Autoimmune Hepatitis Profiles After Liver Transplantation in Latin American Patients: A Cluster Analysis of Clinical...

Most existing evidence regarding liver transplantation (LT) for autoimmune hepatitis (AIH) originates from European cohorts, hindering the applicability of current guidelines to other populations because of possible geographic variations. There is an urgent need for region-specific prognostic models.

We analyzed 164 AIH patients from Brazil and Venezuela undergoing LT between 2002 and 2025, the largest AIH LT cohort from Latin America to date. Random survival forests (RSF) identified key predictors of overall survival (OS), graft survival (GS), and AIH recurrence (rAIH). Clustering based on predicted risks identified high-risk groups with significantly worse outcomes (p < 0.001 for all). Penalized models with the most important variables from the RSF assessed risk stratification.

5-, 10-, and 20-year OS were 83%, 78%, and 49%. rAIH rates increased from 6% at 5 years to 34% at 20 years. Mycophenolate sodium use in maintenance therapy was ranked as the most important factor for OS (Hazard ratio [HR] = 0.44) and GS (HR = 0.46) prediction, while younger age was the most important risk factor for rAIH. LASSO models accurately predicted early outcomes (1-year AUCs: OS 0.73, GS 0.73, rAIH 0.81). In a landmark analysis (6 months post-LT), infectious complications emerged as the main determinant for long-term OS, and a new model improved 5- and 10-year OS prediction (AUCs: 0.76 and 0.79, respectively).

This analysis identifies key predictors that define high-risk clusters for early and long-term outcomes in Latin American AIH patients post-LT, highlighting potential region-specific factors.