Many women with small, node-negative breast cancers are essentially overtreated. For example, most Stage I breast cancers are treated with both local and systemic therapies but approximately 80% are effectively cured with local interventions alone. Separating these patients from the approximately 20% who recur, irrespective of their treatment, remains problematic. Consequently, the development of novel methods that can more accurately predict for a nonrecurrent vs. recurrent phenotype is a major priority. We address this issue in our response to PAR-02-010, for which we have established an imaginative and integrated Bioengineering Research Partnership comprising three research teams (Bioengineering & Biostatistics; Clinical & Pathology; Microarray & Molecular Analysis) from two local sister universities (Georgetown University and The Catholic University of America) and the University of Edinburgh (Scotland). We will apply expression microarray and tissue array technologies and powerful new data analysis algorithms to define the gene expression profiles of 600 invasive breast tumors (Stages I-III). Our multidisciplinary teams will use these molecular profiles and established prognostic factors to build artificial intelligence-based classifiers and multivariate models that accurately predict those patients with nonmetastatic disease (especially Stage I) who will not recur. In the long terms, the genes in this classifier and the classifier's algorithms will be used to build custom diagnostic arrays and software for routine clinical use. HYPOTHESES: We hypothesize that differences in the gene expression profiles of tumors determine outcome (recurrence) in patients with nonmetastatic disease. We also hypothesize that computational bioinformatics can discover these differences and use this knowledge to build classifiers that predict each patient's prognosis (especially in Stage I disease).
AIM 1 : We will perform gene expression analysis on breast needle biopsies of 600 invasive, nonmetastatic breast tumors.
AIM 2 : We will build an integrated data processing and management system for data acquisition and retrieval, to support the data analysis algorithms to be optimized and applied in Aim 3.
AIM 3 : We will optimize and apply novel pattern recognition and information visualization technologies, recognizing the high dimensional nature of the data, to discover and validate gene subsets that separate recurrent from nonrecurrent tumors. We will integrate advanced artificial intelligence algorithms and biostatistical models to build predictive classifiers that can more accurately define cancer phenotypes and predict clinical outcomes.
AIM 4 : We will use tissue arrays (multiple cores from archival tissues arrayed on glass slides) to validate and optimize the performance of these classifiers in a retrospective prognostic study of human breast tumors.

Agency
National Institute of Health (NIH)
Institute
National Cancer Institute (NCI)
Type
Research Project (R01)
Project #
5R01CA096483-02
Application #
6800810
Study Section
Special Emphasis Panel (ZRG1-MEP (02))
Program Officer
Lively, Tracy (LUGO)
Project Start
2003-09-12
Project End
2008-07-31
Budget Start
2004-08-16
Budget End
2005-07-31
Support Year
2
Fiscal Year
2004
Total Cost
$466,280
Indirect Cost
Name
Georgetown University
Department
Internal Medicine/Medicine
Type
Schools of Medicine
DUNS #
049515844
City
Washington
State
DC
Country
United States
Zip Code
20057
Gu, Jinghua; Xuan, Jianhua; Riggins, Rebecca B et al. (2012) Robust identification of transcriptional regulatory networks using a Gibbs sampler on outlier sum statistic. Bioinformatics 28:1990-7
Chen, Li; Xuan, Jianhua; Riggins, Rebecca B et al. (2011) Identifying cancer biomarkers by network-constrained support vector machines. BMC Syst Biol 5:161
Gong, Ting; Xuan, Jianhua; Chen, Li et al. (2011) Motif-guided sparse decomposition of gene expression data for regulatory module identification. BMC Bioinformatics 12:82
Cavalli, Luciane R; Riggins, Rebecca B; Wang, Antai et al. (2010) Frequent loss of heterozygosity at the interferon regulatory factor-1 gene locus in breast cancer. Breast Cancer Res Treat 121:227-31
Zhang, Yuji; Xuan, Jianhua; de los Reyes, Benildo G et al. (2010) Reconstruction of gene regulatory modules in cancer cell cycle by multi-source data integration. PLoS One 5:e10268
Wang, Chen; Xuan, Jianhua; Li, Huai et al. (2010) Knowledge-guided gene ranking by coordinative component analysis. BMC Bioinformatics 11:162
Chen, Li; Xuan, Jianhua; Riggins, Rebecca B et al. (2010) Multilevel support vector regression analysis to identify condition-specific regulatory networks. Bioinformatics 26:1416-22
Clarke, Robert; Shajahan, Ayesha N; Riggins, Rebecca B et al. (2009) Gene network signaling in hormone responsiveness modifies apoptosis and autophagy in breast cancer cells. J Steroid Biochem Mol Biol 114:8-20
Chen, Li; Xuan, Jianhua; Wang, Yue et al. (2009) Identification of condition-specific regulatory modules through multi-level motif and mRNA expression analysis. Int J Comput Biol Drug Des 2:1-20
Zhang, Bai; Li, Huai; Riggins, Rebecca B et al. (2009) Differential dependency network analysis to identify condition-specific topological changes in biological networks. Bioinformatics 25:526-32

Showing the most recent 10 out of 30 publications