FDA analysis of MRD data in hematologic malignancy applications.

2541 Background: There is considerable interest in the use of minimal residual disease (MRD) in clinical trials of hematologic malignancies, especially as a potential surrogate endpoint to expedite drug approval. Although surrogacy has not yet been established in most hematologic malignancies (except CML), MRD data has been submitted in applications to the agency to allow for further characterization of the product’s activity. Here we describe the new drug applications (NDA) or biologics licensing applications (BLA) that included MRD data and provide an analysis of the MRD data and the Agency’s decisions. Methods: The authors reviewed internal databases for all original and supplemental NDA and BLA applications submitted to the Division of Hematology Products (DHP) between 2014 and 2016 that pertained to malignant hematologic conditions. The applications were reviewed for inclusion of MRD data. The clinical reviews and prescribing information (PI) of the identified applications were reviewed for assessment of MRD data and FDA’s findings. Results: There were 34 NDAs or BLAs submitted between 2014-2016. Of these, 13 (38%) included MRD data, in the following diseases: CML, CLL, ALL, and MM. MRD data was added to the PI in 6 cases (46%), not included in 4 (31%), and was not proposed for inclusion in 3 (23%). Among the 6 cases in which MRD data was included in the PI, 5 used PCR testing and 1 used flow cytometry. Among the 4 in which MRD data was not included in the PI, the identified issues included: missing data among those in CR, incomplete test performance characteristics data (e.g.-limit of detection), disparate sample sources (blood, bone marrow), high amount of test failure rates (i.e.-inability to identify a clonal rearrangement), lack of test validation in the proposed disease setting, and inappropriate planned statistical analyses. Conclusions: Nearly 40% of applications submitted to DHP between 2014 and 2016 included MRD data. While the data submitted was deemed adequate for inclusion in the PI in 46% of cases, 31% of applications contained MRD data that the Agency deemed un-interpretable. Data collection and assay performance characteristics should be of significant rigor and completeness to allow for comprehensive review.