|
Activity Number:
|
374
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Wednesday, August 9, 2006 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Statistics in Epidemiology
|
| Abstract - #306598 |
|
Title:
|
Statistical Design and Multiple Testing Analysis of Microarray
|
|
Author(s):
|
Jane Chang*+ and Jason Hsu
|
|
Companies:
|
Bowling Green State University and The Ohio State University
|
|
Address:
|
Department of ASOR, Bowling Green, OH, 43403,
|
|
Keywords:
|
microarray ; sensitivity ; specificity ; design ; multiple testings
|
|
Abstract:
|
Microarray experiments are no longer for discovery only. There are microarray-based products marketed for cancer recurrence prognosis. Microarrays for clinical use should meet FDA joint CDRH/CDER statistical requirements. But, it has been observed that reported sensitivity and specificity of biomarker-based cancer prognostics from even major studies are not necessarily reproducible. We will first describe DESIGN issues with microarray experiments that likely have contributed to this apparent lack of reproducibility. Recommendations for statistically designing microarrays and sample hybridization toward reproducible results will be given. Multiple testings ANALYSIS issues of gene expression levels also will be discussed.
|
