DESCRIPTION (provided by applicant): Background: Diagnostic assay for Hsp90 patient selection -an unmet medical need: These days it is often not enough for pharmaceutical companies to bring a drug to market. Regulators and insurers are also requiring companies to develop tests to pinpoint which patients are more likely to benefit from the drug, thereby sparing other patients from needless side effects and expense. The FDA issued guidance to the industry on companion diagnostic assays in July 2011, including its preference for having the test ready for approval at the same time as the drug. This is especially important for the almost 20 Hsp90 inhibitors in clinical studies or about to enter clinical evaluation, where pinpointing th patient population has been especially elusive. For Hsp90 inhibitors, patient selection is currently based on the presence of an Hsp90-dependent oncoclient protein (i.e. HER2 and mutALK), but for most tumors the Hsp90-adicted onco-client protein(s) that drive transformation are difficult to identify. To predict an individual patient's responsiveness, one would need to define, in a tumor-by-tumor manner, the Hsp90's oncoprotein clientele and then understand the make-up and function of chaperone-client complexes, together with the molecular networks in which they are involved, a daunting technical challenge. Hypothesis: We propose here a simple alternative: instead of measuring the network of tumor-driving Hsp90 clientele, we propose to measure in each tumor the abundance of an Hsp90 species, the "oncogenic Hsp90", that allows for the existence of the aberrant tumor-driving clientele. While the tumor becomes addicted to survival on a network of Hsp90-oncoproteins, these proteins become dependent on "oncogenic Hsp90" for functioning and stability. This symbiotic interdependence suggests that addiction of tumors to Hsp90 oncoproteins equals addiction to "oncogenic Hsp90". Measuring the abundance of the latter is a read-out of the first, and therefore a potential biomarker for Hsp90 therapy enrichment. Approach: But how to measure this "oncogenic Hsp90" species? Its abundance is not dictated by Hsp90 expression alone, however certain Hsp90 inhibitors, such as PU-H71 developed by the Chiosis lab and currently in clinical evaluation at MSKCC and the NCI Clinical Center, specifically interact with this "oncogenic Hsp90" species. Labeled derivatives of PU-H71 therefore can be used as tools to measure its presence and its abundance. Indeed, we have created both a fluorescent and a radiolabeled version of PU-H71 that we optimized for use in flow cytometry (for liquid tumors) or positron emission tomography (PET) imaging (for solid tumors), respectively. These tools interact specifically with the "oncogenic Hsp90" and provide a means for the quantification of this Hsp90 species in clinic. Here we propose to (a) conduct exploratory analyses towards validation of the "oncogenic Hsp90"species as a biomarker for patient selection on Hsp90 inhibitor therapy and (b) demonstrate the use of the two chemical tools, PU-FITC and 124I-PU-H71, to non-invasively measure the presence and abundance of this biomarker. Specifically, we plan: (1) To determine in in vitro preclinical models of liquid and solid tumors the ability of the proposed biomarker to predict the subset of cancers that will respond to Hsp90 inhibitors (Guzman and Chiosis). (2) To determine in in vivo pre-clinical models of liquid and solid tumors whether the "oncogenic Hsp90", as measured by PU-FITC labeling in liquid tumors and by 124I-PU-H71 uptake and retention in solid tumors, predicts for anti-tumor activity (Guzman, Lewis and Chiosis). (3) To conduct an exploratory study of the proposed biomarker in clinic (Dunphy, Lewis, Guzman, Chiosis in collaboration with clinical colleagues Gerecitano (Phase 1, PI), Modi and Hudis (breast cancer, Phase 2), Roboz, Tallman (AML, Phase 1,2 PIs) and Larson (nuclear medicine)). This exploratory correlation analysis of tumor sensitivity vs biomarker profile, once validated in follow- up large clinical studies, will ultimately provide an assay for predictive response of tumors to Hsp90 inhibitors. The ultimate goal is to provide a means by which patient selection for future Hsp90 inhibitor treatment would be routinely performed by analysis of the presence and abundance of the "oncogenic Hsp90" as measured by multiparameter flow cytometry using PU-FITC in liquid cancers and by PET using 124I-PU-H71 radiotracer imaging in solid tumors. Significance: To our knowledge, these are the first reported non-invasive companion diagnostic technologies with potential predictive power for patient selection in Hsp90 therapy. In the two described forms, the assay proposes a solution for both liquid cancers and solid tumors. Because the assays offer data never- before available to clinicians, they promise to help accelerate the development of Hsp90 inhibitors in cancers and to inform clinical decision-making with Hsp90-targeted agents. They could therefore, revolutionize the clinical development and the use of Hsp90 inhibitors in an individualized, patient-specific manner.

PUBLIC HEALTH RELEVANCE: Despite of the large number of potential new agents entering clinical evaluation every year, only 5% to 8% ever reach registration. Of particular concern is the high rate of failures in Phase 3, where an estimated 50% of oncology agents are stopped in development. Such failures are especially expensive and deprive many patients of potentially more effective treatments. These dire statistics clearly speak for the need to discover and implement predictive biomarkers for patient selection and trial enrichment. What have we learned from the results obtained in the last years with targeted agents? When a new drug has been administered, either as a single agent or in addition to chemotherapy in a study population not selected by any biomarker, most trials have produced negative results, while in a small minority of cases a statistically significant benefit has been demonstrated. This benefit, however, consisted -- at best -- of a small or moderate absolute prolongation of overall survival. On the other hand, examples of a greater absolute benefit obtained with the use of targeted agents based on a biomarker-driven patient selection are constantly increasing. Biomarkers provide the possibility to use tumor and patient characteristics to integrate an accurate predictor of efficacy with a specific mechanism based therapy, guiding the selection of treatment for each individual patient. In particular, a validated predictive marker can prospectively identify individuals who are likely to have a positive clinical outcome from a specific treatment. The technology described in this grant application provides an approach to improve our use of Hsp90 cancer therapeutics and promises to revolutionize the way we implement them in oncology. In short, we offer a way to identify which patients will benefit from these therapies. Itis the quest of modern oncology to find such technologies and this proposal is first ever to attempt to do so non-invasively in both blood cancers and solid tumors.