Is Model for End-stage Liver Disease 3.0 Better Than Model for End-stage Liver Disease? Evaluating the Association of Liver Disease Severity Scores With Perioperative Complications in Liver Transplant Recipients.

Abstract

Background: Sequential adaptations to Child-Pugh (CP) and MELD have improved prediction of waitlist mortality in liver transplant (LT). Despite its widespread use as a prognosticator, the association between the MELD score and perioperative adverse events during LT has yet to be evaluated. this study seeks to evaluate whether advances in MELD score calculations correspondingly improve predictions for massive transfusion (MT) and renal failure.

Methods: Adult patients undergoing LT at a tertiary institution between 2015 and 2023 were enrolled. MELD, MELD-Na, MELD 3.0, and CP were calculated at time of LT. Massive transfusion (MT) was >6 units of red blood cells before hepatic artery ligation. Renal failure (RF) was defined as requiring dialysis on postoperative-day one. Area-under-the-receiver-operating-characteristic curves (AUC) was estimated for each score and outcome and compared using the DeLong method. Score performance was evaluated using receiver operator curves (ROC) with a high performing assay considered as an area under the curve (AUC) >0.800.

Results: Total 265 patients were included; 20 (7.6%) received MT, 31 (11.8%) had RF. For MT, scores performed similarly (CP 0.70 [95% CI: 0.58, 0.81]; MELD 0.69 [0.59, 0.80]; MELD-Na 0.71 [0.61, 0.81]; MELD 3.0 0.69 [0.59, 0.80]). For RF all MELD scores outperformed CP, and MELD-Na outperformed MELD 3.0 (0.58 [0.48, 0.68], 0.66 [0.55,0.77], 0.67 [0.56, 0.78], and 0.65 [0.53, 0.77]).

Conclusion: MELD 3.0 did not outperform its predecessors. MELD-Na may still have a role in assessment of perioperative complications in LT recipients as well as patients with end-stage liver disease undergoing nontransplant operations.