Selecting breast cancer patients with micrometastases at diagnosis is crucial for deciding who should and who should not receive toxic and expensive adjuvant chemotherapy to eradicate the metastatic cells. Axillary nodal status, the best marker available, still misclassifies about 25 percent of patients, and assay of individual gene products has never been powerful enough for routine clinical use. We therefore propose to apply two new molecular profiling techniques to a unique set of frozen tumor samples from node-negative patients with no adjuvant therapy and very long clinical follow-up (>12 years), in order to generate clinically useful profiles that more accurately predict long-term outcome. We hypothesize that patients who have never recurred will have metastasis suppressor-like gene expression, and conversely that tumors from patients who experience a recurrence will overexpress genes involved in dissemination and tumor growth at the secondary site. These studies will also identify genes and pathways important in the metastatic process for biological studies, and will provide key data for the other projects in this application focusing on interactions of particular molecular pathways.
Our specific aims are: 1) To identify an RNA expression profile that accurately predicts recurrence of node-negative breast cancer. Our training set of 120 tumors will compare primary tumor specimens from patients with no distant recurrences after at least 12 years vs. those who have recurred. Multigene predictive profiles will be identified using supervised statistical gene selection as well as more exploratory unsupervised methods, and crossvalidated. A further 150 tumors with >12 year follow-up will then be analyzed as a validation study. 2) To identify a DNA profile that predicts recurrence, and integrate this profile with RNA expression. Genomic DNA from the unique untreated tumor subsets from Aim 1 will also be analyzed by array comparative genomic hybridization (CGH) to obtain a DNA profile predictive of recurrence. The prognostic DNA profiles will then be compared and integrated with the prognostic RNA profiles from Aim 1, to identify the genetic components most predictive for metastasis of breast cancer. 3) To develop biological models of metastasis-associated gene function to identify critical metastatic signaling pathways. Breast cancer cell lines will be engineered to express a luciferase reporter, and stably transfected with individual gene candidates identified in Aim 1. The selected genes will be tested for their ability to affect growth, invasion, and colonization at distant metastatic sites, using transfected cell line xenograft models in athymic nude mice. Redundancy of signaling pathways will be examined, and critical candidates with metastatic potential will be determined.
Showing the most recent 10 out of 260 publications
