Establishment and validation of a predictive model of immune tolerance after pediatric liver transplantation: a multicenter cohort study

Abstract

JOURNAL/ijos/04.03/01279778-202409000-00031/figure1/v/2024-09-12T150105Z/r/image-jpeg Background: Side-effect of life-long immunosuppressants (IS) administration is a major obstacle for the long-term survival of pediatric liver transplantation (LT) recipients. Immunotolerance is the status that recipients discontinued IS with normal liver function and intrahepatic histology. So far, only a few clinical parameters were identified related with tolerance but failed to accurately discriminate tolerant recipients in clinical practice. Here, the authors aimed to provide a comprehensive view of pre-LT and post-LT risk factors associated with the achievement of tolerance after pediatric LT and established a tolerance predictive nomogram (ITPLT) with high accuracy and specificity. Methods: The authors enrolled 2228 pediatric recipients who received LT in Renji Hospital between October 2006 and December 2020. All participants survived over 3 years after transplantation with comprehensive and intact medical history and follow-up data. They were randomly assigned to training and validation cohorts in accordance with a ratio of 1:1. Univariate and multivariable Logistic regression were used to identify clinical factors associated with post-LT immune tolerance and establish a predictive model. The model was further validated in an independent external validation cohort from Tianjin First Central Hospital. Results: Among all participants, 6% recipients successfully tapered IS with intact allograft function. The most common reason for IS discontinuity was pneumonia. Univariate analysis identified 15 clinical factors associated with tolerance achievement, including age at LT, follow-up time, preoperative total bilirubin, creatinine, INR, CYP polymorphism, types of transplantation, massive postoperative ascites, episodes of acute rejection, and the severity of EBV and CMV infection. Using multivariable Logistic regression, the authors established the predictive ITPLT model for post-LT tolerance, which included seven easily accessible clinical factors (age at LT, CYP3A5 genotype, types of transplantation, post-LT massive ascites, preoperative INR, creatinine, and total bilirubin levels). Then, the authors visualized the model using nomogram. The c-statistics for predicting tolerance achievement in the training, internal validation, and external validation cohorts were 0.854, 0.787, and 0.746, respectively. Conclusion: Multiple pre-LT and post-LT clinical factors affected the process of immune remodeling after pediatric LT. The predictive ITPLT model, composed of seven easily accessible clinical factors, could comprehensively reveal the effect of these clinical parameters on immune remodeling and accurately identify tolerant recipients after pediatric LT. The application of ITPLT could facilitate the individualized IS strategy in the future.