Background: In the AVAglio study, Bv+RT/T prolonged PFS (co-primary endpoint) in pts with newly diagnosed GB v placebo (P)+RT/T. It has been suggested that antiangiogenic therapy may promote infiltrative tumors at PD, so patterns of tumor progression were prospectively assessed in AVAglio (exploratory endpoint). Methods: Randomizedpts (n=921) received: RT/T+Bv or P, 6 wks; 28-day break; maintenance T+Bv or P (x6); Bv or P until PD/unacceptable toxicity. PD was investigator (INV) assessed (adapted Macdonald criteria). Lesion patterns were retrospectively assessed by an independent review facility for each disease assessment (baseline [BL]/during treatment/at INV-assessed PD). Diffuse/infiltrative patterns were defined as extended hypersignal on T2/fluid-attenuated inversion recovery (FLAIR) sequences, and are reported in pts with pattern derived at BL and INV-assessed PD. Results: Most pts had similar tumor invasiveness at BL and PD (Table). Among pts with non-diffuse disease at BL (n=100 and 138), numerically more Bv- v P-treated pts changed to diffuse disease at PD; median OS for these pts was similar (17.4 v 15.0 mo for Bv+RT/T v P+RT/T, respectively; HR 0.85, 95% CI 0.51–1.42, p=0.5355), and comparable with the OS of the ITT population. In all pts who had BL non-diffuse tumors, median OS was 20.1 v 18.4 months (Bv+RT/T v P+RT/T, respectively; HR 0.76, 95% CI 0.59–0.98, p=0.0303). In all pts with BL diffuse tumors, median OS was 15.6 v 16.2 months (Bv+RT/T v P+RT/T, respectively; HR 0.99, 95% CI 0.82–1.19, p=0.8882). Conclusions: In this exploratory analysis, a change from BL non-diffuse disease to diffuse disease at PD did not affect OS; BL pattern of disease did, suggesting that tumor infiltration at diagnosis is an important prognostic factor. Also, in the subset of pts with newly diagnosed non-diffuse GB, there was an OS benefit from adding Bv to standard of care. Clinical trial information: NCT00943826.