Abstract
Patients with compensated advanced chronic liver disease (cACLD) have a significant risk of decompensation. Therefore, this study aimed to evaluate the predictive value of dynamic liver stiffness measurements (LSM) for decompensation risk, and their performance across different clinically significant portal hypertension (CSPH) risk stratification. This retrospective cohort study included 1409 patients with cACLD. Patients were divided into no CSPH, probable CSPH, and certain CSPH groups. Competing risk regression analysis was used to identify the independent predictors. The receiver operating characteristic curve and time-dependent area under the curve were used to evaluate the predictive performance. During follow-up, liver decompensation incidence increased with CSPH severity (22.2% with no CSPH, 37.5% with probable CSPH, and 64.9% with certain CSPH, p < 0.001). Multivariate regression analysis identified age, basal LSM1, delta LSM/LSM1, delta LSM/delta year, spleen diameter, and international normalized ratio as independent risk factors for liver decompensation. In the no CSPH group, spleen diameter showed the best predictive ability (AUC = 0.710). For probable and certain CSPH groups, delta LSM/LSM1 showed superior predictive performance (AUC: 0.777 and 0.782, respectively). The predictive power of basal LSM1 was relatively limited across all groups (AUC: 0.554-0.639). Subgroup analysis revealed interactions between age, sex, different etiologies, and CSPH subgroups. The relative change rate of LSM outperformed basal LSM1 and annual change rate in predicting liver decompensation risk, particularly in patients with existing portal hypertension. Dynamic assessments and differentiated prediction strategies are essential for optimal patient managements.