Longitudinal Fractional Tumor Burden from DSC-MRI Identifies Glioblastoma Patients at Higher Risk of Recurrence
Abstract
Purpose
Reliable biomarkers for predicting glioblastoma (GBM) recurrence timing remain elusive despite widespread use of perfusion MRI. This preliminary analysis evaluates the rate of change in fractional tumor burden (FTB), derived from dynamic susceptibility contrast (DSC) MRI, as a potential longitudinal marker of recurrence dynamics. These data represent an initial subset of a larger dataset under development.
Methods
Eight IDH1-wildtype GBM patients (21 post-chemoradiation DSC-MRI) were analyzed under IRB approval (2012-0441, PI Dr. Puduvalli). Imaging followed a standardized single-dose, low–flip-angle DSC protocol. Relative CBV maps were generated using IB Neuro with BSW leakage correction and registered to T1CE. Automated tumor segmentation used nnU-Net on T1/T1CE/T2/FLAIR images, integrated with tissue classification, and co-registered to DSC space. The fraction of enhancing tumor volume exceeding validated rCBV thresholds (>1.0, >1.37, >1.56) was computed per scan. Longitudinal slopes (ΔFTB/Δt: fraction/month) between scans were classified as rising or falling using cutoffs from the observed ΔFTB/Δt distribution. Robustness of trajectory classifications was evaluated via median separation and Cox proportional hazards models across FTB thresholds.
Results
Patient classifications remained unchanged across FTB thresholds. Rising ΔFTB/Δt trajectories were associated with shorter median time to recurrence (0 vs. 46 days), suggesting closer proximity to recurrence with more rapid changes in high-perfusion tumor fraction. Cox analysis yielded identical hazard ratios across thresholds (HR = 2.19, 95% CI: 0.43–11.09), demonstrating methodological consistency. Continuous analyses showed directionally consistent inverse associations between ΔFTB/Δt and time to recurrence, though not statistically significant in this pilot cohort.
Conclusion
Although from a limited cohort, ΔFTB/Δt appears to be a feasible, internally consistent longitudinal perfusion biomarker, stable across analytic thresholds and rCBV definitions. Faster increases in FTB predicted closer proximity to recurrence, supporting the biological plausibility of this rate-based metric. These findings motivate validation in larger cohorts to assess clinical utility for early recurrence monitoring in GBM.