# Three-Dimensional quantitative analysis of the Peri-Enhancing zone reveals ADC and CBV signatures of glioblastoma recurrence

## TL;DR

This study uses 3D MRI analysis to detect early signs of glioblastoma recurrence by examining diffusion and blood flow patterns in specific brain regions.

## Contribution

A novel 3D framework is introduced to analyze the peri-enhancing zone and identify ADC and CBV signatures predictive of glioblastoma recurrence.

## Key findings

- Recurrence-prone regions in FLAIR-hyperintense PEZ showed significantly lower ADC and higher CBV compared to stable regions.
- No significant ADC or CBV differences were found in FLAIR-normal PEZ regions.
- The study highlights the importance of FLAIR signal in identifying at-risk tissue for glioblastoma recurrence.

## Abstract

Accurate detection of glioblastoma infiltration and early peri-enhancing changes that predict recurrence remains challenging. We present a three-dimensional (3D) quantitative framework to characterize the peri-enhancing zone (PEZ) on baseline and follow-up MRI and examine whether baseline ADC and CBV differentiate regions that subsequently recur from those that remain stable.

We retrospectively analyzed patients with IDH wild-type glioblastoma who developed local recurrence within 12 months post-resection. The contrast-enhancing tumor core including the necrotic region, and surrounding FLAIR hyperintensity were segmented in 3D on diagnostic MRI. Local recurrence was identified on follow-up MRI as abnormal T1 enhancement contiguous with the initial T1-enhancing core and mapped onto baseline images. The PEZ (up to 5 voxels surrounding the T1-enhancing core) was divided by FLAIR signal into hyperintense (PEZFLAIR+) and non-hyperintense (PEZFLAIR-) subvolumes. Within these subvolumes, voxels overlapping with future T1 abnormalities were classified as recurrence-prone tissue (PEZT1+), while voxels without subsequent enhancement were designated as normal remaining tissue (PEZT1-), yielding four analytical subvolumes for quantitative region-of-interest–based analysis. Mean ADC and CBV were calculated per subvolume and compared using Wilcoxon signed-rank tests and Pearson correlations.

Of 101 eligible patients, 59 developed local recurrence and 46 were included in the final analysis. The mean PEZT1+ volume was 6312 ± 4131 mm3, with approximately 30% located in (PEZFLAIR-). Within PEZFLAIR+, PEZT1+ regions showed 8.2% lower ADC (p≈0.0007) and 13.5% higher CBV (p≈0.006) than PEZT1-, with significant ADC–CBV correlation (p < 0.0005). No significant differences or correlations were observed in PEZFLAIR-.

3D volumetric analysis of the PEZ reveals distinct ADC and CBV signatures in regions predisposed to recurrence, which are not apparent on conventional MRI. Differences were confined to FLAIR-hyperintense regions, highlighting the need for novel imaging strategies to detect at-risk tissue in FLAIR-normal PEZ. Importantly, we present a new 3D evaluation approach linking baseline and follow-up MRI that can be applied to investigate other clinical and research questions in brain tumor imaging.