A voxel-based quantitative framework for analyzing the spatial redistribution and directionality of recurrence in glioblastoma

Glioblastoma is an aggressive primary brain tumor that invariably recurs despite maximal resection and chemoradiotherapy. Understanding the spatial and directional dynamics of recurrence is crucial for informing treatment strategies, yet prior studies have relied mainly on qualitative classification schemes. We conducted a quantitative analysis of glioblastoma recurrence patterns using a two-pronged approach: voxel-based lesion mapping and vector analysis. Lesion distribution shifts between initial and recurrent tumors were statistically evaluated using anatomical labeling based on the AAL atlas. For directionality, we computed vectors from initial to recurrent lesion centroids and assessed their alignment with normative white matter fiber orientations derived from the Human Connectome Project. Lesion mapping revealed a significant posterior shift of distribution in recurrence, particularly involving the parietal lobe. Vector analysis demonstrated that the recurrence vectors exhibited significant directional concordance with local white matter trajectories, as indicated by high mean absolute correlation coefficients (0.60 ± 0.23). These findings suggest that white matter pathways may guide tumor cell migration during recurrence. This study introduces a novel quantitative framework for assessing the spatial and directional features of glioblastoma recurrence. Our integrative analysis highlights the influence of structural brain connectivity on tumor spread and may ultimately contribute to refining initial treatment planning strategies.