Bioequivalence and Bioavailability Forum

HI BEQool,

❝ My question is: has anyone ever submitted a DDI study where the acceptance criterion was that only the point estimate (not the confidence interval) needed to fall within a predefined no-effect boundary, and that this boundary was wider than the conventional 80-125% range? If so, was this approach accepted by the regulatory agencies? Did the submission include any additional justifications, such as PK/PD relationship?

I don’t have a perfect example where the CI was completely ignored, but the closest real-life case I’ve seen is the filgotinib + statin cocktail study

They used prespecified no-effect boundaries of 70–143%, but for rosuvastatin the point estimate came out. Nobody batted an eye — the paper and later the EMA assessment simply concluded there was no clinically meaningful interaction and co-administration is fine without any dose adjustment.
You can see the real-world follow-through in the pooled FINCH safety analysis with concomitant statin use and the EMA EPAR for Jyseleca basically copied that conclusion straight into the SmPC
So it’s not the pure PE only unicorn you’re looking for, but it’s a regulatory-accepted situation where the boundary was obviously wider than 80–125%, the PEs slightly breached the prespecified limit, and the agency still said “no issue” purely because of the solid PK/PD reasoning.