2091 Background: Bevacizumab (BEV) with or without irinotecan (IRI) have demonstrated activity in patients with recurrent GBM. However, conflicting results raise the implication of BEV in facilitating infiltrative progression associated with a shorter survival. While Pope et al. (Neuro-Oncol, 2009) proposed 4 pattern of disease progression, it appear that these pattern definition may not fully identify the “gliomatosis phenotype progression”. We tested a more extensive anatomical analysis as previously described in gliomatosis cerebri (Peretti, J Neuro-Oncol, 2002) Methods: Consecutive MRI performed every 2 months from 25 patients with recurrent GBM treated with BEV+IRI were retrospectively analysed. Status (Local, multifocal, diffuse), number of lobes involved, and tumor infiltration of corpus callosum (CC), basal ganglia or thalamus (BG) and brainstem (BS) were analysed at the time of BEV-+IRI initiation up to progression. Results: At the time of BEV+IRI initiation, 9 (36%), 11 (44%), and 5 (20%) pts presented a local (L1), multifocal (M1), and diffuse (D1) phenotype respectively. According to RANO criterias, 7/25 (28%) pts presented an objective response. At the time of progression, among 9 pts with (L1) presentation, 2 progressed as diffuse (DP), 2 as multifocal (MP) and 5 maintained their local pattern (LP). Of the 11 patients with M1 presentation, 3 progressed as diffuse, and 8 maintained their multifocal phenotype. All pts with D1 status maintained their diffuse phenotype. As a whole, 10/25 pts progressed with a diffuse phenotype, that involved 4 to 6 lobes and, in 8/10 cases, basal ganglia or thalamus and brainstem or corpus callosum. Survival at the time of progression under BEV was <3 mo in 6/10 pts, while survival of M1 or L1 was < 3 mo in 4/15 pts. Conclusions: Recurrent GBM frequently present a diffuse or multifocal presentation while the change of pattern from local to diffuse phenotype associated to BEV trt is uncommun. When observed, diffuse phenotype with extended lobes and midleline involvement, may be associated to a shorter survival. Extended anatomical infiltration analysis prior to BEV administration and at the time of progression should be considered in ongoing and future trials.