(Re)Integration Dilemma: Integrated Summaries of Safety and Effectiveness
As promised in my last post prior to PHUSE-EU Connect, I’d like to now share some reflections on my “Integration Dilemma” speech, in which I discussed integrating data from different studies for supporting, for example, integrated summaries of safety and effectiveness.
An Integrated Summary of Safety (ISS) is required by the U.S. Food and Drug Administration (FDA) as a critical component in any New Drug Application (NDA) or similar market approval request. An Integrated Summary of Effectiveness (ISE) might be also required under certain circumstances, although most of the time, efficacy results from an individual pivotal study, or multiple studies, might be sufficient. This is not simply a summary but rather a detailed integrated analysis of all relevant data from the clinical study reports with the aim of providing a more transparent understanding of responses across different populations (demographics, disease-related, etc.) and dosing regimens. By performing an ISS, the sponsor could provide more robust evidence for safety
Both the ISS and ISE allow reviewers to easily compare individual outcomes, tracking a subject’s results across the entire clinical development spectrum, facilitating broad views of the investigational product’s overall efficacy and safety profile. With the ISS and ISE, a “single database” in ADaM (iADAM) is formed by pooling the results of all concerned clinical trials, from which ISS/ISE analyses are generated. These are analysis outputs that will be used for creating reports providing a summary of safety and, eventually, efficacy.
The question is: what should be the source of iADaM? As discussed in the white paper and presentation, there are three options, and in all scenarios, you have the option to combine both legacy and standard datasets:
- SDTM(s) iADaM: ADAM integration using only individual studies SDTM datasets
- ADaM(s) iADaM: ADAM integration using only individual studies ADAM datasets
- SDTM(s) iSDTM iADaM: SDTM and ADAM integration
Regardless of which option you adopt for your next ISS/ISE, traceability and proper documentation are crucial. It is also important to plan for an early discussion of your integration strategy with your relevant regulatory agency. This can be done for example by sharing as soon as possible your Study Data Standardization Plan. In my paper, I also proposed some examples of possible wording and the level of technical details to use to explain to the reviewer your planned data integration approach. For example, how you plan to handle subjects participating in more than one trial, medical dictionaries up-versioning such as MedDRA, terminology alignment, CDISC conformance, and so on, should be clearly outlined.
The option to adopt depends on a number of different factors such as the status of the available study data packages and their conformance, their variability, and, in some cases, on sponsor preference (or I would say “prejudice”). With most of the sponsors, it was often the case that they would change their minds, and sometimes opt for very general and vague ISS SAP definitions. This, in addition to timelines and updates to the ongoing individual study packages, given that Cytel was not appointed for all ongoing studies, were a common denominator of all the experiences we had. However, I believe this is normal due to the nature of the project, the ISS/ISE, where clinical drug development is still ongoing, and new “inputs” that might come from ongoing studies/research. As such, in my opinion, sometimes ISS/ISE are extremely exploratory, meaning that approaches and required analyses could change from time to time.
In general, the option to reuse original ADaMs from individual study CSRs (option 2 above) can be considered the best choice. However, our experience showed that this might not always be the right option even when all concerned studies made use of CDISC standards for both source (SDTM) and analysis datasets (ADaM), unless the sponsor is able to make an appropriate surveillance of CROs’ work if studies are outsourced.
So bottom line, there is not just one approach to take and, as demonstrated by our experiences with submission projects at Cytel, the decision really depends on the data standardization status of the studies you plan to integrate. For example, looking back at the last ten major FDA submissions we did at Cytel in the last three years, one of which is ongoing, we did apply the data integration options shown in the following table:
|Data Integration Option||N|
|Option 1 / Individual Study SDTMs to iADAM||4|
|Option 2 / Individual Study ADaMs to iADAM||2|
|Option 3 / Individual Study SDTMs to iSDTM to iADaM||4|
We also spent a lot of time with documentation, for example, providing a lot of details on changes we did to original study datasets, and we did that in the integrated reviewer guide.
The PHUSE initiative and the white paper have been a great help, but there are still some gaps. For ADaM, for example, some of the current gaps were addressed in one proposed guidance, back in 2019, which was including a set of new ADaM standards (classes), but since then we do not have any news of that project. Also, PHUSE has an ongoing project to develop a template for integrated analysis reviewer guide, for which a public review was completed last April. This template could also, if it became standard, at least clarify some topics and the level of details we should provide in such a document.
If you are interested in knowing more, watch the test recording of the speech I did before the conference:
The full paper, which includes more details on our experiences at Cytel, will soon be available on the PHUSE website. You can also read my prior post on this topic, which coincided with the PHUSE white paper release and subsequent webinar, paneled by, among others, a couple of FDA biostatisticians.
Lastly, I want to acknowledge the PHUSE office and committee for another very successful, and fully booked, event. I have to say, both the agenda and the conference organization were excellent!
About Angelo Tinazzi
Angelo Tinazzi is Senior Director, Statistical Programming, Clinical Data Standards and Clinical Data Submission at Cytel. He is a well-published and recognized expert in statistical programming with over 20 years' experience in clinical research. The application of CDISC standards in different therapeutic areas is part of his core expertise since 2003 in particular in the context of data submission to health authorities such as the FDA and PMDA.
Angelo is an authorized CDISC instructor and member of the CDISC ADaM Team as well as the CDISC European Committee where he also manages the Italian-speaking CDISC User Network.