Evaluation of Substitution of Total Bilirubin by Direct Bilirubin in Prediction of Mortality Models in patients with acute and Acute-on-chronic liver failure

Technical Summary: Evaluation of Substitution of Total Bilirubin by Direct Bilirubin in Prediction of Mortality Models

Problem Statement
Acute liver failure (ALF) and acute-on-chronic liver failure (ACLF) are life-threatening conditions with high mortality rates, necessitating reliable prognostic biomarkers for risk stratification and clinical decision-making. Traditionally, total bilirubin (TB) has served as a core component of established mortality prediction models, such as the Model for End-Stage Liver Disease (MELD), MELD-Na, and the AARC score. However, TB encompasses both conjugated (direct) and unconjugated (indirect) fractions, the latter of which can be influenced by extrahepatic factors such as hemolysis or genetic conditions like Gilbert's syndrome. Recent evidence suggests that direct bilirubin (DB), the conjugated fraction, may offer superior prognostic utility by more accurately reflecting hepatobiliary dysfunction and cholestasis. Despite this, the specific contribution of DB versus TB in predicting mortality within ALF and ACLF cohorts remains under-explored, particularly regarding the modification of existing scoring systems.

Methodology
This prospective observational study was conducted at the National Liver Institute, Menoufia University, Egypt, between August 2022 and July 2023. The study population comprised 104 patients: 19 with ALF and 95 with ACLF.

• Definitions: ALF was defined by rapid coagulopathy (INR > 1.5) and hepatic encephalopathy without pre-existing liver disease. ACLF was defined using APASL criteria (acute hepatic insult with jaundice ≥5 mg/dL and coagulopathy INR ≥1.5, accompanied by ascites or encephalopathy within four weeks in patients with chronic liver disease).
• Data Collection: Demographic, clinical, and laboratory data were collected, including liver function tests (TB, DB, ALT, AST, ALP, GGT, albumin, PT), renal function, and electrolytes. Infection severity and ascites were graded according to standard guidelines.
• Intervention: The study recalculated standard mortality prediction models (MELD, MELD-Na, and AARC scores) by substituting TB with DB.
• Statistical Analysis: Predictive accuracy was assessed using Receiver Operating Characteristic (ROC) curves and Area Under the Curve (AUC) analysis. Univariate and multivariate logistic regressions were performed to identify independent mortality predictors.

Key Results

• ACLF Cohort (n=95): The mortality rate was 84.2% (80 deaths). Significant predictors of mortality included ascites severity, acute kidney injury (AKI), and higher AARC scores.
  • Bilirubin Performance: DB levels were significantly higher in non-survivors (13.4 ± 4.8 mg/dL) compared to survivors (9.1 ± 3.6 mg/dL).
  • Model Improvement: Substituting TB with DB improved the predictive accuracy of the AARC score (AUC increased from 0.74 to 0.81) and the MELD-Na score (AUC increased from 0.72 to 0.79). The optimal DB cutoff for mortality prediction was identified at 5.6 mg/dL.
  • Regression: Multivariate analysis identified ascites severity as an independent predictor, while DB showed borderline significance (p=0.056) in the univariate context but improved model performance when integrated into scores.
• ALF Cohort (n=19): The mortality rate was 63.1% (12 deaths).
  • Bilirubin Performance: DB levels were significantly elevated in non-survivors (15.8 ± 5.6 mg/dL) versus survivors (10.2 ± 4.3 mg/dL). Severe infection was a significant predictor (p=0.002).
  • Model Performance: Both TB and DB demonstrated perfect discrimination in predicting mortality, with an AUC of 1.0 for both the MELD-Na score and the bilirubin levels themselves.

Key Contributions

1. Prognostic Superiority of DB: The study provides evidence that DB outperforms TB in predicting mortality in ACLF patients, offering a more specific measure of hepatic excretory capacity by excluding the confounding effects of unconjugated bilirubin.
2. Enhancement of Scoring Systems: The research demonstrates that substituting TB with DB in established scoring systems (MELD-Na and AARC) significantly enhances their discriminatory power (AUC) in ACLF patients.
3. First Evaluation in ALF: This manuscript claims to be the first study to evaluate the role of DB as a substitute for TB specifically within an ALF cohort, finding that DB maintains robust predictive accuracy comparable to TB in this acute setting.
4. Clinical Applicability: The study highlights that DB measurement is widely available and does not impose additional resource burdens, making it a feasible candidate for routine integration into prognostic algorithms.

Significance and Claims
The authors assert that the substitution of TB with DB represents a transformative approach to refining prognostic algorithms for liver failure. By offering greater specificity, DB-based models reduce potential confounders and provide a more precise representation of hepatic dysfunction. The study concludes that incorporating DB into existing scoring systems enhances predictive accuracy, thereby facilitating better risk stratification, earlier identification of high-risk patients, and more targeted therapeutic interventions.

However, the authors maintain a modest stance regarding the generalizability of these findings. They explicitly acknowledge limitations, including the small sample size (particularly the n=19 ALF group) and the lack of an external validation cohort. Consequently, the paper concludes that while the results are promising, large-scale, multicenter validation studies are warranted to confirm these findings and establish the utility of DB-based models across broader populations.