Effects of Cognitive Reserve in Alzheimer’s Disease and Cognitively Unimpaired Individuals

Dong Hyuk Lee, Sang Won Seo, Jee Hoon Roh, Minyoung Oh, Jungsu S. Oh, Seung Jun Oh, Jae Seung Kim, Yong JeongFrontiers in Aging Neuroscience 2022 · DOI

TL;DR

This study introduces a novel, multimodal neuroimaging-based measure of cognitive reserve (CR) that better captures its complexity than traditional proxies. They found that this CR measure significantly influences how brain pathology affects cognition and clinical progression, with distinct effects in Alzheimer's disease (AD) patients versus cognitively unimpaired individuals.

What Problem Does This Paper Address?

Alzheimer's disease (AD) is characterized by brain pathology (amyloid plaques and tau tangles), yet not everyone with this pathology develops cognitive symptoms at the same rate. The concept of "cognitive reserve" (CR) attempts to explain this discrepancy, suggesting that some individuals can better withstand brain damage without showing significant cognitive decline. However, existing measures of CR, often relying on proxies like years of education, have limitations. This paper aims to develop a more robust, multimodal measure of CR that integrates neuroimaging data (amyloid PET, tau PET, and MRI) and demographic factors. The researchers want to understand how this refined CR measure modulates the relationship between AD pathology and cognitive function, and how it influences the rate of cognitive decline and disease progression across different stages of AD and in healthy aging.

Methods

The core innovation of this study is the development of a novel CR marker. Instead of relying solely on proxies like education, they conceptualized CR as the "residual" between a person's actual cognitive performance and an estimated cognitive performance. This estimated performance was derived from a linear regression model that incorporated key AD neuropathological markers (global tau, amyloid-beta deposition, and cortical thickness), demographic factors (age, sex), and a genetic factor (ApoE ε4), along with total intracranial volume (TIV). This multimodal approach aims to capture a more comprehensive picture of the factors influencing cognitive function.

They then applied this CR marker in two ways:

  1. Modulating Brain Pathology Effects: They investigated how their CR marker interacted with AD pathology (amyloid and atrophy) to predict cognitive function (ADAS-cog 11 and MMSE) in the AD spectrum. This involved voxel-wise regression and testing for significant interaction effects.
  2. Predicting Clinical Progression: Using longitudinal data, they examined the association between their CR marker and disease conversion rates (progression from mild cognitive impairment or cognitively unimpaired to AD) and changes in cognitive scores over time. This was done using Cox regression and linear mixed-effects models.

Datasets

The study utilized two distinct datasets:

  1. Alzheimer's Disease Neuroimaging Initiative (ADNI 3): This dataset provided cross-sectional data for 91 participants across the AD continuum (AD, amnestic mild cognitive impairment (aMCI), and cognitively unimpaired (CU)). Crucially, these participants had all three imaging modalities (tau PET, amyloid PET, and T1 MRI). Longitudinal data from a subset of these participants (77 individuals) was also used.
  2. MEMORI Dataset: This dataset provided longitudinal data for 53 AD spectrum participants and 34 cognitively unimpaired (CU) individuals, also with tau PET, amyloid PET, and T1 MRI data.

Key Insights and Ideas

  • Beyond Education: The study strongly suggests that traditional CR proxies like years of education, while correlated with their novel CR marker, don't fully capture the complex interplay of factors influencing cognitive resilience. Their multimodal approach offers a more nuanced understanding.
  • Differential Effects: The most striking finding is that CR doesn't act uniformly. In individuals with AD pathology, higher CR appears to accelerate cognitive decline once it starts, suggesting a "threshold effect" where the brain's compensatory mechanisms are eventually overwhelmed. Conversely, in cognitively unimpaired individuals, higher CR seems to mitigate cognitive decline, acting as a protective buffer.
  • Pathology-Cognition Interaction: The novel CR marker was shown to modulate the relationship between brain pathology (amyloid and atrophy) and cognitive function. This means that the impact of brain damage on cognition is not solely determined by the pathology itself, but also by an individual's cognitive reserve.
  • Clinical Progression Predictor: The CR marker was a significant predictor of disease conversion, with higher CR associated with a lower risk of progressing to AD in prodromal AD and CU groups.

