Utlitilizing Neuroquant To Measure Supratentorial Volumetric Brain Changes Over Time And Associations With Dose Parameters In Pediatric Patients Undergoing Radiation Therapy

Radiation dose to the pediatric supratentorial (ST) brain is associated with volumetric changes and resulting cognitive impairment. However, dose-volume relationship has been difficult to define due to growth of pediatric brain confounding analyses, as well as limited detailed dosimetry. NeuroQuant (NQ) is an FDA approved software that can identify and quantify the ST brain based on MRI images standardly acquired in brain tumor imaging. We report the feasibility of implementing this software to investigate volumetric changes over time, and to correlate these changes with RT dose. Pediatric patients at our institution that underwent RT in 2015-2016 and had pre and post T1 weighted MRI available were eligible for analyses. NQ Brain Atrophy Report (Coretech labs) was generated for each scan. Clinical factors and dosimetric parameters were correlated with volumetric changes using Pearson’s correlation coefficients and linear regression modeling including a stepwise multivariate selection. ST brain for dosimetric purposes was defined as a standard brain contour excluding the infratentorial structures such as brainstem and cerebellum in attempt to mirror the NQ ST brain. Mean, V12, V20, V30, V40, and V50Gy in percent doses were correlated with total and ipsilateral ST brain volumes. Thirteen pediatric patients, mean age 12 yrs (range 4-17) were identified. Of these patients, 5 underwent craniospinal irradiation (CSI). The majority of patients were diagnosed with glioma (46%) or medulloblastoma (23%). The median time from preRT to postRT MRI was 15.7 months (range 8-24). The median decrease in preRT to postRT ipsilateral ST brain volume was 19.66 cc, though with high variability and not significantly different from baseline. On univariate analysis, the ipsilateral ST brain V30Gy was the variable most strongly associated with ipsilateral ST brain volume loss (p=0.007), whereas the total ST brain V30Gy was most strongly associated with the total ST brain and total grey matter loss (p=0.007 and p=0.044, respectively). In the stepwise multivariate model, these parameters remained highly significant. However, when excluding CSI patients, these results no longer remained significant, though with limited statistical power (n=8). The impact of dose was most pronounced for ipsilateral ST brain volume changes, where for each percentage increase in V30Gy there was an associated 0.45 cc volume loss. Automated analysis using NQ was successfully implemented to measure ST volumetric changes in the brain of pediatric patients after radiation therapy. Ipsilateral brain volume decreases of 0.45 cc per 1% increase in V30Gy were found in our model, driven by patients receiving CSI. Further analysis with larger sample size and longitudinal analysis is warranted.