Presentation (CTI)

OBJECTIVES: Advanced therapy medicinal products (ATMPs) pose a persistent challenge for health technology assessment (HTA). Launch decisions often rely on small single-arm trials with surrogate endpoints, creating substantial uncertainty around long-term clinical and cost-effectiveness outcomes in the context of high treatment costs. As real-world evidence (RWE) accumulates after launch, decision makers need approaches that can formally update these early assessments. Bayesian methods offer a natural framework for this purpose, yet remain rarely implemented in practice. We demonstrate a practical Bayesian framework for life-cycle cost-effectiveness analysis of ATMPs, using onasemnogene abeparvovec for spinal muscular atrophy (SMA) type I as a case study.
METHODS: We reformulate a validated frequentist microsimulation model (Broekhoff et al., 2021), comparing onasemnogene abeparvovec, nusinersen, and best supportive care in the Netherlands, within a Bayesian framework. The same clinical trial evidence that informed the original cost-effectiveness analysis is represented as prior distributions. These priors are subsequently updated with aggregate RWE from the newborn-screened RESTORE subgroup with two SMN2 copies. Power priors with varying discount parameters are used to regulate the contribution of registry evidence to posterior estimates. The resulting posteriors yield incremental costs, QALYs, ICERs, and cost-effectiveness acceptability curves.
RESULTS: Analyses are ongoing. The study evaluates the impact of incorporating RWE on cost-effectiveness estimates and decision uncertainty. It also examines key methodological challenges associated with evidence updating. Key challenges include asymmetric RWE availability across comparators, population heterogeneity between trial and registry populations, and the limited follow-up available in registry data. These challenges are characterised systematically.
CONCLUSIONS: A Bayesian reformulation provides a practical framework for updating cost-effectiveness estimates as RWE accumulates across the product life cycle, supporting more adaptive HTA decision making for ATMPs.