A New Liver Allocation Score (LivAS) to Facilitate the Weighing of Urgency Against Utility in Liver Transplantation.

Abstract

Introduction: A valid prognostic model with donor and recipient data for the prediction of 3-month mortality after transplantation would be useful to improve donor organ allocation by preoperative weighing of the predicted outcome after transplantation against the urgency of transplantation as predicted by the MELD-score. Methods: This is a multicentre analysis with data from two German transplant centres within the Eurotransplant community. Included were all consecutive liver transplants performed inadult recipients (minimum age 18 years) in Hannover from 01.09.2001-31.12.2010(n=770) and in Kiel from 01.01.2006-08.02.2013 (n=234). Excluded were all combined transplants and living-related transplants. The cohort from Hannover was randomised retrospectively into an internal training group for model development with multivariate binary logit link regression and an internal validation group. The cohort from Kiel was used as an external validation group. Results: Recipient score component: Recipient y = -1.60 + (0.023 x age at transplant in years) + (-0.129 x haemoglobin in g/dl) + (-0.001 x thrombocytes in tsd/μl) + (-0.003 x factor II in %) + (0.003 x creatinine in μmol/l) + (0.474 x artificial ventilation, if yes=1, if no=0) + (0.195 x preTX portal vein thrombosis, if yes=1, if no=0) + (0.894 * risk indication re-TX for primary non-function, if yes=1, if no=0). Donor score component: Donor y = -2.281 + (0.071 x donor liver with macrovesicular steatosis in %, if no biopsy = 0) + (0.544 x ICU stay >11.5 days) + (1.266 x donor sodium >160mmol/l). LivAS = 1,342 + (1,162 x Recipient y) + (0,616 x Donor y). The area under the receiver operating curve for the prediction of 3-month mortality after transplantation was >0.700 in the in the training cohort and the internal and external validation cohorts. The influence of the LivAs above versus below cut-off was significant in all three cohorts (p<0.001). The Goodness-of-Fit test results demonstrated good model fit for all cohorts. Conclusion: The proposed LivAS is a validated prognostic model that could improve organ allocation rules significantly.