(249) Comparator Cohort Selection: a Review of Prospective Pregnancy Registries for Multiple Sclerosis (MS) Products

Background: Per the 2019 FDA draft guidance, pregnancy exposure registries should enroll at least one internal comparator cohort. Often, the FDA postmarketing requirement (PMR) will provide specifications for the study cohorts. To limit bias (e.g., confounding by indication) in pregnancy registries, the compared cohorts should align in terms of disease severity. At the same time, it may be appropriate to exclude patients exposed to other products in the same class (with the same mechanism of action) as the treatment effect of these products may mask a difference in risk between the cohorts.

Objectives: To compare and assess strengths/weaknesses of internal comparator cohorts for ongoing pregnancy registries for MS products (as an example), by reviewing publicly available information.

Methods: We identified 4 ongoing pregnancy registries with at least 1 internal comparator cohort for MS products: 2 S1P therapies, 1 anti-CD20 monoclonal antibody, and 1 fumarate. For these, the FDA PMR specified that the study population should include an exposed cohort and 2 comparator cohorts: 1 with MS and 1 without MS. We reviewed the internal comparator cohorts for each registry with regard to disease status.

Results: All registries had at least 1 MS cohort; 1 had 2 MS cohorts (1 with and 1 without exposure to other MS medications), and 2 had 1 healthy (non-MS) cohort. Registries without internal non-MS cohorts, used external rates from the literature, other studies, or population-based surveillance systems to put the registry rates into context. None of the registries mentioned disease severity (e.g., moderate/severe) enrollment criteria on their website or clinicaltrial.gov profile; however, all specified exclusionary exposures: 1 S1P registry excluded patients exposed to other S1P therapies; 1 S1P registry and the anti-CD20 registry excluded patients exposed to S1P therapies and anti-CD20 therapies; the fumarate registry excluded patients exposed to any disease-modifying therapies.

Conclusions: Based on publicly available information, these registries include internal comparator cohorts that may not align with their respective exposed cohorts in terms of disease severity, thus increasing the risk for confounding by indication and other sources of bias. Although all specified exclusionary exposures to medications in the same class, the allowable treatments across cohorts do not necessarily align in terms of indication/effectiveness.