Research report: Comparison of non-invasive serum markers for discriminating significant liver fibrosis in children with choledochal cyst

Children born with a choledochal cyst — a congenital dilation of the bile ducts that is the most common congenital biliary malformation in childhood — often carry hidden damage in their livers long before surgeons ever touch the duct. Persistent cholestasis from biliary obstruction can quietly drive chronic liver injury and fibrosis, and in some young patients the scarring is already severe, or even cirrhotic, by the time of diagnosis. Knowing how much fibrosis is present before an operation helps surgeons plan perioperative care, informs conversations with parents, and shapes postoperative follow-up. The trouble is that the only definitive way to stage fibrosis is a liver biopsy, which is invasive, carries a bleeding risk, and suffers from sampling error — hardly an ideal routine screening tool.

A new retrospective study from Shanxi Provincial Children’s Hospital in Taiyuan, China, published in Frontiers in Pediatrics, set out to answer a deceptively simple question: which of seven routine blood-based markers best distinguishes children with significant liver fibrosis from those without it? The researchers, led by Peng Zhang and Hongwei Xi, analyzed 219 children with choledochal cyst who underwent radical surgery with intraoperative liver biopsy between January 2020 and December 2025. Because liver biopsy was performed routinely at their center during the study period, cases were not selected according to the degree of fibrosis, which reduces one common source of bias in diagnostic studies.

The team compared seven markers calculated from blood tests taken within three days before surgery: gamma-glutamyl transpeptidase (GGT) on its own, the GGT-to-platelet ratio index (GPR), the aspartate aminotransferase-to-alanine aminotransferase ratio (AAR), the aspartate aminotransferase-to-platelet ratio index (APRI), the Fibrosis-4 index (FIB-4), the fibrosis cirrhosis index (FCI), and the S-index. All were judged against the METAVIR histological staging system applied to wedge biopsies taken from the anterior edge of the right hepatic lobe, with specimens required to contain at least five portal tracts. Two pathologists staged the biopsies independently, blinded to laboratory results and group assignment, with disagreements resolved by consensus and senior review when needed.

The cohort skewed heavily female — 165 of 219 children, or 75.3 percent — with a median age of 47 months. Most children had Todani type Ia (55.3 percent) or type IVa (38.8 percent) cysts. Fibrosis was widespread: only 10 children (4.6 percent) had completely normal livers, 103 (47.0 percent) had mild portal fibrosis without septa, and 106 children — 48.4 percent of the entire cohort — met the study’s threshold for significant fibrosis, meaning METAVIR stage F2 or above. Among the significant fibrosis group, 25 children had numerous septa without cirrhosis (F3) and 7 had established cirrhosis (F4). Preoperative cholangitis was strikingly common, affecting 54.8 percent of all children and 77.4 percent of those with significant fibrosis, compared with 33.6 percent of the no-or-mild fibrosis group.

When the numbers came in, GGT — an enzyme that reflects bile duct injury and cholestasis — emerged as the strongest single performer. It showed the highest correlation with METAVIR fibrosis stage (a Spearman coefficient of 0.638) and the highest area under the receiver operating characteristic curve: 0.826, with a 95 percent confidence interval of 0.770 to 0.882. The data-driven optimal cutoff, derived from the Youden index, was 123.0 U/L, yielding a sensitivity of 89.6 percent and a specificity of 69.9 percent. GPR followed at 0.814, S-index at 0.809, and FCI at 0.802, while APRI managed 0.771, AAR 0.705, and FIB-4 a dismal 0.528 — essentially no better than a coin flip.

Here is where the study becomes a lesson in statistical humility. After correcting for multiple pairwise comparisons with the Holm method, GGT’s apparent superiority evaporated against its closest rivals: the differences versus GPR, FCI, S-index, and APRI were not statistically significant. GGT significantly outperformed only AAR and FIB-4. The authors are explicit about this: their results suggest GGT has relatively good discriminatory performance, but the evidence is insufficient to conclude it is superior to other markers of similar strength. Correlation analysis reinforced the redundancy — GPR, S-index, and GGT correlated with each other at coefficients between 0.963 and 0.985, reflecting the shared GGT component and the largely overlapping information they carry.

