Validation of novel scoring systems for acute decompensated cirrhosis identifies PBC as an independent poor prognostic factor: a single-center Japanese cohort study.

Background: The Model for End-Stage Liver Disease 3.0 (MELD 3.0) was developed in the United States to improve prioritization for liver transplantation (LT); however, its utility in Japanese patients with liver cirrhosis (LC) and acute decompensation (AD) remains invalidated.

Methods: We retrospectively analyzed 312 patients with LC and first-time AD admitted to our institution between 2012 and 2022. Prognostic accuracy of MELD 3.0 was evaluated at 90 and 180 days after admission by comparison with other predictive models. Prognoses according to cirrhosis etiology and contributing factors were also examined.

Results: MELD 3.0 demonstrated superior prognostic accuracy at day 180 (C-index: 0.770) compared to the original MELD and MELD Na, although its C-index up to day 90 was comparable to that of MELD and MELD Na. A cut-off value of MELD 3.0 > 20.5 predicted LT or liver-related death at day 180. Patients with primary biliary cholangitis (PBC) had poorer outcomes than non-PBC cases through 180 days and remained an independent risk factor in the Cox proportional hazards model incorporating MELD 3.0. A progressive increase in both MELD 3.0 and total bilirubin from day 0 to day 90 after AD was observed specifically in the PBC group, which may have been associated with the poor prognosis at day 180.

Conclusions: MELD 3.0 was effective in predicting 180-day outcomes in Japanese patients with LC and AD. Progressive bilirubin elevation in PBC may be associated with poor prognosis. These findings suggest that early consideration of LT is warranted in patients with PBC.