FDA Guidance on External Control Arms: A Checklist For Drug Developers

The FDA has released new draft guidance on the use of external control arms (ECAs) providing insights into the agency’s thinking. This is a substantively positive intervention providing an opportunity for the drug development ecosystem to better coalesce around standards for these studies.

While reemphasising the effective trinity of ECAs; early scientific engagement, robust justification and magnitude of effect size, this guidance goes further by highlighting the importance of study design & conduct.

Specifically, the guidance calls out:

1. The need to pre-specify an external control arm study protocol and SAP as opposed to selecting an ECA after completion of a single arm trial.
   
   
2. The need for a thorough feasibility assessment including the ability for a data source to enable accurate inclusion criteria application, capture important prognostic variables and for appropriate comparable endpoints of interest to be derived. I would suggest sections in study documents be added that address methods and analysis showing the above to be satisfactory. 
   
   
3. The need to for sponsors to find an external data source that fits the characteristics of the intended study as opposed to the trying to shoe-horn a study into a sub-optimal data source. 
   
   
4. The need for sponsors to have a principled approach to outcomes assessment including treatment status blinded outcome review. Outcomes will need to be consistently analysed across both arms for the study to be credible.
   
   
5. The recognition that both historical trial data (HTD) and RWD can serve as ECAs with HTD having some advantages over RWD.
   
   
6. The fact an RWD dataset exists for a given disease doesn’t automatically mean its suitable for an ECA; aspects of prognostic factors, temporality, missingness, comparability etc must be assessed.

In all, the agency raises 10 key areas that must be considered when determining data comparability for an ECA. These are:
‍

a. Time periods both in relation to the intervention as well as observation time and time zero.

b. Accounting for the impact of geographical aspects including differential SoC as well as patient distribution.

c. Accounting for differential aspects of how a diagnosis is made and any changes that may have occurred in diagnostic criteria.

d. Ensuring the prognostic factors of both the ECA and intervention arm are sufficiently similar.

e. Ensuring the treatment in the ECA is comparable to the interventional arm in the context of timing, dose, RoA etc with specific focus on the impact of dose modification or interruption.

f. Ensuring treatment related aspects such as line of therapy, concomitant medications and predictive biomarkers (if relevant) are similarly distributed across both study arms.

g. Ensuring time 0 is the same and comparable between both study arms.

h. The relevance of intercurrent events should be assessed across both treatment arms.

i. Ensuring the endpoints are consistently able to be measured across both arms and their method of assessment is the same.

j. The impact of missing data should be assessed and accounted for in a principled manner.

k. The requirement that IPD & source documentation be readily available for both submissions and inspections.

While this guidance appears to be setting the foundations for a regulatory ECA framework it would have been nice to see the FDA thinking on hybrid control arms (intervention + dynamic/static borrowing for supplementation) however this was determined to be out of scope.

References:

1. Paper on selecting time zero for ECA studies (great work by Anthony Hatswell et al) https://journals.sagepub.com/doi/full/10.1177/0272989X221096070
2. Link to full guidance https://www.fda.gov/regulatory-information/search-fda-guidance-documents/considerations-design-and-conduct-externally-controlled-trials-drug-and-biological-products