Validation of prognostic scores for predicting acute liver failure and in-hospital death in patients with dengue-induced severe hepatitis

Abstract

BACKGROUND

Acute liver failure (ALF) in dengue is rare but fatal. Early identification of patients who are at risk of ALF is the key strategy to improve survival.

AIM

To validate prognostic scores for predicting ALF and in-hospital mortality in dengue-induced severe hepatitis (DISH).

METHODS

We retrospectively reviewed 2532 dengue patients over a period of 16 years (2007-2022). Patients with DISH, defined as transaminases > 10 times the normal reference level and DISH with subsequent ALF, were included. Univariate regression analysis was used to identify factors associated with outcomes. Youden’s index in conjunction with receiver operating characteristic (ROC) analysis was used to determine optimal cut-off values for prognostic scores in predicting ALF and in-hospital death. Area under the ROC (AUROC) curve values were compared using paired data nonparametric ROC curve estimation.

RESULTS

Of 193 DISH patients, 20 developed ALF (0.79%), with a mortality rate of 60.0%. International normalized ratio, bilirubin, albumin, and creatinine were independent predictors associated with ALF and death. Prognostic scores showed excellent performance: Model for end-stage liver disease (MELD) score ≥ 15 predicted ALF (AUROC 0.917, sensitivity 90.0%, specificity 88.4%) and ≥ 18 predicted death (AUROC 0.823, sensitivity 86.9%, specificity 89.1%); easy albumin-bilirubin (ALBI) score ≥ -30 predicted ALF and death (ALF: AUROC 0.835, sensitivity80.0%, specificity 72.2%; death: AUROC 0.808, sensitivity 76.9%, specificity 69.3%); ALBI score ≥ -2 predicted ALF and death (ALF: AUROC 0.806, sensitivity 80.0%, specificity 77.4%; death: AUROC 0.799, sensitivity 76.9%, specificity 74.3%). Platelet-ALBI score also showed good performance in predicting ALF and death (AUROC = 0.786 and 0.699, respectively). MELD and EZ-ALBI scores had similar performance in predicting ALF (Z = 1.688, P = 0.091) and death (Z = 0.322, P = 0.747).

CONCLUSION

MELD score is the best predictor of ALF and death in DISH patients. EZ-ALBI score, a simpler yet effective score, shows promise as an alternative prognostic tool in dengue patients.

Key Words: Fulminant; Clinical outcomes; Liver injury; Prognostic assessment; Predictive models; Mortality prediction

Core Tip: Although acute liver failure (ALF) from dengue infection is rare, its consequences are severe, often resulting in high mortality. Timely identification of at-risk individuals and prompt intervention are crucial for optimizing clinical outcomes. Parameters such as international normalized ratio, bilirubin, albumin, and creatinine were identified to be independent predictors associated with ALF and mortality. Incorporating these parameters into prognostic models, the model for end-stage liver disease score was the most reliable predictor of ALF and mortality in dengue-infected patients. Additionally, the easy albumin-bilirubin score, a simpler metric, exhibited excellent predictive performance, indicating its potential as an alternative prognostic tool in this patient population.

INTRODUCTION

Dengue infection is a significant mosquito-borne disease with a global impact, particularly in the Americas, Southeast Asia, and the Western Pacific. Dengue infection can be classified into two severity levels: Dengue fever (DF), characterized by mild symptoms like fever, anorexia, myalgia, and headache; and severe dengue, known as dengue hemorrhagic fever (DHF). Severe dengue is marked by capillary leakage syndrome (CLS), leading to plasma leakage in the body’s third spaces. In the more severe form of CLS, known as dengue shock syndrome (DSS), there is profound plasma leakage resulting in inadequate tissue perfusion and organ failure. Common organ failures in DSS include the liver, kidney, lungs, heart, and brain[1,2].

