Presentation (CTI)

OBJECTIVES: Clinical trials of etranacogene dezaparvovec demonstrated increases in factor IX (FIX) activity levels in adult patients with hemophilia B (PWHB). However, the long-term durability of gene therapy still relies upon model-based prediction. Consequently, we developed a Bayesian linear mixed model to predict mean FIX activity levels and tested the validity of the exponential decay assumption.
METHODS: A combined dataset of 55 responding PWHB from Phase 2b (2.5 years) and Phase 3 (2 years) studies of etranacogene dezaparvovec were used for developing the model. The predicted FIX activity levels post-infusion were compared to 5-year observations. The model was then re-estimated incorporating the 5-year data and compared to the actual observation.
RESULTS: Mean predicted FIX levels at 5 years post infusion (33.5%) were lower than observed (36.7%), but within the 95% credible interval [6.5, 104.3]. When 5-year observed data was included, the predicted mean FIX level at 5 years post infusion still underestimated observed results (32.4% predicted vs 36.7% observed).
CONCLUSIONS: Although the exponential decay assumption appeared to align with the 2-year HOPE-B FIX trend, it may not have fully captured the plateau observed in longer-term (5-year) results. Including the 5-year data did not improve predictions suggesting the exponential decay model doesn’t accurately represent the observed FIX trajectory and may underestimate the gene therapy durability. Using non-linear modelling methods which reflect the more recent clinical data may increase confidence in the durability of gene therapy.