2072 Background: PD-1/PD-L1 inhibition has demonstrated limited efficacy for recurrent glioblastoma (rGBM) across prior RCTs. VEGF, a proangiogenic factor upregulated in rGBM, contributes to tumor-associated immunosuppression. Preclinical and observational clinical data indicate a potential dose-dependent effect of anti-VEGF therapy on immune response. This RCT evaluated anti-PD-1 plus anti-VEGF therapy for rGBM. Methods: NAVAL was a multicenter, open label, phase 2 RCT that enrolled 90 patients (≥18 years) with GBM at first recurrence. Participants received nivolumab (240 mg IV Q2 weeks), and were 1:1 randomized to concurrent 10 mg/kg bevacizumab ( standard-dose arm) or 3 mg/kg bevacizumab ( low-dose arm) administered IV biweekly. Stratification factors included age, performance status, extent of resection, and MGMT methylation status. Primary endpoint was overall survival (OS) at 12 months (OS-12). Other outcomes were overall objective response rate (ORR), median progression-free survival (PFS), PFS at 6 months (PFS-6), and safety. Exploratory genomic and immune profiling using CITEseq was performed in 16 patients. Results: 90 patients were enrolled with median age of 60.6 years (range 27-86). 35 patients were MGMT methylated, 53 were unmethylated and 2 were indeterminate. Final analysis had 87 evaluable patients, with median follow-up of 146 days (range 7-1239). There were 3 complete responses (3.4%), 29 partial responses (33.3%), 37 stable disease (42.5%), and 18 progressive disease (20.7%). OS-12 was 41.1% and 37.7% in standard-dose and low-dose arms respectively, with PFS and ORR data in table. Post-hoc analysis demonstrated survival benefit for patients aged ≥60 years in standard-dose arm. Most frequent toxicities (>20%) included fatigue (45.6%), proteinuria (34.4 %), diarrhea (28.9%), hypertension (23.3%) and lipase increase (21.1%). Grade 3-4 toxicities included hypertension (7.8%), fatigue (5.6) and non-neurological toxicities. Differential changes across trial arms were seen in myeloid-derived suppressor cells (MDSCs) post-therapy. Network medicine identified VEGF as target to reduce MDSCs. VEGF was found expressed by MDSCs of mainly responder patients. Differential gene expression analysis identified increased pro-inflammatory gene signatures in standard-dose arm. Conclusions: Overall PFS and OS were similar for nivolumab plus standard- or low-dose bevacizumab for rGBM, with post-hoc survival benefit seen in standard-dose arm in elderly. Standard-dose bevacizumab was associated with increased systemic inflammatory response and reduced immunosuppressive MDSCs. Clinical trial information: NCT03452579 . [Table: see text]