A P value of less than 0.05 indicated that the difference was statistically significant.

Results
General information A total of 41 children with acute liver failure were enrolled in the present study.

Comparison of the clinical indicators between the groups of children with different GGT levels
In the present study, the UNL of GGT, 50 U/L, was used as the demarcation line. Fourteen children had GGT levels less than 50 U/L; these children were included in the normal GGT group. The remaining 27 children constituted the high GGT group. Compared to the high GGT group, INR, PELD score and 6 mortality rate were significantly elevated in the normal GGT group

Comparison of the GGT levels in children with PALF in different prognosis groups
The children with PALF were divided into a death group and a survival group based on different prognoses. The death group had a GGT level of 36 U/L (22,73), whereas the survival group had a GGT level of 91.5 U/L (74, 170). The difference between the 2 groups was statistically significant (P=0.0115) ( Table 3). Abbreviation: GGT=Gamma-glutamyl transpeptidase

The correlation between serum GGT level and PELD score
Serum GGT level and PELD score displayed a nonbivariate normal distribution. Based on the Spearman rank correlation test, the Spearman correlation coefficient between GGT level and PELD score was -0.3908. There was a significant negative correlation between the GGT level and PELD score (P=0.0005) (Figure 1).

ROC curve analysis
ROC curve analysis was performed on the GGT levels in children with PALF. The area under the ROC curve was 0.8197 (95% CI: 0.680, 0.959) (Figure 2). The optimal diagnostic cutoff value was 60 U/L.
At this cutoff value, the sensitivity of GGT level for predicting the prognosis in PALF was 86.36%; the specificity was 73.68%.

Discussion
PALF is a severe condition with a high mortality rate. The results of the present study indicated that hereditary metabolic liver disease was the main cause of PALF. The mortality rate of the children with PALF who did not undergo liver transplantation was 46.34%. GGT could serve as one of the useful indicators of poor prognosis in children with PALF. Good sensitivity and specificity were achieved when a GGT level of 60 U/L was used as the cutoff to predict poor prognosis. In 2002, PELD scores started to be used to evaluate children with chronic liver failure who were under 12 years of age awaiting liver transplantation. At a cutoff value of 33, PELD score showed a sensitivity of 81% and a specificity of 86% in predicting a poor prognosis, the area under the curve was 0.88 [6].
Núñez-Ramos et al. found that at a cutoff value of 28, PELD score exhibited a predictive sensitivity of 72.7% and a specificity of 100% in predicting a poor prognosis [27]. Patients with high scores had an increased risk of death. In the present study, a significant negative correlation was found between the GGT level and PELD score, which may indicate that low GGT level is also efficient in predicting the prognosis in PALF.
In a variety of pediatric intrahepatic cholestasis liver diseases, low GGT levels also play a prognostic

Conclusion
We found that the poor prognosis cases had a lower level of GGT, which had a significant negative correlation with PELD score. ROC analysis indicated that the diagnostic cutoff value of GGT was 60 U/L. By the using of this cutoff value, the sensitive and specificity were 86. 36%