Abstract
IMPORTANCE: Long-term survivors of childhood cancer may be at elevated risk for accelerated brain aging. OBJECTIVES: To determine the brain age gap estimation (BrainAGE), defined as the difference between estimated brain age and chronological age; the association between BrainAGE scores and poor neurocognitive outcomes; and the correlations of plasma biomarkers of oxidative stress, neuroinflammation, and cardiovascular health with older BrainAGE scores and previous exposure to central nervous system-directed therapy in adult survivors of childhood cancer. DESIGN, SETTING, AND PARTICIPANTS: This cross-sectional study analyzed participants in the St. Jude Lifetime Cohort Study. Included participants were survivors of childhood cancer (brain tumor, acute lymphoblastic leukemia, and Hodgkin lymphoma) and community controls (no history of childhood cancer) aged 18 years or older with a whole brain scan following a neurocognitive assessment. BrainAGE was calculated from January 2016 to February 2021 and analyzed from February 2022 to March 2025. Treatment information, plasma biomarkers, and neurocognitive outcomes were collected. MAIN OUTCOMES AND MEASURES: The primary outcome was estimated brain age (using previously trained machine learning pipelines) and associations of brain age with neurocognitive outcomes. Associations among BrainAGE scores, plasma biomarkers, and neurocognitive outcomes were examined using multivariable linear regression models and Spearman correlations. RESULTS: A total of 253 survivors of childhood cancer (127 males [50.2%]; mean [SD] age, 31.7 [8.6] years; mean [SD] time since diagnosis, 21.2 [9.8] years) and 43 community controls (24 females [55.8%]; mean [SD] age at evaluation, 31.4 [9.6] years) were included in analyses. Survivors had significantly higher mean BrainAGE scores than controls (6.6 [95% CI, -12.5 to 41.1] years vs 0.7 [95% CI, -1.0 to 2.4] years older than chronological age; P < .001). A 10-year increase in BrainAGE score was associated with lower cognitive flexibility (β = -0.63; 95% CI, -0.82 to -0.45), processing speed (β = -0.49; 95% CI, -0.67 to -0.31), working memory (β = -0.28; 95% CI, -0.53 to -0.03) and visual memory (β = -0.40; 95% CI, -0.58 to -0.22), vocabulary (β = -0.33; 95% CI, -0.48 to -0.18), and reading (β = -0.31; 95% CI, -0.45 to -0.18) z scores after adjusting for age at diagnosis and sex. Neurofilament light was correlated with BrainAGE score in female survivors (Spearman ρ = 0.24 [P = .04]) and controls (Spearman ρ = -0.47 [P = .03]). Female survivors who were younger than 10 years at cancer diagnosis and treated with 40 Gy or higher cranial radiation had mean BrainAGE scores higher than 30 years (37.34 [95% CI, 17.4-41.0] years) older than chronological age (P = .004) and had lower sex hormone precursor correlated with higher BrainAGE score (Spearman ρ = -0.44 [P = .01]) and higher cranial radiation dose (Spearman ρ = -0.48 [P = .01]). Among male survivors, correlations of BrainAGE score and cranial radiation dose with malondialdehyde (Spearman ρ = -0.25 [P = .04] and -0.26 [P = .04]) and oxidized low-density lipoprotein (Spearman ρ = 0.24 [P = .049] and 0.40 [P = .001]) were identified. CONCLUSION AND RELEVANCE: This cross-sectional study found that accelerated brain aging was associated with poorer neurocognitive outcomes and correlated with cranial radiation, biomarkers of atherogenesis, neuronal damage, and sex hormone precursors. These findings can guide future research efforts to reduce the risk of vascular disease in this at-risk population.