(325) Use of population-based real world data sources and disease specific registries to conduct post approval pregnancy safety studies for a biologic indicated for a complex, uncommon indication - challenges and opportunities

Background: Anifrolumab is a human monoclonal antibody that inhibits type I IFN, indicated for treatment of SLE in patients with moderate/severe disease despite standard therapy. SLE predominantly affects females of child-bearing potential. There is a need to characterize safety of anifrolumab use in pregnancy. The intravenous (IV) mode of administration and uncommon nature of the indication present challenges in the use of real-world data (RWD) to assess safety in pregnancy.

Objectives: To evaluate the ability of RWD sources to generate evidence on safety of anifrolumab use in pregnancy using the EMA data quality framework (DQF).

Methods: An extensive review of 46 data sources in 8 countries, including 2 SLE specific international registries and 1 rheumatology pregnancy registry was conducted to determine “fit for purpose”, using 13 criteria mapped to EMA DQF for regulatory decision making: extensiveness, coherence, timelines, relevance, and reliability. Of which, relevance (ability to capture anifrolumab use, availability of pregnancy outcomes and mother-offspring linkage) was prioritized. Additional dimensions specific to the nature of the disease and uncommon disease prevalence, outside of the EMA DQF were included.

Results: RWD from 5 countries and 1 rheumatology pregnancy registry were selected. The greatest challenge in identifying fit for purpose data was the inability to capture anifrolumab use (given the IV administration route, initial prescription in a specialist care setting, and subsequent prescriptions in primary care setting in some data sources, hence requiring a data source that captures both primary and specialist care), with 50% of the selected data sources completely capturing exposure while the other 50% had some limitations. 25% (2/8) of the countries evaluated had no data sources that partially or fully captured anifrolumab exposure. Other reasons for exclusion of data sources included absence of outcome (100% for the international registries) and other required variables to meet the study objectives. None of the SLE specific registries met the EMA data quality framework for selection. However, the rheumatology pregnancy registry contained sufficient information to be selected.

Conclusions: This feasibility assessment highlights successes in the use of the EMA DQF for database selection and challenges of identifying suitable RWD sources to address safety endpoints in pregnancy in the setting of biologics administered in specialist settings and for relatively uncommon indications. Despite increasing recommendations to consider disease specific registries for pregnancy studies, they are often not fit for purpose for pregnancy studies.