An optimized tissue sampling scheme guided by MRI features reveals intratumoral heterogeneity in glioblastoma.

Glioblastoma (GBM) displays marked intratumoral heterogeneity and high invasiveness, making en bloc resection difficult. In particular, total/subtotal resection is challenging for lesions near the diencephalon, midbrain, or brainstem or in motor areas, as damage to these sites can compromise postoperative function and reduce quality of life. Resection of GBM, especially deep-seated tumors, rarely achieves circumferential margins; pathological examination can only be performed on a portion of the tumor. Notably, traditional random intraoperative sampling introduces bias, limiting molecular profiling. The use of imaging-based biomarkers may mitigate this sampling bias, but critical molecular-region signatures remain undefined. In this study, we identified imaging parameters on structural magnetic resonance perfusion imaging to facilitate targeted sampling of representative glioma regions. Whole-exome sequencing of 29 samples from 6 patients revealed significant intratumoral heterogeneity in gene mutations across sampling sites, with edge regions showing the highest heterogeneity. Key cancer driver and drug-target gene mutations also exhibited heterogeneous spatial distributions. Critically, regions with elevated perfusion-weighted imaging cerebral blood volume (PWI-CBV) values demonstrated greater mutation heterogeneity and enrichment of key mutation events. To ensure comprehensive detection of heterogeneous mutations intraoperatively, we propose a three-site sampling strategy within high PWI-CBV regions. This approach may increase the accuracy and completeness of molecular pathological data, ultimately supporting personalized therapeutic planning.