AUTHORS
Pengling Sun, PhD, Bryan McComb, PhD2, Yangmei Zhou, PhD3, Jasmine Healy, PhD4, Delphine Agathon, PhD5, Travis Gould, PhD6, Wenquan Wang, PhD7, Bo Huang, PhD8
AFFILIATIONS
1 Pfizer Inc., Cambridge, Massachusetts, United States
2 Pfizer Inc., New York, New York, United States
3 Pfizer Inc., Shanghai, China
4 Pfizer Canada ULC, Kirkland, Canada
5 Pfizer Inc., Paris, France
6 Pfizer Inc., New York, New York, United States
7 Pfizer Inc., Collegeville, Pennsylvania, United States
8 Pfizer Inc., New York, New York, United States
ABSTRACT
Background: Marstacimab is a monoclonal antibody approved for prophylactic use in adults and adolescents with Hemophilia A (HA) or B (HB) without inhibitors. In June 2026, approval was expanded to adults and adolescents with inhibitors and pediatric patients aged 6 to 11 years in the US. The BASIS KIDS study (NCT05611801) evaluated the safety and efficacy of marstacimab in pediatric participants aged ≥1 to <18 years. Here, we describe a prospectively planned Bayesian extrapolation analysis that supplemented the interim analysis (based on May 2025 data cutoff) to support an accelerated pediatric label expansion from 6 years old, before the estimated completion of the study in 2028.
Methods: Efficacy was assessed in participants who received marstacimab at least 12 months before the data cutoff date. The primary endpoint was annualized bleeding rate (ABR) for treated bleeds compared to historical standard treatment: on-demand [OD] bypassing therapy for patients with inhibitors or prophylactic factor replacement therapy [RP] for patients without inhibitors. To support regulatory evaluation for indication expansion, a Bayesian approach employing robust mixture prior was performed to dynamically borrow information from the completed adult/adolescent BASIS study (NCT03938792). The Bayesian mixture prior was a weighted combination of a skeptical component for the pediatric population and an informative component for the BASIS study with a pre-specified weight (ω) of 0.5. The pre-specified success criteria were i) inhibitor cohort (superiority): upper bound of the 95% credible interval (CrI) of the ABR ratio estimate (marstacimab prophylaxis/OD treatment) < 0.5; ii) non-inhibitor cohort (non-inferiority), upper bound of the 95%CrI of the mean ABR difference (marstacimab prophylaxis – RP) < 2.5. Descriptive summaries were provided for Inhibitor with RP patients for supportive evidence without Bayesian analysis. Posterior probability of efficacy was reported and the extent of borrowing was quantified using the expected local information ratio (ELIR)-based effective sample size (ESS). Tipping point analyses evaluated robustness of results: i) to the degree of borrowing by varying prior weight and ii) to the choice of decision threshold by progressively increasing the stringency of the success criterion from 95% to 99.99%.
Results: 57 male patients (Inhibitor with OD: n=14; Inhibitor with RP: n=7; Non-Inhibitor with RP: n=36) were evaluated for efficacy and the mean age was 11 years (range 6 to 18); 59.6% of patients aged ≥6 to <12 years and 40.4% aged ≥12 to <18 years. Most participants were White (35.1%) or Asian (49.1%). The patient population included 35 with HA (10 with inhibitors; 25 without inhibitors) and 22 with HB (11 with inhibitors; 11 without inhibitors).
With a pre-specified mixture prior with prior weight of 50% from the BASIS study, the posterior estimated ABR difference for Non-Inhibitor cohort with RP was -1.94 (95%CrI: -3.47, -0.34), meeting the non-inferiority criterion. Posterior probability of efficacy exceeded 99.99% and the corresponding ESS was 5. For Inhibitor cohort with OD, the posterior estimated ABR ratio was 0.07 (95%CrI: 0.05, 0.12), representing a 93% reduction in ABR relative to historical OD treatment, satisfying the superiority criterion. The posterior probability of efficacy exceeded 99.99% and the associated ESS was 13. Tipping point analyses demonstrated that conclusions were preserved even with zero borrowing, confirming the robustness of the Bayesian inference. Furthermore, conclusions remained unchanged under the most stringent success threshold (99.99% CrI). Results from ≥6 to <12 years group were consistent with ≥6 to <18 years group, demonstrating superiority against on-demand bypass agent therapy and non-inferiority against prophylaxis factor replacement therapy.
Conclusion: Prospective Bayesian pediatric extrapolation complemented the interim analysis of the BASIS KIDS study. It demonstrated clinically meaningful treatment effects with controlled, dynamic borrowing, and prespecified decision criteria, providing a mathematically rigorous framework that successfully supported the accelerated regulatory approval of marstacimab in the US by approximately 3 years in pediatric patients with hemophilia A and B aged 6 to <12 years with a high unmet need.
DISCLOSURES
Pengling Sun, PhD: Pfizer, Inc. - Current equity holder in publicly-traded company; Pfizer, Inc. - Current Employment