Validation of breast cancer biomarkersOne of the most important challenges in the validation of breast cancer biomarkers is determining whether a potential biomarker is specific for breast cancer or it is simply a marker of acute disease. Controversy on whether C3a is a specific marker for breast cancer illustrates this issue well. The problem with many biomarker identification studies is that they do not include samples that are representative of acute disease as controls. The advent of high-throughput technologies for gene expression and protein measurement has been accompanied by a plethora of articles describing potential biomarkers. Given the experimental design limitations of the original experiments and of initial and secondary data analyses, many biomarkers are hypothesized without a firm basis and, not surprisingly, cannot be validated in further experiments. We hypothesize that it is possible to invalidate most biomarkers in-silico, before any resources are spent in large scale validation studies.
Our aims are to:
- Determine whether it is feasible to reliably align and rescale biomarker measurements from different technologies spanning different biological scales (e.g., transcriptome and proteome) for comparative analysis;
- Build a resource that will enable researchers to assess the predictive ability of hypothesized breast cancer biomarkers using information from other studies.
In addition to DBMI members (Lucila Ohno-Machado, MD, PhD, Jihoon Kim, Kiltesh Patel, Staal Vinterbo, the following collaborators are involved:
- Pedro A. F. Galante and Sandro de Souza, from the Ludwig Institute for Cancer Research
- Ronilda Lacson, from the Decision Systems Group, Brigham and Women's Hospital
- Enrico Capobianco, from the CRS4 Bioinformatics Laboratory
- Winston Kuo, from Harvard Medical School
- Chris Hinske, from Ludwig-Maximilian Medical School
- Erik Pitzer, from University of Linz
This project is funded by the Komen Foundation.