A novel approach for active surveillance of pregnancy safety in multiple sclerosis

Background: Pregnancy safety studies aim to reduce uncertainty around the risk of adverse outcomes associated with a new medication. The study size required for timely exclusion of the relative risk (RR) of various outcomes is of interest.

Objectives: To estimate the enrolment required to exclude specific levels of risk in a pregnant population over the first 3‐6 years of drug approval using a dual‐surveillance approach.

Methods: This case study is based on ocrelizumab (OCR), a newly approved drug for multiple sclerosis (MS), and two key outcomes: major congenital malformations and preterm births (baseline risk 3% and 10%, respectively). Two studies have been designed to assess these and other outcomes in women exposed to OCR: (1) a prospective pregnancy registry; (2) a multi‐database pregnancy outcomes study. Drug‐specific pregnancy registries can enrol patients soon after approval and provide detailed case information, but recruitment is generally slow. Existing health care databases accumulate large amounts of data, but limitations include time to data accrual and lack of detailed information on individual cases. We projected the expected number of exposed pregnancies in both studies and estimated the levels of risk this approach would be able to rule out over time, assuming a 1:3 exposed‐unexposed ratio in the multi‐database study (1:1 in the pregnancy registry study) and a true RR of 1.

Results: The OCR pregnancy registry will collect data in the United States and Germany. Approximately 8, 46, and 55 pregnancies with live births are expected to be enrolled at years 1, 5, and 6. These sample sizes provide 80% power to exclude RRs of 2880, 28, and 21 for major congenital malformations, and 67, 6, and 5 for preterm births. The OCR multi‐database study will use US health care claims databases and Danish nationwide registries. An estimated 300 exposed pregnancies will be captured at 5 years after approval. Assuming 215 live births with successful mother‐child linkage, the study will have 80% power to exclude RRs of 3.5 for major congenital malformations and 2 for preterm births.

Conclusions: Combined, the registry, which will provide early reassurance regarding substantial risks, and the multi‐database study, which will provide narrower safety margins after 5 years, will progressively reduce uncertainty regarding potential risks of adverse outcomes, overcome the limitations of each individual study design, and provide timely and meaningful safety information for women with MS and their physicians.