The natural history and individualized prediction of liver stiffness-based fibrosis risk in metabolic...

Area of Science:

• Hepatology
• Data Science in Medicine
• Predictive Modeling

Background:

• Liver stiffness measurement (LSM) is crucial for risk stratification in MASLD.
• Static LSM thresholds do not fully capture dynamic risk transitions.
• A need exists for time-updated, individualized risk prediction models.

Purpose of the Study:

• To characterize LSM-defined risk transitions in MASLD.
• To develop a dynamic, time-updated model for predicting state transitions and outcomes.
• To generate individualized risk trajectories for liver-related events and death (LREs/death).

Main Methods:

• Applied a multi-state, time-homogeneous Markov model to a real-world MASLD cohort (n=11,514).
• Quantified annual transition probabilities and mean state occupancy times across LSM risk strata.
• Developed a dynamic, multi-state Markov model (DYNAMO) incorporating covariates like age, sex, T2D, and hypertension.

Main Results:

• The low-risk LSM stratum showed high stability (92% unchanged at 1 year, mean occupancy 8.43 years).
• The intermediate-risk LSM stratum was highly dynamic (39% unchanged at 1 year, mean occupancy 0.92 years).
• Type 2 diabetes, hypertension, and obesity shortened low-risk occupancy; antidiabetic medication improved transitions.

Conclusions:

• LSM-based risk strata in MASLD exhibit dynamic trajectories, especially the intermediate-risk state.
• Frequent reassessment of the intermediate-risk state is warranted.
• The DYNAMO model offers personalized surveillance intervals and risk-adapted management strategies.