Altered structural hub connectivity and its clinical relevance in glioma

Background and PurposeStructural network analysis of diffusion imaging is increasingly used to study neurological disease, its pathophysiology and symptoms. We therefore evaluate structural hub connectivity in glioma patients and its association with molecular subtype and clinical status.\n\nMaterials and MethodsUsing retrospective diffusion imaging, structural connectivity was investigated in 65 newly diagnosed glioma patients (36 males; mean age 52 {+/-} 14 years) and 60 healthy controls (23 males; mean age 50 {+/-} 7 years). Probabilistic tractography was performed between 39 cortical nodes per hemisphere. In patients, tumors were drawn in to exclude each tumor-containing voxel from analysis. As previous connectomic research in glioma and other neurological diseases has shown particular importance of hub nodes and connections, the numbers of connections between hubs, hubs and non-hubs, and non-hubs were calculated for each hemisphere separately. Clinical and molecular characteristics were assessed as part of routine clinical care. Group differences in connectivity and its associations with performance and molecular subtypes were tested non-parametrically through Mann-Whitney U-tests, corrected for multiple comparisons.\n\nResultsGlioma patients had more hub-related connections in the hemisphere contralateral to the tumor (hub-hub P = 0.002, hub-non-hub P = 0.005), despite being comparable to controls in terms of total and ipsilateral connections. Within patients, hub-related connectivity related to performance status (P = 0.009) and molecular subtype (P = 0.045).\n\nConclusionWe present experimental evidence for the relevance of structural connectomics as a tool to pick up on the clinical impact of glioma on the rest of the brain.