Article
Shrinkage estimation for dose-response modeling in phase II trials with multiple subgroups
Search Medline for
Authors
-
Burak Kürsad Günhan
- Universitätsmedizin Göttingen, Göttingen, Germany
-
Paul Meyvisch
- Galapagos NV, Mechelen, Belgium
-
Tim Friede
- Universitätsmedizin Göttingen, Göttingen, Germany
| Published: | February 26, 2021 |
|---|
Outline
Text
Recently, phase II trials with multiple subgroups (e.g. dose regimens or patient populations) have become more popular, for instance in the development of treatments for atopic dermatitis. If the relationship of the dose and response is described by a parametric model, a simplistic approach is to pool doses from different subgroups. However, this approach ignores the potential heterogeneity in dose-response curves between subgroups. A more reasonable approach is the partial pooling, that is certain parameters of the dose-response curves are shared, while others are allowed to be different [1]. Rather than using subgroup specific fixed effects, we propose a Bayesian hierarchical model with random effects to model the between-subgroup heterogeneity with regard to certain parameters. Subgroup specific dose-response relationships can then be estimated using shrinkage estimation [2]. Shrinkage estimation borrows strength across subgroups by assuming similarity (not identity) between subgroups. The model is formulated using interpretable parameters [3]. A weakly informative prior is used for the heterogeneity parameter, which is crucial when only a small number of subgroups is included. In a simulation study, we compare the proposed method with pooling and partial pooling approaches. Considering Emax models, the proposed method displayed desirable performance in terms of the mean absolute error and the coverage probabilities for the dose-response curve compared to the pooling approach. Furthermore, it outperformed the partial pooling by producing lower mean absolute error and shorter credible intervals. The methods are illustrated by a phase II trial example in atopic dermatitis [4]. A publicly available R package, ModStan, is under development to automate the implementation of the proposed method (https://github.com/gunhanb/ModStan).
Acknowledgement: We thank Monika Jelizarow who contributed valuable comments during the conceptualization of the study.
The authors declare that they have no competing interests.
The authors declare that an ethics committee vote is not required.
References
- 1.
- Feller C, Schorning K, Dette H, Bermann G, Bornkamp B. Optimal designs for dose response curves with common parameters. The Annals of Statistics. 2017;45:2102–2132.
- 2.
- Jones HE, Ohlssen DI, Neuenschwander B, Racine A, Branson M. Bayesian models for subgroup analysis in clinical trials. Clinical Trials. 2011;8:129–143.
- 3.
- Röver C, Friede Tim. Dynamically borrowing strength from another study through shrinkage estimation. Statistical Methods in Medical Research. 2020;29(1): 293-308.
- 4.
- Thaçi i D, Simpson EL, Beck LA, et al. Efficacy and safety of dupilumab in adults with moderate-to-severe atopic dermatitis inadequately controlled by topical treatments: a randomised, placebo-controlled, dose-ranging phase 2b trial. The Lancet. 2016;387:40–52.