Liver injury from dengue infection is common, typically mild and self-limited. The incidence of dengue-induced severe hepatitis (DISH), defined by a transaminase elevation exceeding ten times the upper normal reference level, ranges from 4% to 15%. While DISH rarely progresses to acute liver failure (ALF)[3], with the incidence ranging from 0.7%[4] to 2.0%[5], its prognosis is notably poor, with the mortality rates of 47.0%[5] to 58.8%[4]. Current treatment approaches are predominantly reactive, with liver failure assessments typically occurring only after symptoms like jaundice or confusion emerge. This delay can hinder timely intervention. Predicting high-risk patients early could enable earlier treatment and improve outcomes. Despite the rarity of ALF, its prognosis underscores the need for prompt recognition and intervention. Although liver-specific prognostic scores have shown promising performance in predicting ALF across various conditions, the predictive scores' utility in dengue-induced ALF remains underexplored. Hence, the present study aimed to evaluate the effectiveness of liver-specific prognostic scores in predicting ALF and death in patients with DISH.

MATERIALS AND METHODS

Patients

We retrospectively reviewed 2532 dengue patients from 2007 to 2022. The inclusion criteria were as follows: (1) Age ≥ 10 years; (2) Dengue infection and classification were identified according to the World Health Organization (WHO) criteria[6] with serological confirmation by the detection of dengue immunoglobulin M antibody or nonstructural 1 antigen; and (3) Severe hepatitis, diagnosed based on serum glutamate-oxaloacetate transaminase (SGOT) > 350 U/L or serum glutamate-pyruvate transaminase (SGPT) > 400 U/L. Liver function tests were needed to be collected within 48 h after hospitalization. The exclusion criteria were as follows: (1) Blood test was not performed within 7 d after the onset of fever; (2) Incomplete clinical information; and (3) Having other concomitant cause of severe liver injury, such as the use of acetaminophen with a dose over 150 mg/kg within the recent 96 h; a prior consumption of ≥ 40 (female) or ≥ 60 (male) g of alcohol/d within the recent 60 d (Figure 1).

Figure 1 Study algorithm. DISH: Dengue-induced severe hepatitis; ALF: Acute liver failure.

Study procedure

Baseline demographic data, clinical presentation of dengue infection, hospital course, and treatment outcome at 30 d were reviewed from the electronic medical records. Clinical presentation and diagnostic criteria for dengue infection and severities were categorized based on the WHO guidelines issued in 2009. Biological samples were collected within 48 h after admission. During this 48-h window, parameters used in the liver-related prognostic scores, such as bilirubin, albumin, and international normalized ratio (INR), have a small variation in dengue cases. As a result, these factors are unlikely to significantly impact the accuracy of the predictive scores in our study. Blood samples were immediately processed at the central laboratory because this is part of the standard medical practices in our hospital. DISH patients who developed hepatic encephalopathy and had an INR level ≥ 1.5 within 26 wk after the onset of fever were diagnosed with ALF according to the American Association for the Study of Liver Diseases 2011 criteria[7]. Factors that might contribute to ALF and liver-specific treatment outcomes were evaluated. Regarding liver-specific prognostic scores, we chose scores that had been studied in the ALF and acute-on-chronic liver failure (ACLF), including the model for end-stage liver disease (MELD) score, albumin-bilirubin (ALBI) score, easy (EZ)-ALBI score, and platelet-ALBI (PALBI) score[8] (Table 1). These liver-specific prognostic scores have been evaluated in various liver injury situations, such as acute viral hepatitis and hepatocellular carcinoma (HCC) undergoing intervention[8]. Laboratory parameters were calculated based on the initial blood test upon admission. The primary outcome of the study was the diagnostic performance of each prognostic score in predicting ALF and in-hospital death in patients with DISH. Secondary outcomes were predictive value of clinical parameters and laboratory parameters in detecting ALF and death from DISH.

Table 1 Liver-specific prognostic score formulas.

Score	Formula
MELD	3.78 × loge total bilirubin (mg/dL) + 11.2 × loge INR + 9.57 × loge creatinine (mg/dL) + 6.43
ALBI	0.66 × log10 total bilirubin (µmol/L) + [0.0085 × albumin (g/L)]
PALBI	2.02 × log10 total bilirubin (µmol/L) - 0.37 × [log10 total bilirubin (µmol/L)]2 - 3.48 × log10 platelets (1000/µL) + 1.01 × [log10 platelets (1000/µL)]2
EZ-ALBI	Total bilirubin (mg/dL) + [9 × albumin (g/dL)]

MELD: Model for end-stage liver disease; INR: International normalized ratio; ALBI: Albumin-bilirubin; PALBI: Platelet-albumin-bilirubin; EZ-ALBI: Easy albumin-bilirubin.

