10503 Background: The phase III GRID study showed that REG provides a significant improvement in progression-free survival (PFS) compared with PL in pts with advanced gastrointestinal stromal tumors (GIST) following failure of at least imatinib (IM) and sunitinib (SU; HR 0.27, p<0.0001). Determining GIST genotype in TKI-refractory disease has proven challenging due to inter-tumoral heterogeneity and pt preference to avoid serial biopsies. To overcome this, we analysed circulating DNA in plasma as a source of tumor DNA and studied the correlation between mutational status and clinical outcome. Methods: DNA was isolated from both archival tumor tissue (n=102) and plasma at baseline (n=163) and analyzed for mutations via Sanger sequencing (tissue) or BEAMing (plasma). Results: Mutational frequencies for tumor tissue samples were: KIT, 66%; PDGFRA, 3%; KRAS, 1%; BRAF, 0%. For plasma, frequencies were: KIT, 58%; PDGFRA, 1%; KRAS, 1 out of 2 samples, BRAF, 0%. Detection of primary KIT mutations showed 84% concordance between tissue and plasma. Secondary KIT mutations were more commonly detected in plasma (47%) than in tissue (12%). Subgroup analysis based on mutational status showed an improved PFS in REG-treated pts vs PL in all subgroups by both central and local review of imaging studies. The presence of a secondary KIT mutation in plasma was associated with shorter PFS in pts receiving PL (HR 1.82, p=0.05). Pts with a KIT-exon 9 mutation received IM for a shorter period of time, and SU for a longer period of time, relative to other GIST genotypes. Pts with a PDGFRA mutation showed variable clinical responses, while 1/1 KRAS-mutant GIST did not respond well to IM, SU, or REG. Conclusions: KIT mutational status correlated to IM and SU treatment duration. While consistent with prior reports using tissue sampling, this validates the utility of plasma-based circulating DNA analysis of target oncogenes. Secondary KIT mutations appear to have a negative prognostic impact in GIST, while the clinical benefit of REG vs PL was not influenced by KIT mutational status. Clinical trial information: NCT01271712.