Development of a Novel Prognostic Nomogram for High Model for End-Stage Liver Disease Score Recipients Following Deceased Donor Liver Transplantation

BACKGROUND: A high model of end-stage liver disease (MELD) score (>30) adversely affects outcomes even if patients receive prompt liver transplantation (LT). Therefore, balanced allocation of donor grafts is indispensable to avoid random combinations of donor and recipient risk factors, which often lead to graft or recipient loss. Predictive models aimed at avoiding donor risk factors in high-MELD score recipients are urgently required to obtain satisfactory outcomes. METHOD: Data of patients with MELD score >30 who underwent LT at three transplantation institutes between 2015 and 2018 were retrospectively reviewed. Early allograft dysfunction (EAD), length of intensive care unit (ICU) stay, and graft loss were recorded. Corresponding independent risk factors were analyzed using stepwise multivariable regression analysis. A prediction model of graft loss was developed, and discrimination and calibration were measured. RESULTS: After applying the exclusion criteria, 778 patients were enrolled. The incidence of EAD was 34.8% (271/778). Donor graft macrovesicular steatosis, graft-to-recipient weight ratio (GRWR), warm ischemia time (WIT), cold ischemia time (CIT), and ABO blood incompatibility, together with donor serum albumins, were independent predictors of EAD. The incidence of ICU stay over 10 days was 64.7% (503/778). Donor age, recipient's MELD score, Child score, and CIT were independent predictors of ICU stay. The 3-year graft survival rates (GSRs) in the training and validation cohorts were 64.2 and 59.3%, respectively. The independent predictors of graft loss were recipient's Child score, ABO blood type incompatibility, donor serum total bilirubin over 17.1 μmol/L, and cold CIT. A nomogram based on these variables was internally and externally validated and showed good performance (area under the receiver operating characteristic curve = 70.8 and 66.0%, respectively). For a recipient with a high MELD score, the avoidance of ABO blood type incompatibility and CIT ≥6 h would achieve a 3-year GSR of up to 78.4%, whereas the presence of the aforementioned risk factors would decrease the GSR to 35.4%. CONCLUSION: The long-term prognosis of recipients with MELD scores >30 could be greatly improved by avoiding ABO blood type incompatibility and CIT ≥6 h.