Identifying Combinations of Biomarkers for Treatment Selection

• Activity Number: 341
• Type: Contributed
• Date/Time: Tuesday, July 31, 2012 : 10:30 AM to 12:20 PM
• Sponsor: Biometrics Section
• Author(s): Chaeryon Kang*+ and Holly E Janes
• Companies: Fred Hutchinson Cancer Research Center and Fred Hutchinson Cancer Research Center
• Keywords: Treatment selection ; Combination of biomarkers

Abstract

Biomarkers that predict treatment effect have the potential to improve patient outcomes. For example, the Oncotype DX recurrence score has some ability to predict the effect of adjuvant chemotherapy over and above hormone therapy for the treatment of estrogen-receptor-positive breast cancer. Using this marker may allow the treatment to be restricted to women likely to benefit from it while sparing others unnecessary treatment and its toxicity. Oftentimes there are multiple biomarkers and deriving marker combinations is a key statistical task. However the vast majority of methodology on combining biomarkers focuses on deriving combinations that predict outcome under a single treatment. These combinations are not optimized for treatment selection. We address the problem of how to combine biomarkers into a score which is optimized for predicting treatment effect, using an approach that is designed to be robust to model mis-specification. We evaluate the performance of the proposed method in comparison to exiting methods, where each combination is evaluated using a measure of the improvement in expected outcome obtained using the marker combination to select treatment.