Statistical analysis

Categorical variables are described as counts and percentages and were compared using the χ² test or Fisher's exact test. Continuous variables are presented as the mean. We evaluated the normality of the data distribution using skewness. For comparisons between groups, a two-tailed independent sample t-test was used for continuous variables with a normal distribution, and the Mann-Whitney (Wilcoxon rank) test was used for those with a nonnormal distribution. The relationship between each clinical parameter and ALF or mortality was initially assessed by logistic regression analysis. Significant parameters and scores were subsequently included in multivariate Cox regression analysis (forward-stepwise model). The odds ratio (OR), calculated by logistic regression analysis and the Mantel–Haenszel method, was used to estimate the degree of each prognostic factor. Youden’s index in conjunction with receiver operating characteristic (ROC) analysis was used to determine the best cut-off value of prognostic scores in predicting ALF and in-hospital death, and area under the ROC curve (AUROC) values were compared using paired data nonparametric ROC curve estimation. Regarding missing data management, we used a complete case analysis strategy. All tests were two-sided, and the adopted P value for the significant level was < 0.05. All statistics were analyzed with the SPSS for Mac (version 26.0; SPSS Inc., Chicago, IL, United States). The statistical methods of this study were reviewed and approved by Chonlada Phathong, a biomedical statistician from the Division of Gastroenterology, Department of Medicine, Faculty of Medicine, Chulalongkorn University.

RESULTS

A total of 2532 patients were hospitalized due to dengue infection during the 16-year study period. Eighty-five patients were excluded. Of 2447 eligible dengue patients, 198 (8.10%; female 50.5% and male 49.5%) had DISH and were enrolled in the study with an average age of 28.8 ± 15.2 years. There were 126 (63.6%) patients who had CLS; 95 (75.4%) patients had DHF and 31 (24.6%) had DHF with DSS.

Of the 198 DISH patients, 174 (87.9%) could maintain their liver function whereas 24 (12.1%) deteriorated and developed ALF. Age, gender, and liver co-morbidities did not relate to ALF. Patients with DSS had a higher risk of ALF than those without, with an OR of 34.62 (95% confidence interval [CI]: 11.71-102.33, P < 0.001); however, among those without DSS, there was no difference in prognosis between DF and DHF without DSS patients (P = 0.360). The overall mortality rate was 7.1% (14 cases), of which 92.9% (13 cases) had DISH with ALF and 7.1% (1 case) did not have ALF. Mortality rate from ALF was 54.2% vs 0.6% in non-ALF patients.

Based on clinical parameters and the development of ALF, there was no difference in hemoconcentration, atypical lymphocyte count, or platelet count between the two groups. ALF patients had a higher degree of inflammation and end-organ damage than non-ALF patients, as demonstrated by higher mean levels of white blood cells, meutrophils, INR, creatinine (Cr), total bilirubin (TB), SGOT, and SGPT (Table 2 and Supplementary Table 1).

Table 2 Parameters at initial presentation.

Parameter	All DISH patients (n = 198)	DISH without ALF (n = 174)	DISH with ALF (n = 24)	P value
Laboratory profiles
WBC count (103/L)	5.52 ± 3.85	5.24 ± 3.63	7.48 ± 4.78	0.007
Neu (103/L)	3.20 ± 2.78	2.99 ± 2.73	4.53 ± 2.86	0.021
INR	1.38 ± 0.64	1.16 ± 0.25	2.42 ± 0.85	< 0.001
Cr (mg/dL)	1.04 ± 1.01	0.91 ± 0.89	1.93 ± 1.30	< 0.001
Albumin (g/dL)	3.60 ± 0.54	3.69 ± 0.49	3.09 ± 0.55	< 0.001
TB (mg/dL)	1.95 ± 4.34	1.28 ± 1.54	6.34 ± 10.44	< 0.001
SGOT (U/L)	1515.86 ± 2258.05	1031.51 ± 1195.21	5007.25 ± 4275.91	< 0.001
SGPT (U/L)	683.64 ± 749.00	550.41 ± 458.53	1649.54 ± 1454.57	< 0.001
Liver prognostic scores
MELD score	12.20 ± 8.01	9.52 ± 4.46	24.71 ± 9.05	< 0.001
EZ-ALBI score	-30.27 ± 7.48	-31.81 ± 4.99	-20.95 ± 12.17	< 0.001
ALBI score	-2.17 ± 0.68	-2.29 ± 0.62	-1.48 ± 0.62	< 0.001
PALBI score	-1.06 ± 0.53	-1.15 ± 0.49	-0.61 ± 0.47	< 0.001

