Adaptive Designs in Individualized Therapy (ADIT)

Basic data for this project

Description

The overall objective of this project is to combine the flexibility of adaptive designs for clinical trials with new requirements and dynamic developments of individualized therapy approaches. Confirmatory analyses and statistical designs are essential parts of modern clinical trials. Thereby, the requirements for statistical designs have increased. We have to face more complex objectives in clinical trials, especially in pediatric oncology and hematology. More extensive and specific genetic information are available. Individualized therapy approaches including genetic information and criteria to assess the course of disease are changing frequently with increasing refinement. Hence, we are confronted with a very dynamic development of targeted-drug-therapies and a very dynamic risk-score-development. This dynamic research environment has to be taken into account by adequate statistical designs and analyses. Modern clinical trials in pediatric oncology and hematology require the possibility of including promising new therapy approaches and dropping inefficient ones during the course of the trial. Moreover, the development is challenged by a rather small and already highly stratified patient population. New statistical methods and principles that enable corresponding adaptations are needed. Sequences of conventional parallel group trials are reaching their limits, and cannot include all these dynamic therapy developments without a violation of the integrity of the trial. Classical multiple testing principles are too conservative and inflexible to meet the new challenges. In order to combine individualized therapy and their rapid development in pediatric oncology and hematology with confirmatory analyses, more sophisticated designs and statistical analyses-approaches like trials which sequentially overlap in the timeline, with sequentially increasing level of evidence and with a more liberal but still sufficiently confirmative multiple type I error control are needed. In summary, new challenges in individualized medicine require statistical approaches which guarantee a high level of flexibility and a proper multiple error rate control despite small sample sizes and highly fragmented study populations by individualized strata. The results of this project can be applied to virtually any study in individualized medicine.