Diagnostic gene expression profiles were first identified ten years ago to identify subsets of childhood leukemia. Since then, untold thousands of studies (2909 listed in PubMed using 'cancer diagnosis microarrays') have been published purporting to be useful tools for cancer diagnosis and treatment. In reality, few have been incorporated for prospective cancer diagnosis and treatment stratification of patients on NCI funded clinical protocols. The Children's Oncology Group has long conducted trials of childhood cancer treatment that include virtually the entire childhood population of North America for patients under the age of 16. As part of those studies, each patient to be admitted for treatment on a COG protocol in one of the nearly 250 participating institutions undergoes central diagnostic review. In the case of childhood rhabdomyosarcoma, the most common form of sarcoma in the young, this review is used to establish eligibility as well as treatment stratification. Currently, separate protocols are open for those deemed to have low, intermediate, or high risk rhabdomyosarcoma. Unfortunately, the criteria used to identify these strata are cumbersome and imprecise, involving histopathology review, clinical criteria like age, stage, anatomic site, and extent of post-surgical disease. In recent years, identification of a unique chimeric gene (PAX-FKHR/FOXO) found only in the alveolar subgroup has been used to identify this intrinsically poor prognosis group. Unfortunately, the histopathology and genetic findings are often discordant (~30% of histopathologically defined alveolar RMS lacks a translocation). Further, there are clearly additional genetic factors that are involved in accurate and reproducible diagnosis and prognosis that go beyond the current criteria. In this proposal, we seek to create clinically useful diagnostic and prognostic biomarker profiles that can be applied to routinely processed (formalin fixed, paraffin embedded) tumor tissue and thus incorporated into the workup of every patient to be admitted on a COG STS RMS protocol. These biomarker profiles have been created over the past five years as part of an NCI funded effort (Director's Challenge and SPECS) and are recently published. They were derived by whole genome gene expression profiling of hundreds of patients'frozen tumor tissue, but they have not been validated on FFPE specimens. Further, the original technology is not readily applied in a timely, cost- effective manner for clinical use, and an alternate technology platform must be identified and validated. Here we propose in the first year to perform extensive cross validation within a CLIA certified laboratory on COG STS RMS protocol treated patients using a variety of technologies and choose the one that best meets these criteria. In the second year we propose to prospectively test the sensitivity and specificity of the new assay on current protocol patients that have been whole-genome profiled as part of the separately funded SPECS initiative. The validated profiles will then be incorporated into future COG STS RMS protocols for risk-stratified therapy using a central reference lab model widely used by COG for other diagnostic studies.
Genomically identified features of cancer like gene expression levels, when correlated with important clinical parameters like diagnosis and prognosis, can be beneficially employed to create predictive biomarker profiles that can predict prior to therapy the diagnosis and prognosis of an individual patient. This information in turn can be used to stratify patients for optimal therapy. Here we propose to translate robust and reproducible diagnostic and prognostic profiles created with the support of the NCI SPECS program into clinical tools applicable to all tumor specimens, frozen or fixed, and to apply them prospectively to new clinical trials for the treatment of childhood rhabdomyosarcoma being developed by the Soft Tissue Sarcoma committee of Children's Oncology Group.