DISH: Dengue-induced severe hepatitis; ALF: Acute liver failure; WBC: White blood cell; Neu: Neutrophils; INR: International normalized ratio; Cr: Creatinine; TB: Total bilirubin; SGOT: Serum glutamate-oxaloacetate transaminase; SGPT: Serum glutamate-pyruvate transaminase; MELD: Model for end-stage liver disease; ALBI: Albumin-bilirubin; EZ-ALBI: Easy albumin-bilirubin; PALBI: Platelet-albumin-bilirubin.

Results from the univariate analysis showed that INR, TB, albumin, and Cr were independent laboratory factors associated with ALF, whereas age, gender, and liver co-morbidity were not associated with in-hospital mortality (Table 3 and Supplementary Table 2). The presence of DSS increased mortality with an OR of 28.05 (95%CI: 7.21-109.18, P < 0.001). High INR and low albumin were laboratory profiles associated with death from DISH with an OR of 5.83 (95%CI: 2.59-13.12, P < 0.001) and 0.15 (95%CI: 0.05-0.44, P < 0.001), respectively.

Table 3 Univariate regression analysis of factors associated with severe adverse outcomes from dengue-induced severe hepatitis.

Parameter	Acute liver failure			In-hospital mortality
	OR	95%CI	P value	OR	95%CI	P value
Clinical parameter
Presence of DSS	34.62	11.71-102.33	< 0.001	28.05	7.21-109.18	< 0.001
Laboratory parameters
INR	72.27	10.98-475.89	< 0.001	5.83	2.59-13.12	< 0.001
SGOT	1.00	1.00-1.00	< 0.001	1.00	1.00-1.00	< 0.001
SGPT	1.00	1.00-1.00	< 0.001	1.00	1.00-1.00	0.002
Albumin	0.11	0.04-0.31	< 0.001	0.15	0.05-0.44	< 0.001
Cr	2.08	1.31-3.31	0.002	1.52	1.00-2.31	0.051
Liver prognostic scores
MELD score	1.34	1.21-1.48	< 0.001	1.15	1.07–1.22	< 0.001
ALBI score	5.01	2.65-11.43	< 0.001	3.22	1.52-6.80	0.002
PALBI score	7.67	2.90-20.26	< 0.001	3.51	1.26-9.82	0.017
EZ-ALBI score	1.28	1.15-1.41	< 0.001	1.07	1.00-1.14	0.038

OR: Odds ratio; DSS: Dengue shock syndrome; INR: International normalized ratio; Cr: Creatinine; SGOT: Serum glutamate-oxaloacetate transaminase; SGPT: Serum glutamate-pyruvate transaminase; MELD: Model for end-stage liver disease; ALBI: Albumin-bilirubin; EZ-ALBI: Easy albumin-bilirubin; PALBI: Platelet-albumin-bilirubin.

Multivariate analysis using significant clinical and laboratory parameters from univariate analysis showed that INR remained the only significant predictor of both ALF and death, with an adjusted OR of 19.54 (95%CI: 3.37-113.38, P < 0.001) and 3.86 (95%CI: 1.13-13.18, P = 0.031), respectively (Table 4).

Table 4 Multivariate regression analysis of factors associated with severe adverse outcomes of dengue-induced severe hepatitis.

Parameter	Acute liver failure			In-hospital mortality
	OR	95%CI	P value	OR	95%CI	P value
Clinical parameters
Age	0.97	0.91-1.04	0.404	1.01	0.95-1.08	0.710
Presence of DSS	3.59	0.55-23.39	0.181	7.28	0.85-62.08	0.070
Baseline laboratory parameters
INR	19.54	3.37-113.38	< 0.001	3.86	1.13-13.18	0.031
TB	1.16	0.79-1.70	0.450	0.91	0.81-1.04	0.166
Albumin	0.40	0.08-1.94	0.254	0.54	0.10-3.05	0.480
Cr	1.17	0.73-1.88	0.511	1.22	0.76-1.97	0.409

OR: Odds ratio; DSS: Dengue shock syndrome; INR: International normalized ratio; TB: Total bilirubin; Cr: Creatinine.

