Abstract
BACKGROUND: Normothermic machine perfusion (NMP) maintains physiological metabolism ex situ and enables functional assessment of liver grafts. However, the clinical performance of predefined viability criteria in liver transplantation from brain-death donors (DBD) remains uncertain. METHODS: This single-center study, an ancillary of the OVERNIGHT NMP program (Rennes University Hospital, 2021-2023), evaluated predefined viability thresholds applied during prolonged end-ischemic NMP. All adult DBD grafts retrieved between 16:00 and 00:00 underwent NMP until scheduled transplantation the following morning. Grafts meeting predefined viability criteria were transplanted. Outcomes were compared with 1:3 propensity-matched controls preserved by static cold storage (SCS) in exploratory, hypothesis-generating analyses. The primary endpoint was early allograft dysfunction (EAD); a modified definition (mEAD, excluding transaminase criteria) was used as a sensitivity analysis. Secondary endpoints included primary non-function (PNF), biliary and vascular complications, and death-censored graft survival. RESULTS: Among 382 liver transplants, 18 NMP and 54 matched SCS recipients were analyzed. All 18 NMP grafts fulfilling predefined thresholds were transplanted, while two non-viable grafts were discarded. No PNF occurred. EAD and mEAD rates were comparable between NMP and SCS (44.4% vs. 26.0%, p = 0.204; 17% vs. 24%, p = 0.550). Biliary complications, requiring endoscopic or surgical intervention, were similar (43.8% vs. 36.6%, p = 0.684). Death-censored graft survival at 24 months did not differ (88.5% vs. 96.0%, p = 0.584). CONCLUSIONS: Predefined viability criteria applied during NMP allowed safe identification of transplantable DBD grafts. However, functional assessment based solely on hepatocellular markers was insufficient to anticipate biliary complications, underscoring the need for integrative viability algorithms incorporating cholangiocellular or molecular indicators.