P14.53 Deconstructing pathologically increased MEG network clustering in glioma patients. (6th September 2019)

• Record Type: Journal Article
• Authors: Kulik, S D; Derks, J; Numan, T; Hillebrand, A; de Witt Hamer, P C; Klein, M; Geurts, J J G; Reijneveld, J C; Stam, C J; Schoonheim, M M; Douw, L
• Is Part Of: Neuro-oncology. Volume 21(2019)Supplement 3
• Page Start: iii79
• Page End: iii79
• Publication Date: 2019-09-06
• Subjects: Brain Neoplasms -- Periodicals; Brain -- Tumors -- Periodicals; Brain -- Cancer -- Periodicals; Nervous system -- Cancer -- Periodicals; 616.99481
• DOI: 10.1093/neuonc/noz126.288
• Languages: English
• ISSNs: 1522-8517

Abstract

Abstract: Introduction: Functional brain networks in glioma patients are characterized by higher global clustering than healthy controls, indicating stronger connectivity in triads of brain regions when averaging across the entire brain. However, this could be due to either primary increased local clustering of (peri)tumor regions or higher local clustering throughout the entire brain. METHODS: Magnetoencephalography recordings of 71 glioma patients and 53 HCs were analyzed by calculating functional connectivity with the phase lag index between source-localized time series of 78 cortical regions of the automated anatomical labelling atlas. Per participant, we calculated (1) global average clustering, (2) local clustering of tumor and non-tumor regions, and (3) Euclidean distance between tumor centroids and of all other region centroids. RESULTS: Glioma patients had higher global average clustering (p=0.002) than HCs. This increase was indeed global: there was no difference between tumor and non-tumor regions (p=0.154) and no association between distance and local clustering (p=0.759). When splitting patients into high (top 25%, n=18) and normal global clustering (other 75%, n=53) to more specifically pick up on the determinants of pathological global average clustering, again no localized or distance-dependent effects were found. High clustering patients were younger than patients with normal global clustering (p=0.027). Posthoc analysis into tumor localization preference for particular network regions in the entire patient cohort revealed greater tumor occurrence in regions with high clustering in HC (p<0.001), while patients with high global clustering showed tumors localized in regions with lower clustering in HC (p=0.032). CONCLUSION: The functional brain network of a subset of (relatively young) glioma patients is disturbed on a global level, suggesting that treatment thereof might benefit patients. Moreover, our exploratory analyses suggest that gliomas occur more often in normally highly clustered regions, but that tumors occurring in less clustered regions are associated with more extensive global network alterations. These findings may speculatively indicate that patients with and without such pathologically altered global clustering represent distinct phenotypes (both in terms of age and tumor localization) and may also need to be treated as such.