Beyond discrimination, the researchers probed whether GGT was independently associated with significant fibrosis using multivariate logistic regression. In the primary model, adjusting for age, sex, maximum cyst diameter, jaundice history, and cholangitis history, each 10 U/L increase in GGT was associated with an odds ratio of 1.041 for significant fibrosis (95 percent CI 1.021 to 1.062). The association persisted in an extended model that added total bilirubin, alkaline phosphatase, and total bile acid, and it survived even after excluding all 120 children with a history of preoperative cholangitis, leaving an odds ratio of 1.055 per 10 U/L. But the discriminatory power of GGT dropped from an AUC of 0.826 to 0.771 in that cholangitis-free subgroup — a clear signal that part of its diagnostic signal rides on the biliary inflammatory background rather than fibrosis itself. The authors caution that GGT, bilirubin, alkaline phosphatase, and bile acids all sit on the same cholestasis axis, so the extended model should be read as exploratory and cannot establish that GGT acts independently of cholestatic factors, let alone causally.

A stratified analysis made the graded relationship vivid. Children with GGT below 100 U/L had significant fibrosis 19.0 percent of the time; those between 100 and 300 U/L, 56.3 percent; and those above 300 U/L, 69.6 percent — a significant linear trend (chi-square for trend 42.90). Yet the authors warn against treating 123.0 U/L, or any of these strata, as a universal diagnostic threshold. The cutoffs of 100 and 300 U/L were borrowed from infantile cholestasis literature, while this cohort’s median age was 47 months, and the ROC-derived cutoff is data-driven and vulnerable to overfitting; bootstrap resampling placed its range at 86 to 161 U/L. In an exploratory analysis of advanced fibrosis (F3 or worse, only 32 cases), FCI edged out GGT with AUCs of 0.906 versus 0.901, but the authors stress this analysis was too small to establish clinical decision thresholds.

Why did the adult-derived scores fare so poorly? FIB-4 was built in adult chronic liver disease populations, combining age, AST, ALT, and platelet count — variables whose pediatric reference ranges shift dramatically with age and whose pathological meaning changes in a disease dominated by biliary obstruction rather than hepatocellular injury. Notably, platelet counts were actually higher in the significant fibrosis group, the opposite of the thrombocytopenia pattern seen in adult cirrhosis, likely reflecting inflammation rather than portal hypertension. AAR showed an inverse relationship with fibrosis, probably because ALT rose far more than AST in the fibrotic group. The study has real limitations: it is single-center and retrospective, all patients underwent biopsy (inviting spectrum bias), METAVIR staging is not specific to biliary fibrosis, there was no systematic follow-up, and no elastography comparison was possible. The authors’ bottom line is measured: GGT can serve as an auxiliary clue in preoperative assessment, but it cannot replace histopathological evaluation, and multicenter validation is needed before any of these markers earns a place in routine pediatric CDC care.

Subject of Research: Comparing serum markers for detecting significant liver fibrosis in children with choledochal cyst

Article Title: Comparison of non-invasive serum markers for discriminating significant liver fibrosis in children with choledochal cyst

Article References: Zhang, P., Zhen, Y., Liu, Y., Xu, Z., & Xi, H. (2026). Comparison of non-invasive serum markers for discriminating significant liver fibrosis in children with choledochal cyst. Frontiers in Pediatrics, 14, 1992446. https://doi.org/10.3389/fped.2026.1992446

Image Credits: AI Generated

DOI: 10.3389/fped.2026.1992446

Keywords: choledochal cyst, liver fibrosis, GGT, non-invasive markers, pediatric surgery, METAVIR staging, APRI, FIB-4, cholestasis, ROC analysis, biliary obstruction, retrospective study