Open Access Highly Accessed Research article

Protein expression based multimarker analysis of breast cancer samples

Angela P Presson12*, Nam K Yoon3, Lora Bagryanova3, Vei Mah3, Mohammad Alavi3, Erin L Maresh3, Ayyappan K Rajasekaran6, Lee Goodglick35, David Chia35 and Steve Horvath145*

Author Affiliations

1 Department of Biostatistics, UCLA, Los Angeles, CA, 90095, USA

2 Department of Pediatrics, David Geffen School of Medicine, UCLA, Los Angeles, CA, 90095, USA

3 Department of Pathology and Laboratory Medicine, David Geffen School of Medicine, UCLA, Los Angeles, CA, 90095, USA

4 Department of Human Genetics, David Geffen School of Medicine, UCLA, Los Angeles, CA, 90095, USA

5 Jonsson Comprehensive Cancer Center, David Geffen School of Medicine, UCLA, Los Angeles, CA 90095, USA

6 Nemours Center for Childhood Cancer Research, Wilmington, DE, USA

For all author emails, please log on.

BMC Cancer 2011, 11:230  doi:10.1186/1471-2407-11-230

Published: 8 June 2011

Additional files

Additional File 1:

Patients clustered by 26 biomarkers and colored by WGCNA, WGCNA* and COX groups. The WGCNA and WGCNA* groups are similar in terms of their assignments of patients to low (white), moderate (grey) and high mortality (black) groups. In comparison, the COX groups defined by a more traditional approach (step-wise cox model selection) were quite different. Yellow indicates missing values.

Format: PDF Size: 63KB Download file

This file can be viewed with: Adobe Acrobat Reader

Open Data

Additional File 2:

Summary statistics for traits and markers from three gene expression data sets. A. The Pawitan 2005 and Sotiriou 2006 data sets were most similar to our TMA data in terms of the percent mortality and survival times, which was 16% and 8 years in our TMA data and 17-22% and 7 years in the Pawitan 2005 and Sotiriou 2006 data sets. The Miller 2005 data set had a longer follow-up time which may explain its higher mortality rate (27%). Estrogen receptor positivity was similar across studies (74%-87%) while progesterone receptor positivity differed by 37% between the Miller 2005 data set and our TMA data. B. Marker expression data for HG-U133A probe sets that best matched our TMA marker data, where each of our TMA markers were represented by two probe set IDs. Medians are plotted with interquartile range (IQR) error bars. Distributions were similar (IQR's overlapped) for at least one of the two probe sets for each marker.

Format: PDF Size: 72KB Download file

This file can be viewed with: Adobe Acrobat Reader

Open Data

Additional File 3:

Summary statistics for variables by WGCNA* mortality group. Median and inter-quartile range (25th - 75th percentiles) are reported for skewed continuous variables. Categorical variables are reported as counts and percent total. The total number of observations and percentage of missing variable data are indicated adjacent to the variable name for continuous variables and categorical variables with more than two levels. Other missing variable data are indicated with footnotes.

Format: PDF Size: 81KB Download file

This file can be viewed with: Adobe Acrobat Reader

Open Data