Whole genome gene expression profiles promise to enable better diagnosis and estimation of prognosis in human cancer. We have developed """"""""cancer signatures"""""""" of sarcomas in children and adolescents from a retrospective analysis of over 600 archived sarcomas from Children's Oncology Group (COG), during which time we have developed a variety of sophisticated bioinformatic tools for clinical use. Our preliminary data document their ability to accurately diagnose sarcomas by class, and have offered new insight into clinically relevant groups heretofore unrecognized: e.g., a molecular as opposed to morphologic definition of alveolar rhabdomyosarcoma and other sarcomas. Additionally, we find that prognostic signatures can also be identified that predict outcome for rhabdomyosarcoma, osteosarcoma, and Ewing's sarcoma as well as, or in some cases better than, existing multi-varite clinical predictors such as clinical stage, histopathology, anatomic location, and response to initial therapy. These prognostic profiles are independent of any of these variables, based solely on expression profiling of the initial biopsy. In addition, these profiles have the potential to identify individual genes that associate with outcome and treatment response and which may represent druggable gene targets for targeted therapies like anti-angiogenic agents, kinase inhibitors, and monoclonal antibody therapy. Based on our initial data from this retrospective analysis of sarcoma patients, we now propose to refine these profiles on a larger, prospectively analyzed cohort of patients treated on current COG bone and soft tissue protocols, to confirm the potential clinical utility of these profiles. This multi-investigator, multi-site, COG sponsored study will establish the extent to which gene expression profiles can move into clinical use for improved diagnoses and risk-based therapy.
Showing the most recent 10 out of 31 publications