The combination of laboratory parameters, calculated with liver-specific scores, showed that DISH with ALF had higher mean prognostic scores than DISH without ALF, and subsequent univariate analysis demonstrated that all liver-specific prognostic scores could be useful in predicting ALF (Table 2 and Table 3). MELD scores had the best performance in predicting ALF with very high accuracy with an AUROC value of 0.929 (87.5% sensitivity, 89.3% specificity at cutoff 16), followed by EZ-ALBI, ALBI, and PALBI scores with an AUROC value of 0.865, 0.832, and 0.797, respectively (Figure 2A and Table 5). All liver prognostic scores were valid in predicting death. MELD score had also the best performance and was quite accurate in predicting death with an AUROC of 0.822 (78.6% sensitivity, 87.7% specificity at cutoff 18), followed by EZ-ALBI, ALBI, and PALBI scores with an AUROC value of 0.807, 0.774, and 0.708, respectively (Figure 2B and Table 5).

Figure 2 Area under receiver operating characteristic curve values of prognostic scores. A: Predicting acute liver failure (ALF) from dengue-induced severe hepatitis (DISH). Model for end-stage liver disease (MELD) scores had the best performance in predicting ALF [area under receiver operating characteristic curve (AUROC) 0.929], followed by easy albumin-bilirubin (ALBI), ALBI, and platelet-ALBI (PALBI) scores (AUROC 0.865, 0.832, and 0.797, respectively); B: Predicting death from DISH. MELD score had the best performance in predicting death (AUROC 0.822), followed by easy ALBI, ALBI, and PALBI scores (AUROC 0.807, 0.774, and 0.708, respectively). ROC: Receiver operating characteristic; MELD: Model for end-stage liver disease; EZ-ALBI: Easy albumin-bilirubin; ALBI: Albumin-bilirubin; PALBI: Platelet-albumin-bilirubin.

Table 5 Diagnostic performance of each prognostic score in predicting acute liver failure and in-hospital death in patients with dengue-induced severe hepatitis.

Prognostic score	Acute liver failure			In-hospital death
	AUROC	P value	Cutoff (sensitivity, specificity)	AUROC	P value	Cutoff (sensitivity, specificity)
MELD score	0.929	< 0.001	16 (87.5%, 89.3%)	0.822	< 0.001	18 (78.6%, 87.7%)
EZ-ALBI score	0.865	< 0.001	-30 (85.7%, 74.0%)	0.807	0.001	-30 (81.8%, 69.3%)
ALBI score	0.832	< 0.001	-2.00 (83.3%, 75.2%)	0.774	0.001	-2.00 (78.6%, 72.7%)
PALBI score	0.797	< 0.001	-0.78 (75.0%, 78.0%)	0.708	0.010	-0.81 (71.4%, 70.4%)

AUROC: Area under the receiver operating characteristic curve; MELD: Model for end-stage liver disease; ALBI: Albumin-bilirubin; EZ-ALBI: Easy albumin-bilirubin; PALBI: Platelet-albumin-bilirubin.

A paired sample ROC curve estimation comparing the MELD and EZ-ALBI scores showed no significant difference between them in predicting ALF (Z = 1.688, P = 0.091, 95%CI: -0.014 to 0.194), indicating that while the MELD score was slightly better, the difference was minimal. For predicting death, the results were even closer (Z = 0.322, P = 0.747, 95%CI: -0.141 to 0.197), meaning that there was virtually no difference between the two scores. These findings suggest that while MELD remains the best scoring system for predicting serious liver-related outcomes, EZ-ALBI performs similarly and is a close second.

DISCUSSION

Dengue-induced liver injury is likely caused by a combination of direct viral damage to hepatocytes, dysregulated immune responses to the infection, cytokine storms, and ischemic liver injury resulting from CLS[4]. Although the reported incidence of ALF from dengue is low, it carries an exceptionally poor outcome with a mortality rate ranging from 20% to 68.3%[4,9]. In the natural course of dengue, characterized by three phases - febrile (2-7 d), critical or leakage (24-48 h), and convalescence (2-7 d)[10], ALF typically occurs during the critical phase, often concurrent with multiple organ failure[4]. Given that most patients seek medical attention during the febrile phase, early detection of those at risk of developing ALF and their triage for comprehensive monitoring are crucial for improving outcomes, particularly in resource-limited countries.

