MEDU-15. STRUCTURAL BRAIN NETWORK PROPERTIES AND COGNITIVE IMPAIRMENT IN ADOLESCENTS AFTER RADIATION THERAPY

摘要

INTRODUCTION: Cranial radiation therapy (CRT) has significantly benefitted pediatric brain cancer treatment. The improved prognosis has led to an increasing population of cancer survivors, but CRT is also associated with negative side effects. These include neurocognitive impairments in attention and working memory. Similar declines have been observed in adults with prior pediatric CRT and the newly-diagnosed elderly receiving CRT. Magnetic resonance imaging (MRI) studies have found that CRT has structural effects in the brain, such as altering white matter (WM). Graph network analysis of WM connectivity is a promising method to study the effects of CRT on cognitive functions that engage multiple brain regions. The method models the brain as a network of connected edges and nodes, which are quantitatively described with graph theory. This study used graph analysis to examine associations between neurocognitive deficits and WM connectivity. METHODS: Ten adolescent and young adult subjects (17.7±5yrs.) with prior CRT (7 whole-brain, 3 whole-ventricular) for pediatric brain cancers were scanned on a 7-Tesla MRI. Subjects were also assessed with a computerized cognitive battery (CogState). High resolution T1 anatomical and multiband two-shell diffusion sequences were acquired. A diffusion-based WM connectivity network was constructed and assessed with global and local graph metrics. Local analysis examined node strength at executive function and working memory areas. RESULTS: The analysis found significant negative correlations between graph metrics and cognitive measures of spatial working memory and executive function. Globally, this included measures of segregation (clustering coefficient and transitivity) and integration (efficiency); locally, this included node strength at domain-related regions. CONCLUSION: These preliminary findings suggest that cognitive performance is inversely correlated with graph metrics characterizing WM structural connectivity. The tumors’ infratentorial locations suggest that the findings were treatment-related rather than tumor-related. Our work suggests that CRT-related cognitive effects may potentially be assessed using WM structural graph analysis.