|
Abstract:
|
A commercially developed diagnostic device or test to detect or measure predictive biomarker test is said to be a "companion" to a therapeutic product. Many companion diagnostic tests provide a binary output (test negative or positive) to identify a population most likely to benefit from a therapeutic product or a population at risk for a serious adverse event if treated with the therapeutic product. For binary-valued diagnostic devices that determine eligibility of treatment, common Phase III trial designs include the (all-comers) biomarker-stratified (or interaction) design, the biomarker-strategy design, the enrichment (or targeted) design. When a trial is used to compare two binary biomarkers, two designs are the biomarker-stratified design and the discordant risk randomization (enrichment) design. The strengths and limitations of such designs have been described by many authors. However, in the literature the performance measure of the predictive biomarker is not given properly for each of these designs. In this talk, we will discuss common designs of clinical trial for predictive biomarkers and their weakness and strengthens based on the performance measure of the device.
|
