Treatment of epithelial ovarian cancer is currently determined by stage and tumor grade. Although treatment response and disease outcome can be predicted for many patients, this is not always true. Technologies to measure gene and protein expression offer the opportunity to evaluate large sets of genes and proteins in parallel. Our hypothesis is that a molecular classification offers the potential to better understand the origin and progression of epithelial ovarian cancer, which in turn will provide the opportunities to investigate new strategies for early diagnosis and novel treatment approaches. Importantly, results of our research will enable targeted diagnosis and therapy to provide good prognosis and quality of life for each patient.

Addressing important clinical questions in ovarian cancer requires systematic knowledge management and analysis of large gene and protein expression datasets. Given the molecular heterogeneity thought to exist in cancer, no single database or computational algorithm will be sufficient to answer both biological and clinical questions in epithelial ovarian cancer. We propose to systematically integrate multiple gene and protein expression data sets from ovarian cancer tissues, provide and apply diverse analysis algorithms to derive the best candidate markers for early diagnosis and prognosis, and validate them by multiple quantitative biological techniques. Being able to systematically generate hypotheses about epithelial ovarian cancer and comprehensively validate them is a unique aspect of this proposal. Although not part of the proposed research, our involvement with clinical trials and drug discovery program provides assurance that viable markers will be further studied to provide clinically relevant outcome in ovarian cancer.

We will create an integrated ovarian cancer repository and intelligent computational tools to analyze existing and emergent data sets on altered gene and protein expression patterns ovarian cancer in conjunction with clinical information. Our goal is to improve ovarian cancer treatment by providing a sophisticated reasoning and data analysis system enabling a more comprehensive understanding of the disease. Our long-term goal is to translate findings from this research to the clinical setting.

By having access to banked ovarian cancer tissues with linked clinical data, we will be able to discern molecular differences of aggressive and less aggressive forms of ovarian cancer, hypothesize different mechanisms present in the familial form of this disease in addition to differentiating known BRCA1/2 subtypes, comparing BRCA1/2 normal and cancer samples, and study how modulation of protein interactions by androgen or estrogen activates pathways or complexes that lead to disease progression. Comprehensive algorithmic approach to integrated data analysis, followed by rigorous and multifaceted approach to quantitative biological validation on multiple samples will provide a robust way to discover novel and clinically useful diagnostic and prognostic markers, improve patient's quality of life by providing the most appropriate treatment and reducing over-diagnosis especially for familial BRCA1/2 patients, and provide potential leads for novel treatment design.

The results of this research will help to fathom biological mechanisms of ovarian cancer and will be applicable to improve disease classification, diagnostic measures, therapy planning, and treatment prognosis. Improving the treatment could in turn improve quality of life for ovarian cancer patients. Using the proposed tools and methodology, physicians will have more relevant information available at the time of diagnosis and treatment planning, and the patient will have a better explanation of the disease, its origin, progression path, and treatment alternatives.

We support an open and free academic access to data and algorithms -- generated hypotheses, data, and developed computational tools will be publicly available. Since many more hypotheses will be generated than any single lab or even institution can promptly validate and use, we will disseminate these results for "distributed validation" in addition to our own validation efforts.