Diagnostic Performance of Clinical, Biochemical and Ultrasonographic Parameters in Differentiating Biliary Atresia from Idiopathic Neonatal Hepatitis: A Prospective Study from a Tertiary Care Center of Nepal

Technical Summary: Diagnostic Performance of Clinical, Biochemical, and Ultrasonographic Parameters in Differentiating Biliary Atresia from Idiopathic Neonatal Hepatitis

Problem Statement
Neonatal cholestasis is a critical clinical condition characterized by impaired bile flow, with Biliary Atresia (BA) and Idiopathic Neonatal Hepatitis (INH) being the two most common etiologies. Differentiating between these conditions is urgent because BA requires immediate surgical intervention (Kasai procedure) to prevent irreversible liver damage, whereas INH is typically managed medically and often resolves spontaneously. In resource-limited settings, such as Nepal, access to definitive diagnostic tools like liver biopsy may be constrained, necessitating the identification of reliable, non-invasive markers to guide clinical decision-making.

Methodology
This was a prospective observational study conducted at Kanti Children's Hospital, a tertiary care center in Kathmandu, Nepal, from April 2025 to March 2026. The study enrolled 35 infants (aged 0–12 months) presenting with clinical suspicion of cholestasis, defined by conjugated bilirubin >1.0 mg/dL (if total bilirubin <5 mg/dL) or >20% of total bilirubin (if total >5 mg/dL).

• Reference Standard: Percutaneous liver biopsy histology served as the gold standard for final diagnosis.
• Data Collection: Researchers collected demographic data, clinical history (including stool color), and laboratory parameters (Complete Blood Count, Liver Function Tests including GGT, INR, AST, ALT, ALP, and TSB/DSB). Abdominal ultrasonography (USG) was performed, and a radiologist provided a diagnostic impression.
• Statistical Analysis: Categorical variables were compared using Fisher's Exact test. Continuous variables were analyzed using independent t-tests (for normally distributed data like TSB/DSB) or Mann-Whitney U tests (for non-normally distributed data like GGT, AST, ALT). Diagnostic performance metrics (sensitivity, specificity, PPV, NPV, accuracy) were calculated for individual modalities and their combinations. Parallel (OR) and Serial (AND) testing strategies were evaluated for the statistically significant variables: persistent acholic stool, serum GGT, and USG diagnosis.

Key Results
The study population consisted of 35 infants, with liver biopsy confirming BA in 25 cases (71.4%) and INH in 10 cases (28.6%).

• Clinical Features: Persistent pale (acholic) stool was the most discriminating clinical feature, present in 96% of BA infants and 0% of INH infants (p<0.001). Term gestation was significantly more common in the BA group (96%) compared to INH (60%, p=0.017).
• Biochemical Markers: Serum Gamma-Glutamyl Transferase (GGT) was significantly higher in BA (Median: 498 U/L) than in INH (Median: 170.5 U/L; p=0.004). ROC analysis identified an optimal GGT cutoff of ≥384.5 U/L, yielding a sensitivity of 72% and specificity of 90%. Other liver enzymes (ALT, AST, ALP) and coagulation profiles (INR) showed trends but did not reach statistical significance between groups.
• Ultrasonography: Gallbladder abnormalities (absent or contracted) were significantly more frequent in BA (100% of BA cases showed abnormalities) compared to INH (p<0.001). The radiologist's final USG diagnosis achieved 80% sensitivity and 60% specificity for BA.
• Histology: Liver biopsy revealed bile ductular proliferation, bile plugs, and portal fibrosis in nearly all BA cases (>96%), whereas giant cell transformation was present in 100% of INH cases but only 24% of BA cases.
• Combined Testing Strategies:
  • Parallel Testing (OR): Combining persistent acholic stool, GGT ≥384.5 U/L, and USG diagnosis of BA resulted in 96% sensitivity and 85.7% accuracy. However, specificity was lower (60%), leading to potential false positives.
  • Serial Testing (AND): Requiring all three tests to be positive yielded 100% specificity and 100% Positive Predictive Value (PPV), but sensitivity dropped to 56%, meaning nearly half of BA cases would be missed by this strict criteria alone.

Key Contributions
The study provides a validated diagnostic algorithm for resource-limited settings by quantifying the performance of simple, accessible markers. It establishes a specific GGT cutoff (384.5 U/L) for the local population and demonstrates that while no single non-invasive test is perfect, a stepwise approach can optimize diagnostic accuracy. The paper highlights that persistent acholic stool is a highly specific clinical sign, while GGT and USG offer complementary biochemical and structural data.

Significance and Claims
The authors conclude that while liver biopsy remains the gold standard, a rational, cost-effective strategy for settings where biopsy is not readily available or carries high risks involves a two-step approach:

1. Initial Triage (Parallel Testing): Utilizing the combination of persistent acholic stool, elevated GGT, and USG findings to achieve high sensitivity (96%), ensuring that BA cases are not missed during initial screening.
2. Confirmation/Expedition (Serial Testing): Using the "AND" logic (all three tests positive) to achieve 100% specificity, which can help prioritize infants for urgent surgical referral when the clinical picture is unequivocal.

The paper posits that adopting this stepwise parallel-then-serial testing framework can reduce unnecessary invasive procedures, shorten the time to Kasai surgery, and improve outcomes for infants with neonatal cholestasis in Nepal and similar resource-constrained environments.