An Analysis of the Relationships Between Abnormal Within-Network Functional Connectivity, Preoperative Brain Tumor Variables, and Neuropsychological Test Scores

Abstract Objective Neuropsychological symptoms in brain tumor patients are often incompletely explained by tumor variables and may be more directly related to changes in large-scale functional network connectivity. Here we examine these relationships. Methods Fifty-one participants underwent pre-operative resting-state functional MRIs and three neuropsychological tests—Trail Making Test-Part-B (TMT-B), WAIS-IV Digit-Span Sequencing (WAIS-DS), and Controlled Oral Word Association Test (COWAT). Within-network functional connectivity of the central executive (CEN), default mode (DMN), language (LANG), and salience (SN) networks were compared to healthy controls. Spearman correlations (ρ) were performed between neuropsychological z-scores, abnormal within-network connectivity, and tumor variables. Exploratory, statistical mediation analyses then evaluated if relevant tumor variables mediated neuropsychological performance via functional connectivity. Results Significant correlations included: (1) WAIS-DS performance to lesional-SN (ρ = 0.53, P = .006), lesional-sided-CEN (ρ = 0.42, P = .023), and right-SN (ρ = 0.42, P = .023) connectivity; (2) COWAT performance to right-SN (ρ = 0.50, P = .012), lesional-SN connectivity (ρ = 0.45, P = .017), and lesion laterality (ρ = 0.47, P = .017); and (3) TMT-B to lesional-LANG (ρ = 0.46, P = .017), right-CEN (ρ = 0.45, P = .017), and bilateral-LANG (ρ = 0.42, P = .024) connectivity. Mediation analyses revealed the effect of: (1) lesion laterality on TMT-B was fully mediated via right-CEN connectivity (path a*b; β = 0.273 0.064, 0.478); (2) IDH-status on WAIS-DS was fully mediated via lesional-CEN connectivity (path a*b; β = 0.251 0.015, 0.588); and (3) lesion laterality on COWAT was partially mediated via right-SN connectivity (path a*b; β = 0.333 0.004–0.72). Conclusions Our data support the hypothesis that functional network connectivity may explain some neuropsychological heterogeneity across otherwise anatomically and oncologically similar cases. Notably, more abnormal connectivity correlated with better performance, suggesting compensatory reorganization.