Alzheimer’s disease (AD) significantly contributes to dementia, characterized by disruptions in functional connectivity within the brain’s default-mode network (DMN). Identifying early neurobiological markers of dementia, especially AD, is crucial for developing targeted prevention strategies. This study explores whether a model of DMN effective connectivity can predict future dementia at an individual level.
Methods
We used spectral dynamic causal modeling (DCM) on resting-state functional magnetic resonance imaging (rs-fMRI) data from a nested case-control group in the UK Biobank. The study included 81 undiagnosed individuals who developed dementia within nine years post-imaging and 1,030 matched controls. We aimed to determine if DMN dysconnectivity could predict future dementia incidence and time to diagnosis.
Key Findings
Prediction Accuracy:
DMN dysconnectivity effectively predicted future dementia incidence with an area under the curve (AUC) of 0.82.
Time to dementia diagnosis correlated significantly with DMN dysconnectivity (R = 0.53).
Comparison with Other Models:
The DMN effective connectivity model outperformed models based on brain structure and functional connectivity in predicting dementia.
Risk Factors:
Strong associations were found between DMN dysconnectivity and major dementia risk factors, including polygenic risk for AD and social isolation.
Discussion
Significance of DMN:
The DMN, comprising regions like the medial prefrontal cortex, posterior cingulate cortex, and inferior parietal cortices, is vulnerable to AD neuropathology. Changes in DMN connectivity are observable in preclinical AD stages and those at high risk due to genetic factors or family history.
Effective Connectivity vs. Functional Connectivity:
Effective connectivity, determined via DCM, provides a more nuanced understanding of neural circuit interactions than traditional functional connectivity measures. It accounts for causal influences between brain regions, offering a detailed neural and hemodynamic model.
Clinical Implications:
Early detection through effective connectivity could lead to individualized prevention strategies. The ability to predict dementia at the single-participant level using rs-fMRI data represents a significant advancement in preclinical diagnosis.
Conclusion
This study demonstrates that DMN effective connectivity is a powerful predictor of future dementia, outperforming other models based on brain structure and functional connectivity. The findings highlight the potential for neurobiological models to facilitate early detection and support targeted dementia-prevention strategies at the population level.
Future Directions
Further research should focus on:
- Expanding the model to include diverse populations.
- Integrating DMN effective connectivity with other biomarkers for a comprehensive early detection framework.
- Exploring intervention strategies tailored to individuals identified as high risk based on effective connectivity patterns.
