[Characteristics and predictive factors of recompensation in patients with hepatitis B-related decompensated cirrhosis: a retrospective cohort study].

Abstract

Objective: To investigate and evaluate the incidence rate, time distribution, related factors, and prediction model discriminative ability for recompensation in patients with hepatitis B-related decompensated cirrhosis. Methods: Four hundred and five patients hospitalized for the first time due to hepatitis B-related decompensated cirrhosis were retrospectively enrolled, using the date of initial hospitalization for decompensation as the follow-up initial point. Recompensation was defined as the resolution condition of decompensation complications, lasting for at least 12 months without specific maintenance therapy, with reference to the Baveno Ⅶ consensus in combination with real-world clinical practice. A statistical analysis was performed on the recompensation occurrence time in patients who achieved recompensation. Patients followed up for≥12 months (n=382) were included, with 12 months as the cutoff. Univariate and multivariate logistic regression analyses were used to analyze related factors, with recompensation (primary event time of occurrence >12 months) as the endpoint. A recompensation prediction model was constructed based on key variables. The discriminative ability was evaluated using the area under the receiver operating characteristic curve (AUC), and tertile risk stratification was performed. The baseline and recompensated indicators were compared pairwise for patients with recompensation. Results: Recompensation occurred in 94 out of 405 cases, with an overall incidence of 23.2%. The recompensation incidence was 22.7% (92/405) when "time to recompensation >12 months" was used as the primary event criterion. The median time to recompensation was 22.9 (16.0, 35.4) months, mainly concentrated between 12 and <24 months. Multivariate logistic regression showed that persistent virological suppression was independently associated with recompensation (OR=2.387, 95%CI 1.395-4.082, P=0.001). The independent adverse factors for recompensation were initial decompensated complications ≥2(OR=0.525, 95%CI 0.288-0.957, P=0.036) and portal vein thrombosis (OR=0.296, 95%CI 0.108-0.806, P=0.017). There was no statistically significant difference in the composition of the initial decompensation complication types, but the complication burden was closely related to recompensation, with a significantly lower recompensation rate in patients with ≥2 complications [18.2% (25/137) vs. 28.2% (69/245), P=0.031]. The model for end-stage liver disease (MELD), complication burden, portal vein thrombosis, sustained virological suppression, and non-selective β receptor blockers had better discrimination in Model B than Model A (AUC 0.670 vs. 0.532). The actual recompensation rates following tripartite risk stratification were 15.4% (18/117), 19.7% (23/117), and 35.9% (42/117), respectively. Alanine aminotransferase (ALT) and aspartate aminotransferase (AST) levels and MELD and Maddrey discriminant function (MDF) scores were decreased, while albumin levels were increased, and the infection rate decreased from 25.5% (24/94) to 12.8% (12/94) (P<0.05) at the time of recompensation assessment. Conclusion: Recompensation is not uncommon in patients with hepatitis B-related decompensated cirrhosis, and it often occurs more than 12 months after the initial decompensation. Sustained virologic suppression is an independent favorable factor for recompensation, while a severe complication burden and portal vein thrombosis suggest a lower likelihood of recompensation. The predictive models have a certain discriminative ability and can be used for risk stratification and as a reference for follow-up and comprehensive management based on key clinical variables.