Low Dose to White Matter Predicts Attention and Processing Speed Decline After Fractionated Brain RT

Abstract

<h3>Purpose/Objective(s)</h3> Attention and processing speed, an important neurocognitive function, is subserved in the brain by critical white matter (WM): the corpus callosum (CC) and intra-hemispheric white matter tracts (IHWM). We sought to characterize the dose-response relationship of these structures with decline in attention/processing speed at 6 months post-RT in patients with primary brain tumors, with particular attention to low doses. <h3>Materials/Methods</h3> On a prospective trial, 52 primary brain tumor patients receiving fractionated brain RT underwent volumetric MRI, diffusion tensor imaging, and neurocognitive assessments (for processing speed: Delis-Kaplan Executive Function System Trail Making (D-KEFS-TM) Number Sequencing (TM-NS), D-KEFS-TM Letter Sequencing (TM-LS), and Weschler Adult Intelligence Scale (WAIS-IV) Coding; for attention: D-KEFS-TM Visual Scanning (TM-VS) test and WAIS-IV Digit Span Digits Forward (DF) test. Reliable change indices were calculated and adjusted for practice effects (RCI-PE), with decrease in RCI-PE scored as decline. Regions of interest (ROIs) were autosegmented using open source software and atlas-based tractography, including: CC, all WM w/wo CC, total IHWM without CC, and right/left IHWM without CC. Censoring masks excluded tumor, surgical cavity, scarring, and edema. Planning CTs were co-registered to MRIs to extract DVH parameters for each ROI, including V10 Gy and V20 Gy. Univariate logistic regressions were used to test for dosimetric predictors of decline. <h3>Results</h3> Median age was 47y. Most patients (61.5%) had gliomas. Median prescription RT dose was 59.4 Gy. At baseline and 6 months, 37-43 patients completed the neurocognitive assessments, with 61-78% showing decline. Several significant associations between increasing relative volumes exposed to low dose and decline were seen in all assessments except TM-NS (Table). Investigation of mean and maximum dose parameters yielded similar results. <h3>Conclusion</h3> While maximum dose is typically considered the most relevant dosimetric parameter for toxicity endpoints in the brain, we also detected several significant logistic relationships between V10 and V20 to more diffuse white matter networks (CC and IHWM) and decline in attention and processing speed post-RT. Minimization of low dose bath, whether by planning optimization or use of more conformal modalities like protons, may be clinically relevant for preservation of these critical neurocognitive functions.