9001 Background: The success of imatinib (Gleevec) in gastrointestinal stromal tumor (GIST) is proof that a small oncogene targeted molecule can have significant benefit in a solid tumor. Sarcomas, other than GIST, also overexpress one or more of the tyrosine kinases inhibited by imatinib. We are performing a multi-institutional study to evaluate imatinib activity in nine different sarcoma subtypes and desmoid tumors. Methods: Patients ≥10 years old with a sarcoma subtype not curable by multidisciplinary management were eligible. Imatinib was prescribed at 300 mg BID (BSA≥1.5m2), 200 mg BID (BSA=1.0–1.49m2), or 100 mg BID (BSA<1.0m2). Responses at two and four months were collected. Rules for early termination within each disease type were based on a hierarchical Bayesian probability model accounting for correlation of the responses of the 10 disease types. Tissue specimens were analyzed by immunohistochemistry for c-kit, PDGFRα, PDGFR****223'3f NEEDS TO BE ADDED TO TIMES NEW ROMAN GREEK FONT****, AKT, PTEN, FKHR, and by allelic PCR analysis for PDGFRα exon 18. Results: 181 patients have been enrolled; 161 currently evaluable. Patients received prior chemotherapy and all had progressive disease. Four month progression-free survival (PFS) rates follow: all sarcoma subtypes 18% (27/147), angiosarcoma 10% (1/10), Ewing sarcoma 0% (0/13), fibrosarcoma 29% (2/7), liposarcoma 32% (9/28), leiomyosarcoma (LMS) 20% (6/30), malignant fibrous histiocytoma 1/15 (7%), osteosarcoma 18% (3/17), peripheral nerve sheath tumor 20% (1/5), rhabdomyosarcoma 0% (0/2), synovial sarcoma 20% (4/20). Among patients with LMS, one had a complete remission for 11 months while four others had PFS greater than one year. Conclusions: Further investigation of imatinib in the therapy of liposarcoma, LMS, and fibrosarcoma is warranted. Imatinib is less likely to be active in the other sarcoma subtypes. The postulated mechanism of action is through inhibition of polymorphisms of PDGFR and/or downstream effectors. The hierarchical Bayesian probability model is an effective method for studying rare diseases and their subtypes, when it is reasonable to assume that their response rates are exchangeable. Author Disclosure Employment or Leadership Consultant or Advisory Stock Ownership Honoraria Research Funding Expert Testimony Other Remuneration Novartis