| authors | Nikolakopoulos, S. |
| source | Faculty of Social and Behavioural Theses (2011) |
| full text | The full text of this item is not available.
|
| document type | Master thesis |
| disciplines | Sociale Wetenschappen |
| abstract | In drug development programs, early phase II clinical trials typically have a biomarker as a primary outcome.
Decision on whether to proceed with further development is then based on frequentist analysis of
the biomarker data. We suggest an additional rule based on Bayesian predictive probabilities. We adopt the
methodology suggested by Stallard et al. (2005) for evaluating the probability of success of a subsequent
phase III study. We estimate this probability based on phase II results on the biomarker. For a dichotomous
clinical endpoint of interest to be measured in phase III, a Beta prior distribution is suggested. This distribution
quantifies information from both the data observed on the biomarker and its predictive accuracy, which
provides insight on how much information from the biomarker is passed on to inform decision-making. We
quantify the substantial impact of the predictive ability of biomarkers, aiding decisions concerning both the
choice of the biomarker and also further clinical development through a phase III trial. The approach is
illustrated with a practical example where a biomarker has to be chosen as a primary endpoint for the phase
II trial. A more extensive simulation study is also presented. |
| keywords | Clinical trials, Decision making, Surrogate marker, Beta Distribution, Probability
of success. |
