Validation of a novel ordinal endpoint for an adaptive platform trial in paediatric haematopoietic stem cell transplant

BackgroundHaematopoietic stem cell transplant (HCT) is a curative therapy for various paediatric conditions but is associated with significant morbidity and mortality, particularly in children requiring intensive care, facing delayed immune reconstitution or prolonged viral reactivation. Due to the rarity and heterogeneity of paediatric HCT, traditional randomised controlled trials are challenging. Adaptive platform trials (APTs), which evaluate multiple interventions across multiple subgroups, offer a solution, but typically rely on a shared short-term primary outcome that is relevant to clinicians, patients and families and can be used for all interventions/subgroups. No such outcome currently exists in paediatric HCT. In this article we propose and validate four novel ordinal outcomes to assess HCT-related morbidity and mortality within the first 100 days post-transplant for use in An adaptive platform trial Designed to Improve the COmplications, cost-effectiveness and health Outcomes for children receiving a stem cell Transplant (BANDICOOT) APT, designed. MethodsThe proposed outcomes were validated using real-world data from n=202 paediatric patients who underwent allogeneic HCT. The validation process included examining the distribution of patients across outcome categories, assessing the association with key long-term outcomes post HCT, and evaluating whether exposures with known efficacy had the expected associations with the proposed endpoints and whether the proportional odds assumption used in the analysis is likely to be reasonable. We also sought feedback on the outcomes from clinicians and family representatives. ResultsThe results showed strong associations between each ordinal endpoint and long-term HCT complications, including relapse, chronic-graft-versus-host disease, and death. Associations with key exposures (e.g. donor type and positive minimal residual disease pre HCT) were mostly in the expected direction. Moreover, expert feedback from clinicians and family representatives indicates that one of the proposed endpoints, which incorporates viral-related patient states and single/multi-organ support days, was both feasible and relevant for use in BANDICOOT. ConclusionsThe selected ordinal endpoint provides a robust and clinically applicable framework for evaluating interventions in paediatric HCT that offers broad applicability across various HCT outcomes.