In the present study, the prevalence of DISH was 7.89%, consistent with other reports (3.80%-15%[3,4,11,12]). Previously, the prevalence of ALF from dengue ranged from 0.31 to 1.1%[9,13], which aligns with our finding of 0.98%. However, within the subset of patients with DISH, the prevalence of ALF increased significantly from 0.98% to 12.1%. Moreover, DISH patients with ALF had a markedly high mortality rate of 54.2% compared to those without (mortality rate 0.6%). Therefore, we propose screening all dengue patients with SGOT and SGPT detected at the time of diagnosis and considering repeat testing 5 d after illness onset (the peak enzyme levels according to previous research[14]) to detect severe hepatitis early. Patients with SGOT or SGPT levels exceeding ten times the upper normal limit should undergo ALF risk assessment, utilizing either clinical parameters or liver-specific scoring systems.

Previous studies identified age, atypical lymphocytes, platelet count[13], and the presence of DSS[3] as clinical parameters associated with ALF. However, those results were based on small studies[3,13] and were inconsistent with our results. Our study found that inflammatory parameters such as white blood cell and neutrophil counts, along with markers of end-organ damage such as INR, Cr, and hypoalbuminemia, were independent predictors of ALF and death. These parameters reflect severe inflammation or liver dysfunction, aligning with the major pathogenesis of dengue-induced liver injury[4,9]. Additionally, the presence of DSS, indicating severe systemic inflammation in the critical phase, was associated with increased liver injury from hepatic hypoperfusion, further correlating with ALF or death. For non-statistically significant parameters like platelet levels, SGOT, and SGPT, associated with poor liver outcomes, they showed consistent patterns with some studies[4,15] but differed from others[3,13,16]. These discrepancies may be due to study population heterogeneity, including variations in ethnicity, age, and treatment.

Prognostic scores in liver diseases offer enhanced reliability compared to individual factors by considering multiple variables and their interactions. Scores like the MELD score, ALBI, EZ-ALBI, and PALBI incorporate parameters such as TB, INR, Cr, and platelet levels, all identified as independent predictors of ALF in our study. By evaluating a combination of these factors, these scores provide a more comprehensive assessment of disease severity and predict patient outcomes more accurately than individual markers alone. Various liver-specific prognostic scores are utilized to assess the severity and prognosis of liver conditions. Some notable ones used in ALF include the MELD score, Clichy criteria, and King’s College Hospital criteria[17]. However, the clinical feasibility and prognostic predictive superiority of MELD score over others in ALF have been proposed by several studies[18-20]. In our study, the MELD score demonstrated the best predictive performance for ALF in dengue infection consistent with previous dengue studies[4,15] and other viral-induced acute liver injuries such as those caused by hepatitis A virus[21], hepatitis E virus[22], ctomegalovirus[23], and severe acute respiratory syndrome coronavirus 2[24]. Therefore, we suggest calculating the MELD score at the time of DISH diagnosis in all patients to promptly identify those at risk of poor prognosis.

The ALBI score was first introduced as an alternative to the Child-Pugh score for assessing liver function in HCC patients, irrespective of the degree of underlying liver fibrosis[25]. Although the ALBI score was originally designed for different purposes, its advantages over the MELD score, i.e., simpleness due to the use of only two laboratory parameters and continuous scale, have contributed to its increasing popularity and expanded applicability to predict various cirrhosis conditions, such as survival in ACLF, the development of HCC, and portal hypertension complications[26]. Outcomes from our study showed that the ALBI score could be used in predicting ALF and death from DISH with an excellent AUROC value of 0.832. This suggests its potential in predicting acute liver events in patients with healthy baseline liver function and that it could serve as an alternative to the MELD score, particularly in specific conditions where it may be limited by certain factors like the absence of some laboratory results, presence of chronic kidney diseases, or concurrent anticoagulant use[27].

