Evaluating alterations in longitudinal white matter brain age in survivors of pediatric craniopharyngioma.
AI interpretation is pending for this paper.
Open original publication →What the AI sees
Not AI summarized yet.
Research significance
Pending deeper interpretation.
Source abstract
BACKGROUND AND PURPOSE: Brain tumors and subsequent tumor-directed treatments including surgery, radiotherapy, and chemotherapy may disrupt brain maturation in children, but there are limited quantitative methods to assess this outcome beyond the behavioral and mental constructs of cognitive testing. Here, we developed a brain age estimation tool that uses standard clinical imaging to measure changes in neurodevelopmental integrity after cranial radiotherapy. METHODS: Diffusion-weighted imaging data were collected from 90 pediatric patients with craniopharyngioma who received 54-Gy proton radiotherapy and completed a 5-year follow up regimen. In this retrospective analysis, a ridge regression model was developed from fractional anisotropy scores obtained from major white-matter tracts. Model accuracy was validated on an independent test set. Training and testing samples included data from 72 and 18 patients, respectively. The brain age and age-adjusted brain age gap from this validated model were then calculated at 6 timepoints: at post-operative baseline before radiotherapy and during each of the 5 annual follow-up visits. Longitudinal changes in brain age gap were evaluated by multivariable linear mixed-effects analysis to identify possible predictors of brain age gap change. RESULTS: The median participant age was 9.17 [1.85 to 20.1] years at baseline. Forty-nine participants (54%) were female. The ridge regression model performed strongly on both training datasets (mean absolute error = 1.88 years; root mean square error = 2.42 years) and test datasets (mean absolute error = 2.00 years; root mean square error = 2.57 years). The median brain age gap was 0.01 years at baseline and -2.78 years after 5 years. The brain age gap was negatively associated with time in the linear mixed-effects regression (βTime [95% CI] = -0.25 [-0.41 to -0.09], P < .001), with a significant time-by-age interaction (βTime:Age [95% CI] = -0.24 [-0.34 to -0.13], P < .001) and time-by-sex interaction (βTime:Sex[Male] [95% CI] = -0.23 [-0.42 to -0.05], P = 0.016) interactions. Age and baseline IGF-1 z-scores were positively correlated with brain age gap (β [95% CI] = 0.27 [0.03-0.51], P = 0.027). CONCLUSIONS: We used diffusion-weighted imaging data to model brain age in children treated for craniopharyngioma. Our findings suggest that patients treated for craniopharyngioma demonstrate delayed brain aging compared to their baseline status. Future work will evaluate causative factors for craniopharyngioma-related maturation disruptions.