Background: BEV + radiotherapy/temozolomide (RT/TMZ) improves progression-free survival (PFS) (vs placebo [PBO] + RT/TMZ) in newly diagnosedGBM, but this does not translate into an OS benefit (AVAglio/NCT00943826). Data from recurrent GBM (Tabouret, Neuro Oncol 2014, 2015) suggest that MMP9 and MMP2 plasma levels may predict BEV-related tumor response and correlate with improved OS. Evaluation of MMP9 and MMP2 as potential biomarkers for patients likely to derive OS benefit from BEV is needed. We present data from a retrospective analysis of AVAglio, which assessed whether MMP9 baseline levels predicted OS benefit from BEV in newly diagnosed GBM. Methods: MMP9 and MMP2 levels were assessed (ELISA; R&D Systems) in baselineplasma samples (after surgery/before treatment) from 577/921 patients enrolled in AVAglio (PBO n=294; BEV n=283). Post-hoc analysis and PFS/OS multivariate models including MMP9–treatment interactions were carried out. Results: MMP9 distribution at baseline was comparable between arms (median [Quartile1–3]: PBO 82.4ng/mL [44.8–160.3]; BEV 83.6ng/mL [43.8–133.8]). Patient characteristics were generally balanced between MMP9 subgroups (per quartile), including for known prognostic factors. Patients with low MMP9 (<Quartile1) derived significant OS benefit from BEV, which translated into a median 5.2-month OS increase (HR 0.51, 95% CI 0.34−0.76, p=0.0009; median 13.6 [PBO] vs 18.8 [BEV] months); a consistent PFS benefit was seen (HR 0.36, 95% CI 0.24−0.54, p<0.0001). In patients with high MMP9, there was a trend for the HR to favor the PBO arm for OS (>Quartile 3 HR 1.21, 95% CI 0.80−1.81). No clear prognostic value of MMP9 was shown, although an interaction was seen between treatment and MMP9 (p=0.03) for OS. MMP2 (by quartile) did not exhibit prognostic or predictive value. Conclusions: A post-hoc analysis of AVAglio suggested that baselineplasma MMP9 levels were predictive of OS benefit from the addition of BEV to RT/TMZ in newly diagnosed GBM. Clinical trial information: NCT00943826