Key Results

  • CR Marker Validation: The proposed multimodal CR marker was significantly correlated with years of education, providing a degree of validation.
  • Modulation of Pathology: In the AD spectrum, higher CR was associated with a steeper slope of cognitive decline in relation to AD pathological burden. Specifically, the interaction of CR marker × left frontal amyloid-beta SUVR on ADAS-cog 11 was significant (p = 0.034). Similarly, the interaction of CR marker × left temporoparietal atrophy on ADAS-cog 11 was also significant (p = 0.026).
  • Clinical Progression: In prodromal AD and CU groups, higher CR was associated with a lower conversion rate to AD (continuous CR: HR = 0.57, p = 0.037).
  • Cognitive Decline Trajectories:
    • AD Spectrum: Higher CR was associated with worse cognitive decline (e.g., MMSE: p = 0.026, composite score: p = 0.045).
  • Cognitively Unimpaired (CU): Higher CR was associated with attenuated cognitive decline (e.g., MMSE: p = 0.017, composite score: p < 0.001).

Limitations and Future Work

  • Linear Model Simplicity: The authors acknowledge that their CR model is a relatively simple linear model. They suggest future work could explore more complex interactions between AD biomarkers, or incorporate other factors like white matter hyperintensity or vascular components.
  • Sample Size: The longitudinal study sample size was relatively small, and results are considered preliminary.
  • Dataset Heterogeneity: While they leveraged two datasets, there were differences in amyloid positivity among CU individuals, which could influence generalizability.
  • Future Directions: The authors propose that their CR measure could be valuable for developing non-pharmacological interventions for delaying AD onset or promoting prevention. They also highlight its potential for personalized medicine approaches and for improving the interpretation of clinical trial results by stratifying participants based on CR levels. Further research could explore the underlying neurobiological mechanisms driving these differential CR effects.

Summary updated

Abstract

The concept of cognitive reserve (CR) has been proposed as a protective factor that modifies the effect of brain pathology on cognitive performance. It has been characterized through CR proxies; however, they have intrinsic limitations. In this study, we utilized two different datasets containing tau, amyloid PET, and T1 magnetic resonance imaging. First, 91 Alzheimer's disease (AD) continuum subjects were included from Alzheimer's Disease Neuroimaging Initiative 3. CR was conceptualized as the residual between actual cognition and estimated cognition based on amyloid, tau, and neurodegeneration. The proposed marker was tested by the correlation with CR proxy and modulation of brain pathology effects on cognitive function. Second, longitudinal data of baseline 53 AD spectrum and 34 cognitively unimpaired (CU) participants in the MEMORI dataset were analyzed. CR marker was evaluated for the association with disease conversion rate and clinical progression. Applying our multimodal CR model, this study demonstrates the differential effect of CR on clinical progression according to the disease status and the modulating effect on the relationship between brain pathology and cognition. The proposed marker was associated with years of education and modulated the effect of pathological burden on cognitive performance in the AD spectrum. Longitudinally, higher CR marker was associated with lower disease conversion rate among prodromal AD and CU individuals. Higher CR marker was related to exacerbated cognitive decline in the AD spectrum; however, it was associated with a mitigated decline in CU individuals. These results provide evidence that CR may affect the clinical progression differentially depending on the disease status.
BibTeX
@inproceedings{lee2022effects,
  author = {Dong Hyuk Lee and Sang Won Seo and Jee Hoon Roh and Minyoung Oh and Jungsu S. Oh and Seung Jun Oh and Jae Seung Kim and Yong Jeong},
  title = {{Effects of Cognitive Reserve in Alzheimer’s Disease and Cognitively Unimpaired Individuals}},
  booktitle = {Frontiers in Aging Neuroscience},
  year = {2022},
  doi = {10.3389/fnagi.2021.784054},
}
Figure