American Statistical Association
We have previously generated a pair of microRNA array datasets for the same set of tumor tissue samples (96 serous ovarian cancer samples and 96 endometriod endometrial cancer samples). One dataset used blocked randomization and uniform handling to prevent any confounding array effects, while the other followed a typical lab practice using neither blocked randomization nor uniform handling and exhibited array effects. Using the former as the benchmark data and the latter as the test data, we have demonstrated (1) logistic feasibility and significant scientific benefits of using blocking and randomization in the design of molecular biomarker studies and (2) limited benefits of applying post-doc normalization as an attempt to remove array effects, when the study purpose was to discover molecular biomarkers that distinguish between two tumor groups. In this follow up study, we will assess the impact of study design and data normalization on discovering molecular biomarkers that are predictive of progression free survival in serous ovarian cancer, using both empirical and simulated data. We will also examine the impact of several analysis choices (such as the variable form of molecular biomarkers and the adjustment of clinical covariates) on the accuracy of prognostic biomarker discovery.
|Date:||Wednesday, December 17, 2014|
|Time:||4:00 - 5:00 P.M.|
Memorial Sloan-Kettering Cancer Center
Department of Epidemiology and Biostatistics
307 East 63rd Street
(between First and Second Avenues)
3rd Floor Conference Room
New York, New York
Note: To gain access to the building, please follow the directions by the telephone in the foyer.