Portal vein thrombosis compromises the performance of MELD and MELD-Na scores in patients with cirrhosis.

Abstract

The accuracy of model for end-stage liver disease (MELD) and MELD with sodium (MELD-Na) scores in reflecting the clinical outcomes of patients with cirrhosis and portal vein thrombosis (PVT) remains unclear. This study aimed to evaluate the performance of scores in predicting 90-day mortality in patients with cirrhosis and PVT. Post hoc analysis was performed in two prospective cohorts (NCT02457637 and NCT03641872). The correlation between the MELD/MELD-Na score and 90-day liver transplantation (LT)-free mortality was investigated in patients with cirrhosis with and without PVT. In this study, 2826 patients with cirrhosis were included, and 255 (9.02%) had PVT. The cumulative incidence of 90-day LT-free mortality did not significantly differ between patients with and without PVT (log-rank P=0.0854). MELD [area under the receiver operating curve (AUROC), 0.649 vs. 0.842; P=0.0036] and MELD-Na scores (AUROC, 0.691 vs. 0.851; P=0.0108) were compared in patients with and without PVT, regarding the prediction of 90-day LT-free mortality. In MELD<15 and MELD-Na<20 subgroups, patients with PVT had a higher 90-day LT-free mortality than those without PVT (7.91% vs. 2.64%, log-rank P=0.0011; 7.14% vs. 3.43%, log-rank P=0.0223), whereas in MELD≥15 and MELD-Na≥20 subgroups, no significant difference was observed between patients with and without PVT. The performance of MELD and MELD-Na scores in predicting 90-day LT-free mortality of patients with cirrhosis was compromised by PVT. MELD<15 or MELD-Na<20 may underestimate the 90-day LT-free mortality in patients with PVT.