LAUSR

With great interest, we read the article by Dutkowski and Clavien on several scoring systems that are used worldwide to predict outcome or futile outcome after liver transplantation (LT). Their study shows the variety of models that have been developed to get a better indication on donor and/or recipient risk. In the Eurotransplant (ET) region 3 countries use the Model for End-Stage Liver Disease (MELD) for centralized liver allocation: Belgium, Germany, and the Netherlands. The other member states apply a center-based liver allocation (Austria, Croatia, Hungary, and Slovenia). In 2012, we published the ET– donor risk index (DRI), a donor risk model specifically designed for the ET region, adapted from the DRI. Because outcome also (mainly) depends on recipient-related factors, its goal is not primarily to predict outcome after LT. Furthermore, donor quality is not a simple “yes” or “no.” It is not dichotomous, but a continuum and describes the steadily increasing risk depending on the parameters involved. Therefore, the ET-DRI is a continuous score that should ideally not be capped at a certain value, as was done in the current study (Tables 1 and 2; DRI, 1.8). Furthermore, if one wishes to use a threshold for a continuous score, this should be in line with the donor quality (or recipient status) of the country or region for which this threshold is used. For example, the DRI between the United Network for Organ Sharing and ET is not comparable and would automatically lead to another threshold for acceptance or decline in light of the available donor organs. A more complete way of estimating posttransplant outcome would be by combining donor and recipient risks. It would have been interesting if the authors had compared the recently published donor-recipient model (DRM) with the models they describe in the current study. Our preference goes to such a combined DRM that assesses the specific risks of the donor and the specific risks of the recipient separately and/or combined. However, such a model that consists of donor and recipient factors cannot be used to describe donor quality. Obviously, a risk model will never give a definitive answer whether or not to accept a liver allograft. It is a tool to estimate the (pre-)transplant risks involved. In our opinion, risk models are ideal to objectively compare outcomes between transplant centers or regions/ countries, by correcting for the donor and/or recipient risks involved (so-called “case-mix”). A limitation of risk models is that they are constructed based on retrospective data. Looking at the differences between Tables 1 and 2, it could well be that the higher positive predictive value and specificity of the BAR (balance of risk) score is explained by the fact that the authors used their own model to accept or decline a liver allograft, leading to a self-fulfilling prophecy. An important issue that is addressed by the authors is the limitation of MELD with regard to the lack of pretransplant and posttransplant quality of life (or morbidity). The authors show the correlation between patient morbidity, measured by the comprehensive complication index, and several donor and/or recipient models. Again, the BAR score seems superior to other models. We would like to point out that it would be unlikely that the DRI (or ET-DRI), a model Address reprint requests to Joris J. Blok, M.D., Department of Surgery, Division of Transplantation, Leiden University Medical Center, Albinusdreef 2, 2333 ZA Leiden, the Netherlands. Telephone: 1 31 71 526 61 88; FAX: 1 31 71 526 67 50; E-mail: [email protected]