P0624 : Acute viral hepatitis: Epidemiological change during the last 25 years in Spain

MELD score (AUC: 0.81) and better than MELD-NA (AUC: 0.78) and CP (AUC: 0.74) scores. In contrast, HMGB1 had no prognostic value (p =0.681), neither in cirrhotic nor in non-cirrhotic HBVACLF patients. In order to accurately predict the clinical prognosis of HBV-ACLF patients, we test the MELD-65 index [MELD-65 = 0.000372×M65-antigen (U/L) + 0.3432×MELD score − 11.17] established by logistic regression analysis with MELD and M65antigen remaining in the model. The new M65-based MELD score (AUC: 0.92, 95%CI 0.81 to 0.98, youden index: 0.71) provided significantly better predictive values in HBV-ACLF than MELD (AUC: 0.81, youden index: 0.6, P value vs MELD-65, P = 0.03), MELD-NA (AUC: 0.77, youden index: 0.45, P value vs MELD-65, P = 0.01) and CP (AUC: 74, youden index: 0.37, P value vs MELD-65, P = 0.02) scores. Conclusions: The MELD-65 index, a combination of MELD score and M65-antigen, has superior predictive accuracy to MELD, MELD-NA and CTP scores for the prognosis of HBV-ACLF patients from Asiapacific. Further prospective clinical studies are warranted to validate its role in predicting ACLF patients with other etiologies.