Article
An adaptive design for early drug development including interim decision for single-arm trial with external controls or randomized trial
Search Medline for
Authors
-
Heiko Götte
- Merck Healthcare KGaA, Darmstadt, Germany
-
Marietta Kirchner
- Institute of Medical Biometry and Informatics, University of Heidelberg, Heidelberg, Germany
-
Johannes Krisam
- Institute of Medical Biometry and Informatics, University of Heidelberg, Heidelberg, Germany
-
Arthur Allignol
- Merck Healthcare KGaA, Darmstadt, Germany
-
Francois-Xavier Lamy
- Merck Healthcare KGaA, Darmstadt, Germany
-
Armin Schüler
- Merck Healthcare KGaA, Darmstadt, Germany
-
Meinhard Kieser
- Institute of Medical Biometry and Informatics, University of Heidelberg, Heidelberg, Germany
| Published: | February 26, 2021 |
|---|
Outline
Text
Background: In early clinical development, where only few drug candidates have the potential for market approval, the gold standard, a randomized controlled trial (RCT) might not be the most efficient option for ethical and economic reasons. The required trial sample size could be reduced as the control treatment was already previously investigated. Therefore, a single-arm trial with external controls (SATwEC) may be a pragmatic alternative. Using individual patient data, propensity score methods could provide unbiased comparisons between the experimental drug and the external control if all relevant confounders are observable. However, there is a practical challenge: the propensity score methods are applied at the end of the trial. If the resulting propensity score distributions are very different, the SATwEC might lead to inappropriate decisions for further development or require a subsequent RCT before going into late development.
Methods: We aim to develop and assess through simulations an adaptive design that reduces the risk of unreliable decision making at the end of a SATwEC.
In stage I of the adaptive design, subjects are solely assigned to the experimental arm. If the propensity score distributions at interim are comparable based on the preference score [1], further subjects in stage II will also be assigned to the experimental arm; if not, a randomized stage II will be conducted. Futility stop at interim is an additional design option.
In the simulation study, data is generated using a time-to-event model with observable as well as unobservable confounders. The confounder space will be modified to investigate the impact on treatment effect estimates and decision making.
Results: With comparable confounder spaces, the SATwEC is the best approach in balancing correct decisions and sample size. In case of strong deviations, the randomized design is preferable. The proposed adaptive design provides a compromise as being less prone to wrong conclusion in those “strong deviations”-settings than the SATwEC design while requiring less investment than an RCT.
Conclusion: The proposed adaptive design is a viable and pragmatic option for early clinical development if external control data are available for an interim analysis.
HG, AA, FXL, AS have no conflict of interest with the subject matter of this talk while being employees of Merck Healthcare KGaA.
The authors declare that an ethics committee vote is not required.
References
- 1.
- Walker AM, Patrick AR, Lauer MS, Hornbrook MC, Marin MG, Platt R, Roger VL, Stang P, Schneeweiss S. Tool for assessing the feasibility of comparative effectiveness research. Comparative Effectiveness Research. 2013:3 11–20. DOI: 10.2147/CER.S40357
