Background: Isocitrate dehydrogenase mutations (IDHmt) define astrocytomas and oligodendrogliomas. IDHmt results in accumulation of R-2-hydroxyglutarate (2HG), leading to epigenetic dysregulation and defective homologous recombination repair, providing a rationale for poly (adenosine 5’-diphophate-ribose) polymerase (PARP) inhibitors. PARP inhibition upregulates PD-L1 so the combination with immune checkpoint inhibition is potentially synergistic. Methods: Patients (pts) with recurrent high-grade IDHmt gliomas were enrolled in this phase II open-label study (NCT03991832). Eligibility included progressive disease with up to 2 prior lines of systemic therapies and ECOG 0–1. Pts received olaparib 300 mg twice daily continuously and durvalumab 1500 mg IV every 4 weeks. Simon’s optimal two-stage design was used. The primary objective was overall response rate (ORR) and disease control rate (DCR) by RANO criteria. Secondary objectives included overall survival (OS), progression free survival (PFS) and safety.Exploratory biomarkers of response and resistance were assessed with serial blood samples using cell-free methylated DNA immunoprecipitation and high-throughput sequencing (cfMeDIP-seq) as well as multiplex-immunohistochemistry. Results: In the 29 pts enrolled between January 2020–February 2023, median age was 40.5 (range 23–66) and 41% were female. Initial tumor grade was 2 in 9, 3 in 8 and 4 in 12 pts, respectively. Median time to enrollment from tumour diagnosis was 5.9 years. All had prior resection and median number of prior systemic therapies was 2. One patient clinically deteriorated before starting treatment. ORR was 10%, 95% CI 2.2–27%, with responses in 3 pts. One pt with grade 4 astrocytoma had a complete response and remains on treatment after 33 months. The other 2 responders with grade 4 astrocytoma had response durations of 4.1 and 9.8 months. DCR was 28% (95% CI 12.7–47.2%) with 5 additional pts demonstrating stable disease. With a median follow-up of 33 months, mOS was 9.5 months (95% CI 4.3–19.3) and mPFS was 1.9 months (95% CI 1.8–3.0). There was no treatment-related grade 3–4 toxicities. Any grade toxicities included fatigue (48%), nausea (17%), diarrhea (10%), and cytopenias (3%). Using serial cfMeDIP-seq, cell free DNA methylomes comparing responders versus progressors using the top differentially methylated regions can predict treatment response with high accuracy (AUC 0.833). Post progression, differentially methylated genes converge on TGF-β and signal transduction pathways, suggesting possible mechanisms of resistance. Multi-modal analysis of long-term responders will additionally be presented. Conclusions: Combination treatment with olaparib and durvalumab for pts with IDHmt glioma is well tolerated but has limited efficacy in unselected pts. cfMeDIP-seq can reliably predict tumour progression providing a blood-based biomarker of treatment response. Clinical trial information: NCT03991832.