Multicenter validation of the liver graft assessment following transplantation (L-GrAFT) score for assessment of early allograft dysfunction

Vatche G. Agopian, Daniela Markovic, Goran B. Klintmalm, Giovanna Saracino, William C. Chapman, Neeta Vachharajani, Sander S. Florman, Parissa Tabrizian, Brandy Haydel, David Nasralla, Peter J. Friend, Yuri L. Boteon, Rutger Ploeg, Michael P. Harlander-Locke, Victor Xia, Joseph DiNorcia, Fady M. Kaldas, Hasan Yersiz, Douglas G. Farmer, Ronald W. Busuttil

Research output: Contribution to journalArticlepeer-review

33 Scopus citations


Background & Aims: Early allograft dysfunction (EAD) following liver transplantation (LT) negatively impacts graft and patient outcomes. Previously we reported that the liver graft assessment following transplantation (L-GrAFT7) risk score was superior to binary EAD or the model for early allograft function (MEAF) score for estimating 3-month graft failure-free survival in a single-center derivation cohort. Herein, we sought to externally validate L-GrAFT7, and compare its prognostic performance to EAD and MEAF. Methods: Accuracies of L-GrAFT7, EAD, and MEAF were compared in a 3-center US validation cohort (n = 3,201), and a Consortium for Organ Preservation in Europe (COPE) normothermic machine perfusion (NMP) trial cohort (n = 222); characteristics were compared to assess generalizability. Results: Compared to the derivation cohort, patients in the validation and NMP trial cohort had lower recipient median MELD scores; were less likely to require pretransplant hospitalization, renal replacement therapy or mechanical ventilation; and had superior 1-year overall (90% and 95% vs. 84%) and graft failure-free (88% and 93% vs. 81%) survival, with a lower incidence of 3-month graft failure (7.4% and 4.0% vs. 11.1%; p <0.001 for all comparisons). Despite significant differences in cohort characteristics, L-GrAFT7 maintained an excellent validation AUROC of 0.78, significantly superior to binary EAD (AUROC 0.68, p = 0.001) and MEAF scores (AUROC 0.72, p <0.001). In post hoc analysis of the COPE NMP trial, the highest tertile of L-GrAFT7 was significantly associated with time to liver allograft (hazard ratio [HR] 2.17, p = 0.016), Clavien ≥IIIB (HR 2.60, p = 0.034) and ≥IVa (HR 4.99, p = 0.011) complications; post-LT length of hospitalization (p = 0.002); and renal replacement therapy (odds ratio 3.62, p = 0.016). Conclusions: We have validated the L-GrAFT7 risk score as a generalizable, highly accurate, individualized risk assessment of 3-month liver allograft failure that is superior to existing scores. L-GrAFT7 may standardize grading of early hepatic allograft function and serve as a clinical endpoint in translational studies ( Lay summary: Early allograft dysfunction negatively affects outcomes following liver transplantation. In independent multicenter US and European cohorts totaling 3,423 patients undergoing liver transplantation, the liver graft assessment following transplantation (L-GrAFT) risk score is validated as a superior measure of early allograft function that accurately discriminates 3-month graft failure-free survival and post-liver transplantation complications.

Original languageEnglish (US)
Pages (from-to)881-892
Number of pages12
JournalJournal of Hepatology
Issue number4
StatePublished - Apr 2021


  • Early allograft dysfunction
  • Ischemia-reperfusion injury
  • Liver transplantation
  • Risk prediction model

ASJC Scopus subject areas

  • Hepatology


Dive into the research topics of 'Multicenter validation of the liver graft assessment following transplantation (L-GrAFT) score for assessment of early allograft dysfunction'. Together they form a unique fingerprint.

Cite this