Brain Glucose Metabolism is Reduced After Radiation Therapy for Head and Neck Cancer

Abstract

The effects of cancer treatment exposures on the brain are poorly understood, particularly for cancers outside of the central nervous system. Patients with head and neck cancer (HNC) treated with radiation (RT) often have portions of their brain exposed to moderate doses of radiation, particularly within the posterior fossa. The purpose of this study was to characterize changes in brain metabolism using 18-fluorodeoxyglucose (FDG) PET after HNC treatment, with emphasis on accounting for the distribution of the radiation exposure.This study included patients who received RT for newly diagnosed HNC at an academic center since 2012 for whom an FDG PET scan was utilized for RT treatment planning, a subsequent PET was obtained within 6 months of completing RT, and both PET scans included the entire brain. Brain regions of interest (ROIs) were defined using the original RT treatment plans and included the intersection of the brain with the 10 Gy, 20 Gy, 30 Gy isodose lines, typically contained within the posterior fossa. The portion of the brain outside of the 3 Gy isodose line was also defined as an ROI relatively spared from radiation (< 3 Gy ROI). PET scans were individually coregistered to the skull on treatment planning CT. The mean standardized uptake values (SUVMean) for each ROI were extracted. The paired samples t-test was used to assess differences in pre- and post-RT SUVMean for each ROI as well as to compare the percent change in SUV between ROIs.A total of 50 patients met the inclusion criteria. Pre-RT average SUVMean was statistically higher for the < 3 Gy ROI than the 10 Gy, 20 Gy, and 30 Gy ROIs (P < 0.01 for each pairwise comparison), consistent with expected baseline differences between cortical and subcortical brain structures. The average SUVMean of the post-RT PET was statistically lower than the pre-RT PET for all ROIs (Table). The percent decline in SUVMean for the < 3 Gy ROI (-11.7% SD: 15.9%) was statistically greater than for the 10 Gy ROI (-7.9% SD: 16.1%, P < 0.001), 20 Gy ROI (-7.3%, SD: 17.3%, P = 0.007), but not the 30 Gy ROI (-6.7% SD: 22.2%, P = 0.076). The percent decline in SUVMean was not statistically different when comparing the 10 Gy vs. 20 Gy ROIs (P = 0.458) or 10 Gy vs. 30 Gy ROIs (P = 0.569).In this population of patients receiving RT for HNC we observed a decrease in SUVMean when comparing pre- and post-RT FDG PET. The decline in SUVMean was greatest in the portion of the brain that was relatively spared from radiation. Further research is needed to clarify whether the observed changes are a direct result of radiation exposure or differences in regional brain metabolism, assess for a dose effect, and to relate changes on PET with patient-centered outcomes.