White matter cognition mapping in pediatric cancer survivors: a two-cohort NODDI-Bingham study

Abstract

Childhood cancer survivors (CS) treated with chemotherapy frequently experience long-term cognitive impairments, commonly referred to as CRCI. Whether these deficits reflect disrupted white matter microstructure, altered relationships between microstructure and cognition, or both remains unclear. Here, we applied the NODDI-Bingham model, which quantifies neurite density (NDI), isotropic free water (ISO), and orientation dispersion along primary (ODIp), secondary (ODIs), and total (ODIt) fiber axes, to characterize biophysically specific white matter microstructure in CS and examine its relationship to cognition. Twenty-one pediatric cancer survivors and 21 age-matched healthy controls (HC) underwent multi-shell diffusion MRI. Voxel-wise NODDI-Bingham analyses were performed within JHU atlas regions. Group differences were tested by ANCOVA, and associations between microstructure and cognition were examined with the NIH Toolbox Crystallized, Fluid, and Early Childhood composites using group-specific regressions. Confirmatory analyses were performed in an age-matched subsample of the HCP (n = 679). Notably, CS showed reduced NDI in the corona radiata, cerebral peduncles, and corpus callosum, with elevated ODIt and ODIp in the middle cerebellar peduncle, pathways that subserve processing speed and interhemispheric integration, cognitive domains among the most vulnerable in CRCI. Group-specific regressions revealed qualitatively different relationships between microstructure and cognition: HC exhibited mixed positive and negative relationships across four of five metrics (138 clusters spanning approximately 305,000 voxels; peak Cohen's d up to 3.00), whereas CS showed a predominantly uniformly negative pattern (67 clusters across 57,000 voxels), with no positive clusters for ISO, ODIt, ODIp, or ODIs, thus highlighting that reduced dispersion and extracellular free water are linked to better cognitive performance in CS. HCP analyses reproduced the HC pattern across all five metrics at proportionally smaller effect sizes, ruling out sample-size effects. NODDI-Bingham identifies two aspects of chemotherapy-related white matter pathology that standard diffusion metrics cannot resolve. First, pediatric cancer survivors show selective elevation of ODIp and ODIt in the cerebellar peduncles and reduced NDI in the corona radiata and corpus callosum, localizing group-level injury to specific fanning and neurite density signatures. Second, whereas in healthy controls white matter shows mixed positive and negative relationships between microstructure and cognition, survivors retain only negative relationships for ODIp and ODIt and a diminished mixed pattern for NDI, such that in the chemotherapy-exposed brain, only reduced neurite density and increased primary-axis fanning relate to cognitive impairment. These findings implicate a biophysically specific restructuring of the relationships between white matter and cognition as a candidate mechanism for CRCI and support NODDI-Bingham as a targeted biomarker panel for monitoring and intervention trials.