AI-based volumetric vs bidimensional RANO 2.0 progression assessment: Prognostic value for overall survival in glioblastoma.

Abstract

e14099 Background: Glioblastoma (GBM) presents unique imaging challenges due to irregular morphology. The RANO 2.0 criteria recommends both bidimensional (2D) and volumetric (3D) measurements, yet their comparative prognostic value between two methods needs more evidences. While 2D measurements offer simplicity, they may inadequately capture complex tumor geometry and demonstrate high inter-reader variability. Determining which method provides superior prognostic discrimination is essential for RANO 2.0 implementation in clinical trials. We compared the prognostic value of 2D versus 3D RANO 2.0 assessments in predicting overall survival (OS) and evaluated concordance in progression-free survival (PFS) determination. Methods: Nine newly diagnosed GBM patients from a blinded pooled clinical trial cohort were analyzed. All received standard-of-care treatment. Contrast-enhanced T1-weighted brain MRI was performed at 8-week intervals. Progression was independently determined using RANO 2.0 criteria. Time-dependent Cox regression models, treating progression as time-varying covariates, assessed associations between progression and OS for each method. Kaplan-Meier analysis compared survival outcomes. Results: Among 9 patients 2D RANO detected progression in 6 patients (67%) while 3D RANO identified 5 patients (56%). Median PFS was 54 weeks for both methods, demonstrating high temporal concordance. However, one patient showed 2D-detected progression without meeting 3D criteria, indicating measurement-dependent discordance in 11% of cases. In time-dependent Cox regression, 3D volumetric assessment demonstrated a numerically higher hazard ratio for OS (HR=7.75, 95% CI: 0.73-81.83, p=0.089) compared to 2D assessment (HR=4.53, 95% CI: 0.32-64.72, p=0.27), suggesting stronger prognostic discrimination, though neither reached statistical significance in this pilot cohort due to small sample size. The wider confidence intervals reflect limited sample size but indicate that progression by either method was associated with substantially increased mortality risk. Conclusions: Both 2D and 3D RANO 2.0 assessments provide prognostic value for OS in GBM. Volumetric 3D showed numerically superior discrimination and approached significance despite limited sample size. Findings suggest 3D volumetric measurements may offer enhanced prognostic stratification within RANO 2.0, with implications for clinical trial design and response assessment standardization. Validation in larger cohorts (n= 50 - 60) is planned to definitively establish comparative performance.