Change in Hippocampus Volume as a Function of Radiation Dose: Early Results From a Prospective Trial with Standardized Imaging and Morphometric Evaluation.

Abstract

PURPOSE/OBJECTIVE(S)\nPediatric brain tumor patients are at high risk of developing neurocognitive deficits following treatment. The perihippocampal subventricular zone contains a niche of radiosensitive neural progenitor cells linked to memory development, and radiotherapy (RT) to this brain substructure has been associated with neurocognitive impairment in randomized trials. In this prospective study, 3D volumetric MRIs were obtained in pediatric brain tumor patients at baseline and during follow-up to measure volumetric changes in multiple brain substructures along with neurocognitive, endocrine, quality-of-life, and exploratory biomarker assessments. In this planned interim analysis, we model early outcomes for change in hippocampal volume at 6 months following RT.\n\n\nMATERIALS/METHODS\nAs of 2/26/2021, 47 patients had enrolled on this prospective study and 36 had completed their 6-month follow-up assessments after fractionated intensity-modulated proton therapy (IMPT) for primary brain and skull base tumors. Left and right hippocampus volumes were independently measured on T1 sagittal precontrast MRI using automated software at baseline and 6-months after RT and were compared to manual physician contours. The relationship between mean hippocampus dose and change in volume was assessed by Pearson s correlation coefficient. The effect of mean hippocampus dose on change in volume was assessed for mean doses < 10 Gy and ≥10 Gy by t-test. A linear mixed-effects (LME) model was applied to evaluate other predictors associated with change in hippocampus volume, assuming random effects of subjects. Potential factors considered were age, gender, tumor location, focal vs. whole brain RT, prior craniotomy, and chemotherapy.\n\n\nRESULTS\nMean hippocampus dose was strongly correlated with change in hippocampus volume at 6 months following RT (r\u202f=\u202f-0.727, 95% CI [-0.820 -0.596], P < 0.001). Hippocampus volumes and observed changes over time were similar between the software and physician-delineated contours. A significant reduction in hippocampus volume was observed for mean doses ≥10 Gy (mean Δ -10.8% ± 5.5%, P < 0.001), while no significant change in volume was observed for mean doses < 10 Gy (mean Δ +0.7% ± 3.9%). The LME model demonstrated that only mean hippocampus dose was significantly associated with change in hippocampus volume (P < 0.001). The final model predicted a -3.4% change in hippocampus volume for every 10 Gy increase in mean dose. Regression diagnostics showed no evidence of lack-of-fit and no patterns in residuals.\n\n\nCONCLUSION\nChange in hippocampus volume was correlated with hippocampus mean dose at 6 months following RT. A significant reduction in hippocampus volume was observed for mean doses ≥10 Gy compared to no significant change at mean doses < 10 Gy. Future analyses from this study will assess volume change in this and other brain substructures over time as a function of radiation dose and will correlate these findings with measured neurocognitive and other late effects.