The Dose-Dependent Relationship of the Medial Temporal Network, Parietal Memory Network, and Visual Network on Episodic Memory Decline Following Chemoradiation Therapy in Patients With Diffuse Gliomas

Purpose: In this prospective observational study, we evaluated dose-response relationships between radiation dose to the brain’s resting-state networks (RSNs) and changes in neurocognitive function (NCF) following radiation therapy (RT) in adult patients with diffuse glioma. Methods and Materials: Adult patients with isocitrate dehydrogenase-wildtype and isocitrate dehydrogenase-mutant gliomas underwent NCF testing using the National Institutes of Health Toolbox Cognition Battery and resting-state functional magnetic resonance imaging (rs-fMRI) before (baseline) and 6 months after RT. The battery assessed 5 cognitive domains and generated a fluid cognition composite score. NCF change was defined as the percent change in age-adjusted scores from baseline to follow-up. Radiation dosimetric parameters were extracted for 13 RSNs and 3 subcortical regions, defined by 300 rs-fMRI–derived regions of interest. Correlations between NCF change and dosimetric parameters were assessed using Spearman’s rank correlation test (ρ). Linear regression models were compared using a nested analysis of variance, the Akaike information criterion, and adjusted R². Results: Of the 48 patients enrolled, 36 were evaluable with paired rs-fMRI and NCF data. Moderate negative correlations were observed between changes in episodic memory and the mean dose to the medial temporal lobe network (ρ = −0.41; 95% CI, −0.66, −0.08; P = .01), visual network (ρ = −0.42; 95% CI, −0.67, −0.09; P = .01), and parietal memory network (ρ = −0.40; 95% CI, −0.65, −0.07; P = .01). No significant correlations were found for other RSNs or NCF domains. A linear regression model incorporating medial temporal lobe dose and its interaction with age outperformed the age-only model in explaining variance in episodic memory change (P = .046; Akaike information criterion change = −2.95; adjusted R²= 0.313). Conclusions: Focal dose-response relationships were observed between radiation dose to specific RSNs and episodic memory changes following RT, highlighting the prognostic and therapeutic potential of rs-fMRI in identifying targets for cognitive preservation in patients undergoing RT.