Abstract
INTRODUCTION: Radiation therapy (RT) plays an integral role in the treatment of pediatric brain tumors, despite its association with significant long-term adverse effects including cognitive impairments and RT-induced vascular injury. Numerous studies have shown steady decreases in Intelligence Quotient in the years after RT and instances of domain-specific cognitive decline that are more severe in patients treated with a whole brain versus a focal radiation approach. In this study we evaluated the relationship between functional brain network properties derived from resting-state functional MRI (rsfMRI) and cognitive performance measures. rsfMRI may serve as a tool for early prediction of patient outcome and as an adjunct to cognitive testing batteries in instances where patients cannot effectively complete the tasks, e.g. young age. METHODS: Ten patients (ages 6–22 years) treated with RT for 0.10–18 years prior, and two nonirradiated control patients (ages 13–14 years) were scanned on a 7-Tesla system. High resolution T1-weighted anatomical and rsfMRI data were acquired. A computerized cognitive battery (Cogstate) was employed to assess multiple domains of cognitive function. Global measures of network segregation, integration, and regional strength were extracted from rsfMRI connectivity matrices using the AAL (Automated Anatomical Labelling) atlas and compared to cognitive performance measures using the Pearson correlation coefficient. RESULTS: Across all tasks, patients performed poorly relative to age-matched healthy subjects. A visual memory task (continuous paired associate learning) consistently showed good correlations (r > 0.7) with global and regional network measures. Visuospatial and learning/episodic cognitive domains revealed slighter lower correlations (r > 0.6) across the majority of network measures. CONCLUSIONS: This work suggests that rsfMRI functional connectivity matrices can provide network metrics that reflect cognitive performance measures in adolescents after RT. Ongoing work will evaluate these data in a larger cohort with alternative atlases and in conjunction with structural brain networks