RESUMO
BACKGROUND: Predictors of poor outcome associated with variceal bleeding remain suboptimal. In patients with cirrhosis, serum lactate combined with Model for End-Stage Liver Disease (MELD-LA) improved prediction across heterogeneous populations. However, prognostic properties have not yet been assessed in the context of variceal bleeding. AIMS: We aimed to evaluate the predictive performance of MELD-LA compared to MELD, lactate, and nadir hemoglobin in cirrhosis patients with variceal bleeding. METHODS: In this multicenter study, we identified 472 patients with variceal bleeding from a German primary cohort (University Hospitals Hamburg/Frankfurt/Cologne), and two independent external validation cohorts [Veterans Affairs (VA), Baylor University]. Discrimination for 30-day mortality was analyzed and scores were compared. MELD-LA was evaluated separately in validation cohorts to ensure consistency of findings. RESULTS: In contrast to nadir hemoglobin, MELD and peak-lactate at time of bleeding were significantly higher in 30-day non-survivors in the primary cohort (p = 0.708; p < 0.001). MELD-LA had excellent discrimination for 30-day mortality (AUROC 0.82, 95% CI 0.76-0.88), better than MELD and peak-lactate (AUROC 0.78, 95% CI 0.71-0.84; AUROC 0.73, 95% CI 0.66-0.81). MELD-LA predicted 30-day mortality independently of age, sex, severity of liver disease and vasopressor support (HR 1.29 per 1-point-increase of MELD-LA; 95% CI 1.19-1.41; p < 0.001). Similarly, MELD-LA demonstrated excellent discrimination for 30-day mortality in the VA (AUROC = 0.86, 95% CI 0.79-0.93) and Baylor cohort (AUROC = 0.85, 95% CI 0.74-0.95). CONCLUSIONS: MELD-LA significantly improves discrimination of short-term mortality associated with variceal bleeding, compared to MELD, peak-lactate and nadir hemoglobin. Thus, MELD-LA might represent a useful and objective marker for risk assessment and therapeutic intervention in patients with variceal bleeding.