A Statistical Procedure to Evaluate the Most Accurate Animal Model of Human Disease

• Activity Number: 642
• Type: Contributed
• Date/Time: Thursday, August 7, 2014 : 10:30 AM to 12:20 PM
• Sponsor: IMS
• Author(s): Cuilan Gao*+ and Behrouz Shamsaei and Stanley Pounds
• Companies: University of Tennessee at Chattanooga and University of Tennessee at Chattanooga and St. Jude Children's Research Hospital
• Keywords: Genomics ; Genome-wide gene expression ; ANOVA ; Multiple Comparison ; Animal Model

Abstract

Animal models of human disease are commonly utilized to gain preclinical insight into the potential efficacy and mode of action of novel drugs. The development and selection of an animal model that accurately mimics the human disease profoundly reduces the research timeline and resources needed to make meaningful advances in the treatment and prevention of the human disease. We propose a statistical procedure to select the animal model that most accurately mimics the human disease in terms of genome-wide gene expression.Our procedure is designed for studies that have gene expression profiles for a cohort of human disease tissue specimens from different subjects and gene expression profiles for cohorts of disease tissue specimens for each of several animal models. First, we define and compute a metric of similarity of each human gene expression profile with each animal gene expression profile. An ANOVA model is used to compare the mean similarities between different animal models and a follow-up multiple comparison (1984, Hsu) to seek the "best" animal model. The advantages of the proposed method are observed in simulation studies and a real example of pediatric medulloblastoma.