Lack of agreement for defining 'clinical suspicion of rejection' in liver transplantation: a model to select candidates for liver biopsy

The gold standard to diagnose acute cellular rejection (ACR) after liver transplantation (LT) is histological evaluation, but there is no consensus to select patients for liver biopsy. We aimed to evaluate the agreement among clinicians to select candidates for liver biopsy early after LT. From a protocol biopsy population (n = 690), we randomly selected 100 LT patients in whom the biopsy was taken 7-10 days after LT. The clinical information between LT and protocol biopsy was given to nine clinicians from three transplant centres who decided whether a liver biopsy was needed. The agreement among clinicians to select candidates for liver biopsy was poor: κ = 0.06-0.62, being κ < 0.40 in 76% of comparisons. The concordance between indication for liver biopsy and moderate-severe ACR in the protocol biopsy was κ < 0.30 in all cases. A multivariate model based on the product age-by-MELD (OR = 0.81; P = 0.013), delta eosinophils (OR = 1.5; P = 0.002) and mean tacrolimus trough concentrations <6 ng/ml within the prior 4 days (OR = 11.4; P = 0.047) had an AUROC = 0.84 to diagnose moderate-severe histological ACR. In conclusion, the agreement among clinicians to select patients for liver biopsy is very poor. If further validated the proposed model would provide an objective method to select candidates for liver biopsy after LT.