Abstract A065: Pooled Analysis of 7 Multicenter Prospective Studies (n=212) Using an Integrated Multi-Compartment Biomarker Ecosystem to Differentiate True Progression from Treatment Effect in High-Grade Glioma

Abstract Background In this pooled analysis of 7 multicenter prospective studies, Pooled Analysis of 7 Multicenter Prospective Studies (n=212) Using an Integrated Multi-Compartment Biomarker Ecosystem to Differentiate True Progression from Treatment Effect in High-Grade Glioma we confront the pressing clinical dilemma of distinguishing true tumor progression from treatment effect (pseudoprogression/radiation necrosis). We analyzed 212 patients (140 newly diagnosed, 72 recurrent) with high-grade glioma (160 IDH-wildtype GBM; 52 other HGG). Conventional MRI remains frequently ambiguous and blood–brain barrier dynamics limit single-compartment assays, so a multimodal ecosystem may be required. Methods We integrated plasma cfDNA methylation profiling, paired CSF and plasma proteomics, MRI radiomics, and an AI ensemble classifier trained on multimodal features. Tissue/CSF concordance was assessed (κ = 0. 78) and all cases underwent blinded adjudication by two neuro-oncologists. Prespecified thresholds used the Youden index for primary discrimination and a 90% sensitivity rule to prioritize clinical safety. Feasibility: cost ≈ 250/sample, turnaround 48–72 hours, and scalable lab and computational workflows. Thresholding balanced false negatives against harms of unnecessary intervention. Results / Key Findings / New Concepts The ensemble achieved AUC 0. 92, sensitivity 91%, and specificity 84%, with a median lead time gain of 6. 2 weeks over radiographic consensus. Proteomic signals (IL-6, CXCL1) and a myeloid inflammation signature were enriched in treatment effect and correlated modestly with tumor mutational burden (r≈0. 42). We introduce the dynamic biomarker axis and ecosystem monitoring—longitudinal, multi-compartment integration that appears to resolve many ambiguous cases where single modalities fail. Conclusions This pooled, multimodal ecosystem may enable earlier therapy adaptation, reduce unnecessary surgery and steroid exposure, and provide adaptive trial endpoints. Prospective validation and biomarker-guided therapeutic algorithms are warranted; if confirmed, this approach is likely to be paradigm-shifting in neuro-oncology. Citation Format: Mohamed Tharwat. Kamouna, Abdelrahman Wael. Ibrahim, Abdullah Wael. Ibrahim. Pooled Analysis of 7 Multicenter Prospective Studies (n=212) Using an Integrated Multi-Compartment Biomarker Ecosystem to Differentiate True Progression from Treatment Effect in High-Grade Glioma abstract. In: Proceedings of the AACR Special Conference in Cancer Research: Brain Cancer; 2026 Mar 23-25; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2026;86 (6Suppl): Abstract nr A065.