Examining the Causal Mediating Role of Brain Pathology on the Relationship Between Diabetes and Cognitive Impairment: The Cardiovascular Health Study

Ryan M. Andrews, Ilya Shpitser, Oscar L. López, W.t. Longstreth, Paulo H. M. Chaves, Lewis H. Kuller, Michelle C. CarlsonJournal of the Royal Statistical Society Series A (Statistics in Society) 2020 · DOI

TL;DR

This study investigates how diabetes impacts cognitive decline, specifically focusing on the role of brain pathology like hypoperfusion and white matter disease. Using advanced causal mediation analysis on the Cardiovascular Health Study data, they found that about a third of diabetes's negative effect on cognition is explained by these vascular brain issues, suggesting a key pathway for intervention.

What Problem Does This Paper Address?

Diabetes mellitus is a widespread metabolic disorder known to be associated with an increased risk of cognitive impairment, including mild cognitive impairment (MCI) and Alzheimer's disease (AD). However, the precise mechanisms by which diabetes leads to cognitive decline are not fully understood. One prominent hypothesis suggests that diabetes contributes to cerebrovascular damage (like reduced blood flow and white matter lesions), which in turn impairs cognitive function. This paper aims to rigorously test this hypothesis by quantifying the extent to which vascular brain pathology mediates the relationship between diabetes and incident cognitive impairment, moving beyond simple associations to establish a causal link.

Methods

The authors employed causal mediation analysis using the inverse odds ratio weighting (IORW) approach. This statistical framework allows for the decomposition of the total effect of an exposure (diabetes) on an outcome (cognitive impairment) into direct and indirect effects, mediated by specific pathways (brain hypoperfusion and white matter disease). The IORW method is particularly useful as it can handle multiple mediators and does not require strong parametric assumptions about the relationships between variables, making it suitable for observational data where random assignment of diabetes is not ethically feasible. They also conducted sensitivity analyses to assess the robustness of their findings to different model specifications and assumptions.

Datasets

The study utilized data from the Cardiovascular Health Study (CHS), a large, 10-year prospective study of US adults aged 65 years and older. Specifically, they focused on participants enrolled in the CHS Cognition Study, which collected detailed information on cognitive function, brain imaging (MRI), and cardiovascular risk factors. The final analytic sample comprised 2872 individuals.

Key Insights and Ideas

  • Diabetes's indirect pathway to cognitive decline is significant: The study provides strong evidence that diabetes doesn't just directly harm the brain; it significantly impacts cognitive function by damaging the brain's vascular system.
  • Brain pathology as a key mediator: Brain hypoperfusion and white matter disease are identified as crucial intermediaries, explaining a substantial portion of diabetes's negative impact on cognition.
  • Quantifying the mediation: The ability to quantify that approximately one-third of the total effect is mediated is a critical contribution, moving beyond qualitative hypotheses to concrete estimations.
  • Support for a common etiology: The findings lend credence to the idea that diabetes and cognitive impairment share underlying vascular mechanisms, suggesting a unified approach to prevention and treatment.

Key Results

  • Diabetes was associated with a 51% increased odds of MCI or dementia (Total effect OR = 1.51).
  • Approximately one-third of this effect was mediated through brain hypoperfusion and white matter disease.
  • The indirect effect (mediated through brain pathology) was significant, while the direct effect was also notable, though not always statistically significant at the chosen p-value threshold.
  • When MCI and dementia were analyzed separately, similar patterns emerged, with the indirect effect being slightly stronger in some cases.

Limitations and Future Work

The authors acknowledge several limitations:

  • Observational data: The study relies on observational data, meaning causal inference is based on assumptions that cannot be definitively proven.
  • Temporal ordering: While efforts were made to establish temporal order, the definition of diabetes in the CHS population might not perfectly align with the age group recruited.
  • Potential for unmeasured confounding: Despite extensive covariate adjustment and sensitivity analyses, the possibility of unmeasured confounders affecting the relationships cannot be entirely ruled out.
  • Focus on specific mediators: The study focused on brain hypoperfusion and white matter disease, but other vascular pathways might also play a role.

Future work suggested includes:

  • Replication in other longitudinal studies: Validating these findings in other cohorts that started in midlife is crucial.
  • Examining path-specific effects: Investigating the individual contributions of brain hypoperfusion and white matter disease separately could offer more nuanced insights.
  • Exploring midlife risk factors: Understanding how midlife diabetes and cardiovascular risk factors influence late-life cognitive health is essential for a life-course perspective.

Summary updated

Abstract

The paper examines whether <i>diabetes mellitus</i> leads to incident mild cognitive impairment and dementia through brain hypoperfusion and white matter disease. We performed inverse odds ratio weighted causal mediation analyses to decompose the effect of diabetes on cognitive impairment into direct and indirect effects, and we found that approximately a third of the total effect of diabetes is mediated through vascular-related brain pathology. Our findings lend support for a common aetiological hypothesis regarding incident cognitive impairment, which is that diabetes increases the risk of clinical cognitive impairment in part by impacting the vasculature of the brain.
BibTeX
@article{andrews2020examining,
  author = {Ryan M. Andrews and Ilya Shpitser and Oscar L. López and W.T. Longstreth and Paulo H. M. Chaves and Lewis H. Kuller and Michelle C. Carlson},
  title = {{Examining the Causal Mediating Role of Brain Pathology on the Relationship Between Diabetes and Cognitive Impairment: The Cardiovascular Health Study}},
  journal = {Journal of the Royal Statistical Society Series A (Statistics in Society)},
  year = {2020},
  doi = {10.1111/rssa.12570},
}
Figure