A randomized biomarker-stratified design, which uses the biomarker to guide analysis but not treatment assignment, provides a rigorous assessment of the utility of a potential biomarker for guiding therapy, according to a commentary published online January 14 in the Journal of the National Cancer Institute.
In the commentary, Boris Freidlin, Ph.D., of the Biometric Research Branch at the National Cancer Institute in Bethesda, Md., and colleagues discuss both advantages and disadvantages of commonly used randomized trial designs, including biomarker-stratified, enrichment, and biomarker-strategy designs.
According to the authors, clinical biomarker tests will play an important role in achieving personalized medicine for cancer patients.
Based on a comprehensive review of the three designs, the researchers concluded that biomarker-stratified designs were usually best, when feasible, because they show complete information on the relationship between treatment effect and the biomarker. In this design, patients are randomly assigned regardless of biomarker status, but the analysis is centered on the relationship between the biomarker and the effect of the treatment.
"The biomarker-stratified design maximizes the advantage of randomization by providing unbiased estimates of benefit to risk ratios across different biomarker-defined subgroups and for the entire randomly assigned population," the authors write.
Contact: NCI Press Officers, email@example.com, 301-496-6641
Note to Reporters:
The Journal of the National Cancer Institute is published by Oxford University Press and is not affiliated with the National Cancer Institute. Attribution to the Journal of the National Cancer Institute is requested in all news coverage.
Visit JNCI online at http://jnci.oxfordjournals.org and the JNCI press room at http://www.oxfordjournals.org/our_journals/jnci/press_room.html
For the latest cancer news and studies, follow us on Twitter @JNCI_Now
AAAS and EurekAlert! are not responsible for the accuracy of news releases posted to EurekAlert! by contributing institutions or for the use of any information through the EurekAlert! system.