Abstract:
|
Pediatric cancer treatment can have profound and complicated late effects. With the survival rates increasing as a result of improved detection and treatment, there is a critical need for a more comprehensive understanding of the impact of current treatments on neurocognitive function and brain structure. We propose an integrative Bayesian mediation analysis approach to model jointly a treatment exposure, a high-dimensional neuroimaging mediator, and a neurocognitive outcome and to uncover the mediation pathway from the exposure through the mediator to the outcome. The proposed method models the high-dimensional imaging-related coefficients via a binary Ising–Gaussian Markov random field prior (BI-GMRF), which addresses the sparsity, spatial dependency, and smoothness and increases the power to detect brain regions with mediation effects. Numerical simulations demonstrate the estimation accuracy, power, and robustness of the BI-GMRF method. Applied to the SJMB03 data set for pediatric medulloblastoma patients, the BI-GMRF method has identified white matter microstructure that is damaged by cancer-directed treatment and impacts late neurocognitive outcome.
|