Massive blood transfusion after the first cut in liver transplantation predicts renal outcome and survival

Abstract

Transfusion requirements of blood products may provide useful prognostic factors for the prediction of short-term patient mortality and renal outcome after liver transplantation. Two hundred ninety-one consecutive liver transplants in adults were analysed retrospectively. Combined and living-related liver transplants were excluded. The amount of transfused packed red blood cells (PRBC) and units of platelets (UP) within the first 48 h were investigated as prognostic factors to predict short-term patient mortality and renal outcome. Receiver operating characteristic (ROC) curve analysis with area under the curve (AUC), Hosmer-Lemeshow tests and Brier scores were used to calculate overall model correctness, model calibration and accuracy of prognostic factors. Cut-off values were determined with the best Youden index. The potential clinical usefulness of PRBC as a prognostic factor to predict 30-day mortality (cut-off 17.5 units) and post-transplant haemodialysis (cut-off 12.5 units) could be demonstrated with AUCs >0.7 (0.712 and 0.794, respectively). Hosmer-Lemeshow test results and Brier scores indicated good overall model correctness, model calibration and accuracy. The UP proved as an equally clinically useful prognostic factor to predict end-stage renal disease (cut-off 3.5 units; AUC = 0.763). The association of cut-off levels of PRBC with patient survival (p < 0.001, log-rank test) and dialysis-free survival (p < 0.001, log-rank test) was significant (cut-off levels 17.5 and 12.5 units, respectively) as well as the association of UP with dialysis-free survival (p < 0.001, log-rank test) (cut-off level 3.5 units). The impressive discriminative power of these simple prognostic factors for the prediction of outcome after liver transplantation emphasizes the relevance of strategies to avoid excessive transfusion requirements.