Accelerated brain age and associated neurocognitive impairments in adult survivors of childhood cancer

10028 Background: Structural brain aging assessed from routine clinical MRI exams is associated with neurologic and neurocognitive impairments in older non-cancer populations yet is unexplored in childhood cancer survivors. Methods: Neurocognitive assessments and clinical 3-D T1 MRI whole brain images were obtained on 253 childhood cancer survivors (acute lymphoblastic leukemia = 118; brain tumor = 49; Hodgkin lymphoma = 86) and 43 community controls. Brain age was estimated using open-source Gaussian process regression software (BrainAgeR) on visually inspected, segmented, and normalized MRI scans. Benjamin-Hochberg adjusted T-tests were used to compare age-adjusted neurocognitive z-scores between controls and survivors. Linear regression tested associations between brain age acceleration (brain age-chronological age) and treatment (multivariable), and neurocognitive outcomes (univariable). Results: Survivors (mean(SD) age 31.2[8.9]) had poorer executive function (Trails B survivors -0.72 vs controls 0.32, p < 0.0001), processing speed (Grooved Peg Board -0.77 vs 0.05, p < 0.0001), memory (visual selection -0.73 vs -0.10, p = 0.0010; digit backward span -0.45 vs 0.08, p = 0.0304), vocabulary (WASI vocabulary -0.43 vs 0.14, p = 0.0007), and reading (WJ letter word -0.43 vs -0.05, p = 0.0006) compared to controls. In survivors, mean[95%CI] brain age was 7.5[6.4 to 8.6] years older than chronologic age; 0.65[-1.0 to 2.3] years for controls. Compared to males without cranial radiation, for females age≥10 years at cancer diagnosis treated with ≥40Gy cranial radiation, brain age acceleration was 13.35 years (p < 0.001), which worsened by an additional 9.81 years (p = 0.017) for those diagnosed < 10 years. Greater brain age acceleration was associated with a worse performance in executive function (β = -1.73, p < 0.001), processing speed (β = -2.18, p < 0.001), memory (β = -2.07, p < 0.001; β = -1.78, p < 0.029), vocabulary (β = -2.04, p < 0.001), and reading (β = -3.77,p < 0.001). Conclusions: Survivors of childhood cancer experience significant brain age acceleration that is associated with impairments in neurocognitive function. In non-cancer populations, accelerated brain age has been shown to predict late morbidities and mortalities. BrainAge is a promising surrogate biomarker that could predict cognitive decline and improve early detection of dementia risk. Finally, it highlights the risk of cranial radiation in < 10-year-old female cancer survivors.