[A non-invasive diagnostic evaluation study on liver fibrosis in children with autoimmune hepatitis]

Objective: To compare the concordance and diagnostic performance of eight non-invasive diagnostic models with liver biopsy-based fibrosis staging, and to evaluate their clinical utility in staging liver fibrosis in children with autoimmune hepatitis (AIH). Methods: This retrospective cohort study included 52 children diagnosed with AIH who underwent liver biopsy in the Department of Infectious Diseases at Children's Hospital of Fudan University between January 2020 and May 2025. According to the stage of fibrosis (S) in liver biopsy, the patients were divided into three groups: S0-S1, S2, and ≥S3. Relevant laboratory parameters were collected to calculate the indices, including the aspartate aminotransferase (AST) to alanine aminotransferase ratio, AST-to-platelet ratio index (APRI), Forns index, fibrosis-4 (FIB-4) index, γ-glutamyl transpeptidase-to-platelet ratio (GPR), King's index, S index, and Liver stiffness-to-platelet ratio index (LPRI). One-way analysis of variance and the Kruskal-Wallis H test were used for comparisons among multiple groups. Spearman correlation analysis was performed to assess the association between the 2 groups of variables. Draw the receiver operating characteristic (ROC) curve, calculate the area under the curve (AUC), and use DeLong test for comparison of the diagnostic efficacy of each model. Results: Among the 52 children with AIH, 13 were boys and 39 were girls, the age of hospitalized children was (10±3) years. S0-S1 4 cases, S2 13 cases, ≥S3 35 cases, significant differences were observed in the Forns index, GPR, S index, and LPRI among the three groups (all P<0.05). The Forns index, GPR, S index and LPRI were all positively correlated with liver fibrosis stage (r_s=0.48, 0.40, 0.49, and 0.71, respectively; all P<0.05). The AUC values of Forns index, GPR, S index and LPRI diagnosis ≥S3 group were 0.72, 0.70, 0.76, and 0.87, respectively. Parallel testing combining LPRI and the S index yielded a sensitivity of 0.93 for diagnosing ≥S3 fibrosis 0.19 and 0.22 higher than single diagnostic model LPRI and S index, respectively. Conclusions: Among the eight non-invasive diagnostic models evaluated, LPRI showed good concordance with pathological fibrosis staging in children with AIH and demonstrated the highest diagnostic performance for detecting advanced fibrosis (≥ S3). The combination of LPRI and the S index significantly improved the detection rate of advanced fibrosis.