Data mining the serous ovarian tumor transcriptome

Ovarian cancer is the most lethal gynecologic cancer in the United States. If caught in early stages, patient survival rate is 94%, late stage survival rates drop to 28%. It is because most cases are caught in late stages that high mortality is seen. Correct diagnosis is dependent on the presence of symptoms: ~90% of diagnosed ovar- ian cancers are symptomatic. These symptoms tend to be unfocused and not acute. The goal of this project is to develop a transcript-level data set measuring ovarian tumor expression and associated paracrine signaling for later biomarker research. To this end, laser capture microdissection was used with exon based oligonucleotide ar- rays to measure the transcriptome of benign and malignant (Type II) serous ovarian surface epithelial-stromal tumors. In addition to profiling tumor, surrounding stro- mal tissue expression was measured to examine potential paracrine signaling. In total, ~270 million measurements were performed using 50 microarrays. An initial analysis was performed to measure quality, and to compare our measurements against known ovarian cancer properties as established in the molecular genetics literature. Using ontological annotation and de novo pathway generation methods, major trends were defined in the data set including the following: apical surface and tight junction ac- tivity, mitotic activity, tumor suppression in benign tumors, epithelial-mesenchymal transitioning, known ovarian tumor oncogene activity, and evidence of paracrine sig- naling. A list of differentially expressed transcripts was defined which may be explored as biomarkers. The potential for meaningful future analysis is diverse. This data set will contribute to the capacity of the cancer genetics community to perform high resolution exploration of serous ovarian epithelial-stromal surface tumors, aiding in developing better diagnostics and therapeutics