How Does the Source of Transition Probabilities Influence Natural History Simulation and Economic Evaluation in Metabolic Dysfunction-Associated Steatotic Liver Disease (MASLD)?
Author(s)
Victoria Zaitceva, MSc1, Mehdi Javanbakht, PhD1, Amir Ansaripour, PhD2.
1HEOR, Optimax Access, Southampton, United Kingdom, 2HEOR, Optimax Access, Rotterdam, Netherlands.
1HEOR, Optimax Access, Southampton, United Kingdom, 2HEOR, Optimax Access, Rotterdam, Netherlands.
OBJECTIVES: MASLD is a progressive disease that demands rigorous economic modeling, especially as treatment options continue to evolve. The transition probabilities (TPs) used to simulate disease progression can vary considerably depending on the data source, whether from randomized controlled trials (RCTs) or real-world evidence (RWE), and this variation may influence modeling outcomes. This study examined how TP source influences projected progression to cirrhosis (F4).
METHODS: A patient-level simulation model was developed in R to estimate time to cirrhosis (F4) using annual TPs for fibrosis stages F0-F3. TPs were sourced from a published meta-analysis (Le et al. 2023), reported separately for placebo groups in RCTs and observational studies. A synthetic patient cohort was generated using baseline characteristics from a published real-world study (Angulo et al. 2015). Patients were simulated using annual cycles under two scenarios: one applying RCT-derived TPs and the other using RWE-based TPs. The primary outcome was cirrhosis-free survival, stratified by data source and baseline fibrosis stage.
RESULTS: We simulated 10,000 patients with baseline fibrosis stages (F0:53.6%, F1:23.5%, F2:14.0%, F3:8.9%), a mean age of 48 years, and 37.5% with type 2 diabetes. Cirrhosis-free survival varied significantly by data source. RWE-based modeling predicted longer cirrhosis-free survival (28.7 years) compared to RCT-derived inputs (25.4 years). When stratified by baseline fibrosis stage, estimated life-years free of cirrhosis in the RWE model were: F0:29.77, F1:27.77, F2:25.45, F3:24.15; while RCT-based estimates were lower: F0:25.67, F1:25.05, F2:24.02, F3:23.71.
CONCLUSIONS: The choice of transition probabilities clearly influences fibrosis modeling in MASLD. We observed that real-world data led to more favorable projections than RCTs, especially in early-stage fibrosis, likely reflecting differences in diagnostic approaches, such as biopsy use in RCTs vs. non-invasive testing in routine care. These results underscore the importance of thoughtful data source selection and transparent modeling assumptions to support meaningful health economic evaluations.
METHODS: A patient-level simulation model was developed in R to estimate time to cirrhosis (F4) using annual TPs for fibrosis stages F0-F3. TPs were sourced from a published meta-analysis (Le et al. 2023), reported separately for placebo groups in RCTs and observational studies. A synthetic patient cohort was generated using baseline characteristics from a published real-world study (Angulo et al. 2015). Patients were simulated using annual cycles under two scenarios: one applying RCT-derived TPs and the other using RWE-based TPs. The primary outcome was cirrhosis-free survival, stratified by data source and baseline fibrosis stage.
RESULTS: We simulated 10,000 patients with baseline fibrosis stages (F0:53.6%, F1:23.5%, F2:14.0%, F3:8.9%), a mean age of 48 years, and 37.5% with type 2 diabetes. Cirrhosis-free survival varied significantly by data source. RWE-based modeling predicted longer cirrhosis-free survival (28.7 years) compared to RCT-derived inputs (25.4 years). When stratified by baseline fibrosis stage, estimated life-years free of cirrhosis in the RWE model were: F0:29.77, F1:27.77, F2:25.45, F3:24.15; while RCT-based estimates were lower: F0:25.67, F1:25.05, F2:24.02, F3:23.71.
CONCLUSIONS: The choice of transition probabilities clearly influences fibrosis modeling in MASLD. We observed that real-world data led to more favorable projections than RCTs, especially in early-stage fibrosis, likely reflecting differences in diagnostic approaches, such as biopsy use in RCTs vs. non-invasive testing in routine care. These results underscore the importance of thoughtful data source selection and transparent modeling assumptions to support meaningful health economic evaluations.
Conference/Value in Health Info
2025-11, ISPOR Europe 2025, Glasgow, Scotland
Value in Health, Volume 28, Issue S2
Code
MSR121
Topic
Economic Evaluation, Methodological & Statistical Research, Study Approaches
Disease
Diabetes/Endocrine/Metabolic Disorders (including obesity)