# Altered perivascular diffusivity in glioblastoma: integrating DTI-ALPS index with radio-pathomic and histopathologic correlates

## TL;DR

This study shows that glioblastoma tumors disrupt brain fluid drainage pathways, as seen through MRI metrics and tissue analysis.

## Contribution

The study introduces DTI-ALPS as a novel MRI-based measure to assess glymphatic dysfunction in glioblastoma.

## Key findings

- DTI-ALPS index was significantly lower on the tumor-affected side in glioblastoma patients.
- DTI-ALPS metrics inversely correlated with tumor volume and cellularity in affected regions.
- Autopsy samples showed tumor cells in perivascular spaces, supporting glymphatic disruption.

## Abstract

Glioblastoma is an aggressive primary brain tumor that often exhibits perivascular invasion. This behavior may directly interfere with glymphatic flow, hindering perivascular drainage routes. This study aims to assess glymphatic dysfunction in glioblastoma by evaluating the DTI-ALPS index, an MRI-based surrogate of glymphatic activity. We additionally correlate mpMRI-derived tumor features with radio-pathomic maps of hypercellularity.

We included 368 IDH-wildtype GBM patients from the UCSF-PDGM dataset. Preoperative T1, T1C, FLAIR, ADC, and diffusion tensor imaging (DTI) maps were preprocessed using standard co-registration and intensity normalization protocols. Radio-pathomic maps of tumor cellularity were generated using a previously published model which was trained on spatially aligned autopsy samples. The DTI-ALPS index was computed using DTI maps normalized to the JHU-ICMB-FA template, with ROIs on predefined white matter tracts and categorized by tumor laterality.

The DTI-ALPS index was significantly lower on the ipsilateral side for both the GTR and STR cohorts (p < 0.00001). Furthermore, DTI-ALPS mean and DTI-ALPS ipsilateral showed an inverse association with contrast enhancing and FLAIR hyperintensity volumes (both p < 0.00001) and total cellularity within the contrast enhancing and FLAIR hyperintensity regions (both p < 0.00001). Notably, autopsy tissue analysis revealed SOX2 positive tumor cells in the perivascular spaces. DTI-ALPS metrics were not independently associated with overall survival after adjustment for clinical covariates.

Our findings suggest that GBM-associated perivascular invasion disrupts glymphatic function, as evidenced by a significantly lower DTI-ALPS index in the tumor-affected hemisphere. The inverse correlation between the DTI-ALPS index and both volume and cellularity metrics highlights the extent to which GBM alters perivascular fluid dynamics.

The online version contains supplementary material available at 10.1007/s11060-026-05458-x.