Statistical Interaction Term to Assess Treatment Effect in Biomarker Analysis

• Activity Number: 331
• Type: Contributed
• Date/Time: Tuesday, August 5, 2014 : 10:30 AM to 12:20 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Dung-Tsa Chen*+ and James J. Chen and Ying-Lin Hsu and Po-Yu Huang
• Companies: Moffitt Cancer Center & Research Institute and NCTR/FDA and National Chung Hsing University, Taiwan and National Chung Hsing University, Taiwan
• Keywords: interaction ; predictive biomarker

Abstract

In biomedical research, one purpose of biomarker analysis is to identify biomarkers with potential clinical utility, such as cancer risk assessment, prognosis, and treatment prediction. For example, the biomarker, estrogen receptor, is associated with overall survival in patients with breast cancer. Another biomarker, EFGR, is able to predict who will likely respond to gefitinib or erlotinib treatment in patients with non-small-cell lung carcinoma. While there are some success in biomarker discovery, challenge remains to identify useful biomarkers for clinical application in mining genomic data. This is specifically difficult for predictive biomarker development which predicts the likelihood of patient's response to a particular treatment. While there are various ways to develop predictive biomarkers, such as by preclinical experiments and biological mechanisms in molecular level, statistical method can also make significant contribution. This is feasible by employing an interaction model to evaluate if biomarker of interest interacts with treatment. Examples of microarray data analysis are used for illustration.