Beyond Tumor Volume: An Integrated Radiological Model of Tumor Load, Anatomical Spread, and Mass Effect for Survival...

Background and Objectives: Prognostic assessments in grade 4 diffuse astrocytic tumors primarily depend on clinical and molecular characteristics, with radiological attributes frequently assessed in isolation. In this study, we explored whether an integrated radiological approach combining tumor burden, anatomical spread, and mass effect could contribute to survival prediction. Materials and Methods: A total of 310 adult patients with histopathologically confirmed grade 4 diffuse astrocytic tumors, diagnosed between January 2022 and January 2025, were included in this retrospective single-center cohort. Preoperative MRI was used to assess contrast-enhancing tumor volume, edema volume, and brain volume, combined with anatomical spread and midline shift as a marker of mass effect. Tumor burden was defined as the ratio of enhancing tumor volume to brain volume (ETV/BV). Overall survival was analyzed using Kaplan-Meier and Cox regression methods. Model performance was evaluated with the C-index, bootstrap internal validation, and 12-month calibration. Results: Tumor burden was higher in IDH-wildtype tumors, which also showed higher midline shift and more frequent deep structure involvement and contralateral extension. In multivariable analysis, IDH status, age, tumor burden, midline shift, and deep structure involvement were independently associated with overall survival. A greater tumor burden was associated with reduced survival. The addition of molecular and imaging-derived variables improved discrimination, increasing the C-index from 0.69 to 0.76. Following bootstrap validation, the corrected value was 0.73. Calibration at 12 months demonstrated acceptable agreement between predicted and observed outcomes. Conclusions: An integrated radiological approach that combines tumor burden, anatomical spread, and mass effect may support prognostic assessment in addition to established clinical and molecular variables.