The PALBI score is a modification of the ALBI score, incorporating platelet count along with the ALBI score, aiming for a more comprehensive assessment of liver function in HCC patients undergoing interventions. It has been further studied in predicting acute variceal bleeding in cirrhosis[8,28,29]. For patients with normally reserved liver functions, our study showed that the PALBI score was also useful in predicting ALF and death from dengue infection; nonetheless, its accuracy was inferior to the MELD and conventional ALBI scores. In addition, it is noteworthy to consider that severe dengue patients commonly exhibit concurrent thrombocytopenia. Consequently, relying on platelet counts may not offer sufficient specificity in predicting liver dysfunction. Therefore, the prognostic utility of the PALBI score in liver dysfunction assessment may be limited due to the confounding effect of thrombocytopenia. By the way, our study findings suggested that beyond its established role in predicting prognosis in HCC and cirrhosis, the PALBI score may also hold relevance in prognosticating acute liver conditions. Further investigation in this particular clinical context is warranted to validate and elucidate the potential utility of the PALBI score.

While the MELD and ALBI scores have undergone extensive study and validation across diverse liver conditions, a notable limitation lies in their formula’s complexity, posing challenges to clinical applicability. Introducing the EZ-ALBI score, a novel modification of the ALBI score, aims to address this limitation by simplifying the formula for bedside calculation while maintaining high accuracy[30]. The EZ-ALBI score was initially devised to predict prognosis in patients with HCC[8,30-32]. However, research investigating its utility in other liver conditions remains lacking. Previous studies[26] suggested that the ALBI score has a role in predicting prognosis in acute liver injury. Our findings, which support its predictive accuracy in DISH, lead us to hypothesize that the modified ALBI score, EZ-ALBI, is also valid for predicting prognosis in DISH. Our results were surprising, demonstrating that not only is the EZ-ALBI score valid for predicting ALF from DISH, but it also exhibits superior predictive accuracy compared to both the original ALBI score and the PALBI score. In comparison with the MELD score, although its performance was slightly lower, the EZ-ALBI score is simpler and easier to use, requiring fewer laboratory parameters. This makes it promising as an alternative predictor for ALF.

Recent research indicated that ALF deaths often coincide with concurrent multiple organ failure. ALF is thus regarded as a component of multiorgan failure stemming from immune hyperresponsiveness triggered by dengue infection[4]. Without timely intervention to halt the disease progression, individuals may succumb to these complications. Consequently, both independent parameters and liver-specific scores demonstrated similar trends in predicting both ALF and death as shown by results from this study. While DSS emerged as the most robust clinical predictor of both ALF and death, the MELD score consistently exhibited the highest predictivity regardless of the severity of dengue infection, followed by the EZ-ALBI, ALBI, and PALBI scores. The clinical utility of the MELD score in predicting ALF from adult dengue is further supported by previous studies from Thailand and India[4,15].

To integrate our study findings into clinical practice, we recommend testing liver function in all patients initially diagnosed with dengue infection, particularly DHF. If severe hepatitis is detected, characterized by ALT levels exceeding ten times the upper normal limit, further assessment using liver prognostic scores such as MELD or EZ-ALBI should be conducted based on resource availability. The EZ-ALBI score is highly beneficial for hospitals with limited resources or a high volume of patients, as it allows for quick and straightforward patient selection. Conversely, the MELD score is more suited to advanced healthcare facilities that require precise prognostic predictions, particularly liver transplant centers, where the MELD score can be used to assess the urgency of liver transplantation. Patients with a MELD score > 15 or an EZ-ALBI score > -30 should be closely monitored for potential ALF.

This study had some limitations. First, there was heterogeneity in the timing of biochemical blood collection for liver function tests, ranging from immediate venous sampling at the first visit to 48 h after admission, potentially affecting accuracy and leading to underdetection of severe hepatitis cases. Second, conducting a retrospective observational study over an extended duration resulted in variations in data collection practices and changes in treatment knowledge and technology, potentially influencing disease outcomes between older and more recent cases. Third, we used a complete case analysis strategy to handle cases with incomplete information. Since the exclusion was not random, this may introduce selection bias into the study. Lastly, in severe acute liver injury, fluctuations in liver function could lead to variability in prognostic score calculations, highlighting the need for serial assessments. However, the optimal frequency for these assessments remains uncertain. For future research, further study may be required to validate liver-specific models across different populations, including variations in ethnicity and hospital settings.

CONCLUSION

DSS is the strongest predictor of ALF and death in dengue infection. Irrespective of dengue severity, the MELD score is the best predictor. The EZ-ALBI score, simple and easy to use, exhibits excellent predictive performance, offering an alternative tool for prognosis assessment in dengue patients.