Title
WHEN HOPE IS NOT ENOUGH: A NOMOGRAM TO PREDICT OUTCOME OF LIVER GRAFTS TREATED BY HYPOTHERMIC OXYGENATED MACHINE PERFUSION (HOPE)
Authors
Damiano Patrono, Nicola De Stefano, Rebecca Panconesi, Victor Ugo De Donato, Federica Ferraina, Giorgia Catalano, Davide Cussa, Caterina Costanza Zingaretti, Stefano Mirabella, Renato Romagnoli
General Surgery and Transplantation, AOU Città della Salute e della Scienza, Turin, Italy
Background
Hypothermic oxygenated machine perfusion (HOPE) has been associated to superior liver transplantation (LT) outcomes as compared to static cold storage, but clinical indications remain uncertain. Aim of the study was to analyze risk factors of liver grafts treated by HOPE and build an interactive nomogram driving clinical decision making.
Methods
HOPE LT were identified from a single-Institution prospectively maintained database in the period March 2016 – August 2025. Poor outcome was defined as one of the following: 1) primary nonfunction (PNF); 2) graft loss ≤90 days from LT; 3) L-GrAFT score ≥-0.57 (high or very high risk of 90-day graft loss). Poor outcome predictors were identified by logistic and Random Forest regression. The Shiny R package was used to build the interactive app.
Results
446 HOPE LT, representing 31% of all LT performed during study period, were identified. Median (IQR) donor age was 72.1 (60.3-80.9) and 78 (17.5%) livers were from DCD donors. 52 (11.6%) met criteria for poor outcome (PNF, n=1; 90-day graft loss, n=21; L-GrAFT ≥-0.57, n=41). Other clinical outcomes, including patient and graft survival, were also inferior in poor outcome patients. At multivariable logistic regression, graft macrosteatosis ≥15% was the only independent predictor of poor outcome (OR 2.91; 95%CI 1.27-6.40; p=0.009), whereas at Random Forest analysis recipient MELD-Na, albumin level and graft macrosteatosis were the variables with the higher mean decrease gini (11.1, 8,5 and 3.6, respectively), thereby with a stronger association with outcome (Figure). Coefficient from multivariable logistic regression were used to model the risk of poor outcome based on a set of covariates available pre-LT and build an interactive user-friendly tool for clinical decision making.
Conclusions
Recipient factors are main determinant of outcome in HOPE LT. Use of liver grafts with significant macrosteatosis, especially if allocated to sicker recipients, is associated with a risk of poor outcome, which suggests their cautious use, ideally with some kind of viability assessment.
