Brain Connectomics Improve the Prediction of High-Risk Depression Profiles in the First Year following Breast Cancer
Mu Zi Liang1, Peng Chen2, Ying Tang3
1Guangdong Academy of Population Development, Guangzhou, China.
Background:
Prediction of high-risk depression trajectories in the first year following breast cancer diagnosis with fMRI-related brain connectomics is unclear.
Methods:
The Be Resilient to Breast Cancer (BRBC) study is a multicenter trial in which 189/232 participants (81.5%) completed baseline resting-state functional magnetic resonance imaging (rs-fMRI) and four sequential assessments of depression (T0-T3). The latent growth mixture model (LGMM) was utilized to differentiate depression profiles (high vs. low risk) and was followed by multivoxel pattern analysis (MVPA) to recognize distinct brain connectivity patterns. The incremental value of brain connectomics in the prediction model was also estimated.
Results:
Four depression profiles were recognized and classified into high-risk (delayed and chronic, 14.8% and 12.7%) and low-risk (resilient and recovery, 50.3% and 22.2%). Frontal medial cortex and frontal pole were identified as two important brain areas against the high-risk profile outcome. The prediction model achieved 16.82-76.21% in NRI and 12.63-50.74% in IDI when brain connectomics were included.
Conclusion:
Brain connectomics can optimize the prediction against high-risk depression profiles in the first year since breast cancer diagnoses.
