# Multiparametric MRI to stratify risk factors for hemorrhagic complications in inoperable glioblastomas following stereotactic needle biopsy

## TL;DR

This study uses MRI to identify risk factors for bleeding after brain biopsies in inoperable glioblastomas, finding that tumor location is the strongest predictor.

## Contribution

The study introduces a multiparametric MRI approach to predict hemorrhagic complications after stereotactic needle biopsy in glioblastomas.

## Key findings

- Lesion location was the strongest predictor of post-biopsy hemorrhage, with deep lesions showing the highest frequency.
- rCBVmax showed a significant linear association with hemorrhage area but not with hemorrhage incidence.
- Grade 3 ITSS lesions were linked to more extensive bleeding, while peritumoral edema showed no correlation.

## Abstract

Histological confirmation of glioblastoma (GB) is essential for therapeutic planning, even in inoperable cases where stereotactic needle biopsy (STNB) is the only option. However, post-procedural bleeding remains a known risk. This study aimed to evaluate the association between MRI features of GB and hemorrhagic complications following STNB.

This retrospective, single-center study included 78 patients with IDH-wildtype GB (mean age: 61 years; 33 females) who underwent pre-biopsy MRI (including SWI and DSC-perfusion) and post-biopsy CT within 72 h. Lesions were anatomically classified into four groups based on their location: cortical/superficial grey matter (sGM n = 12), subependymal white matter (sWM; n = 36), deep nuclei/thalamus (n = 26), or brainstem (n = 4). Hemorrhage incidence and area were correlated with lesion location, intratumoral susceptibility signal (ITSS) grade, rCBVmax values, and peritumoral edema. Clinical outcomes were also recorded.

Hemorrhage incidence significantly differed by lesion location (p = 0.009), with the highest frequency in deep lesions (85%). Most non-hemorrhagic cases (53%) occurred in sWM. While rCBVmax did not correlate with hemorrhage incidence, a significant linear association with hemorrhage area was noted (p = 0.016, r = 0.331). Grade 3 ITSS lesions showed more extensive bleeding. No correlation was found between peritumoral edema and bleeding. Most hemorrhages were asymptomatic; only two patients experienced transient neurological symptoms.

Lesion location was the strongest predictor of post-biopsy hemorrhage. The absence of correlation between rCBVmax and bleeding risk suggests biopsies can be safely performed even in hyperperfused (and potentially more aggressive) tumor areas. STNB remains a safe and valuable diagnostic tool when appropriate preoperative evaluation and postoperative monitoring are ensured.