This past week, the FDA gave a webcast describing some of the issues that they are seeing with the SEND packages for safety pharmacology studies. The agency informed the industry that they are currently receiving around 200 such studies per year and were facing certain difficulties when using the SEND datasets for analysis for some studies. The presentation then focused on a deep dive into 24 such studies which represented a good cross section of the submissions the agency receives.
In a few cases the issues were due to the SEND standard not being followed, but mostly the issues were a result of inconsistencies in how different organizations had implemented SEND. In one case there were even inconsistencies from domain to domain within the same study!
This is a common issue with the standard and something I’ve often discussed previously. We have a very flexible standard. Such flexibility allows data collected in a variety of ways to be represented in SEND, but at the expense of the person or organization receiving the data not really knowing what they are going to get.
In the presentation, it was stated that the agency uses 2 different tools for reviewing the study data and both have issues with the way that data may be presented in SEND. These tools have both tabular and graphic representations of summary results, yet each tool uses a different method of identifying dose level and the timepoint relative to dose when summarizing the findings.
The webcast stated that for some studies, either tool could be used, but for some, only one could be used, and there were studies where neither tool could be used, such is the variability permitted by SEND. This situation is mostly a result of the flexibility of the timing information. Firstly, inconsistencies in how to describe the various time periods and dose levels within a Latin Square study, and then variability in how to describe the timing of an individual finding so that it can be grouped in the correct dose level. While the agency is not currently specifying any recommended best practices to overcome this issue, they are asking for more information to be included in the nSDRG (non-clinical study data reviewer’s guide), in order to help them understand how the data are rendered and how the data can be best summarized for analysis.
I would imagine that this inconsistency in being able to consume the SEND data must be very frustrating, and I can see a time coming when the TCG (FDA’s study data technical conformance guide) includes directions to remove such variability and the agency specify their preference for how such studies and data should be rendered. They also could add business rules to check that such instructions are followed. I can also see pressure coming on CDISC to tighten the standard and not allow such flexibility, and this would not be a bad thing in my opinion.
If you are working with SEND for these particular studies, and you did not attend the webcast, I’d highly recommend availing yourself of the materials here: November 16, 2023 Webinar: Common Issues with SEND Data Submitted for Safety Pharmacology Studies.
Did you attend the webcast? What insight did it provide? If you’d like to share your thoughts with me mail me at [email protected]
‘til next time
Marc
