(BSPOT32) Important Tool in our Rare Disease Toolbox: Hybrid Retrospective-Prospective Natural History Studies Serve Well as External Comparators for Rare Disease Studies

Background: Real-world evidence (RWE) from Natural History of Disease studies can well-support regulatory approval, including as external comparator arms for single-arm trials. These are typically either prospective or retrospective; less common are hybrid external comparator cohorts, which combine retrospective and prospective data. These can be a tool in creating fit-for-purpose RWD by taking advantage of strengths across different datasets; however, differences in data availability, timing, and quality between sources present methodological challenges.

Objectives: To describe design considerations for hybrid Natural History of Disease studies intended for use as external comparators in rare disease clinical development programs.

Methods: We present two broad categories of hybrid Natural History of Disease studies: 1) “mixed patient”, and 2) “same patient”/ ambispective designs. Both collect retrospective and prospective data on different or same set(s) of patients respectively, combined in a single study. We detail applicability, strengths, and considerations for operational application.

Results: Hybrid design choice depends on disease dynamics (e.g., fatality, evolution of variants), patient heterogeneity, and practical factors inherent to the clinical development program (e.g., anticipated truncation of patient follow-up). Mixed patient designs quicken patient accrual and increase sample size. Same-patient designs capture key aspects of disease course, lengthen available patient follow-up, and can shorten prospective follow-up. For instance, Nulibry®, approved for a rare cofactor deficiency indication, employed a mixed patient design combining retrospective data for 37 patients (17 living, 20 deceased) and prospective data for 14 of those 17; 18 of the 37 were genotype-matched to single-arm treated patients in the external cohort. This design enabled observing the 1st year of life for all 37 patients, including those who did not survive to year 1, possibly representing a unique phenotype/severity. A hybrid external comparator framework would require regulatory-grade feasibility review of each data source, specialized study design (e.g., nuances of time zero definition, availability of prognostic variables, standardized endpoint assessments across sources), and analytical considerations (e.g., missingness and unmeasured confounding differences in the retrospective and prospective periods/patients).

Conclusions: Hybrid Natural History of Disease studies allow for more robust contextual information on natural course of disease, improving suitability of the data for an external comparator. As adoption increases, investigators will benefit from regulatory guidance on acceptable scenarios for these designs.