Neuropsychological Symptoms in Patients with Brain Tumors Correlate More Strongly with Anomalous Large-Scale Network Functional Connectivity than Classical Tumor Variables

Introduction: Patients with brain tumors commonly present with neuropsychiatric symptoms which are only partially explained by tumor variables. Because there is a heterogenous relationship between anatomical localization, presenting symptoms, and recovery, cognitive and behavioral outcomes are more difficult to explain and predict, yet they strongly impact quality of life and clinical decision making. We assessed whether abnormal within-network functional connectivity correlated more strongly with neuropsychological symptoms than classical tumor variables.

Methods: Fifty-one participants with brain tumors underwent resting-state fMRI and neuropsychological evaluations before surgery. Functional connectome matrices were generated with Quicktome™ (Omniscient Neurotechnologies, LLC) from the BOLD signal time series. Pairs of brain parcellations with Pearson’s correlation coefficients (R) ±2 and ±3 standard deviations from healthy control means were deemed anomalous. The percent of within-network anomalies were then calculated for default mode, central executive, dorsal attention, salience, sensorimotor, language, limbic and paralimbic, and visual networks. Relationships between neuropsychiatric test (Trail Making, Digit Span, Animal Naming, Controlled Oral Word Association Tests (COWAT) z-scores, percent within-network anomalies, and tumor variables (pathology, grade, volume, laterality, and location) were examined.

Results: Anomalies on the same side as the lesion within the salience network were highly correlated with Digit Span Total (r=0.558, p=0.000021) and Sequencing (r=0.552, p=0.000027) scores, Animal Naming (r=0.502, p=0.000175) and COWAT (r=0.448, p=0.000978). Anomalies in the right hemisphere within the central executive network were highly correlated with Trail Making Test Part B (r=0.462, p=0.000643) and Digit Span Forward (r=-0.454, p=0.000819) and Backward (r=-0.445, p=0.001068) scores.

Conclusion : Our results demonstrate that neuropsychiatric tests within the executive function domain were highly correlated with anomalies in the central executive network. Additionally, tests within the attention domain were correlated with anomalies in the salience network. Executive function and attention are more broadly distributed cognitive functions, so dysfunction within these domains is better explained using anomalous functional connectivity than common tumor variables.