Regular articleCognitive reserve moderates the association between functional network anti-correlations and memory in MCI
Introduction
Cognitive reserve (CR) designates the ability to cognitively perform relatively well in the presence of neuropathology and is thought to be conferred by higher mental abilities like IQ or prolonged life-time experiences such as education. The CR hypothesis is supported by findings from epidemiologic studies showing that proxy measures of CR including IQ and years of formal education are associated with lower risk of dementia (Esiri and Chance, 2012, Meng and D'Arcy, 2012, Valenzuela and Sachdev, 2006, Wilson et al., 2002). Further support for the CR hypothesis comes from neuroimaging studies on mild cognitive impairment (MCI) and Alzheimer's disease (AD), which showed that patients with higher levels of CR proxy measures exhibited stronger neuropathology (such as amyloid pathology, gray matter atrophy, glucose hypometabolism) at relatively stable levels of cognitive performance (Ewers et al., 2013, Morbelli et al., 2013, Stern, 2006, Vemuri et al., 2011). CR-related effects at the level of functional brain networks, however, are not well understood (Barulli and Stern, 2013). Since previous fMRI studies showed circumscribed changes in functional networks that are linked to cognitive impairment in patients with MCI (for review see (Teipel et al., 2016)), the question arises whether higher levels of CR buffer the effects of lower neural network function on memory in MCI. Motivated by a previous finding in nondemented older adults, which showed that at higher levels of IQ, the association between amyloid pathology and episodic memory impairment was weaker (Rentz et al., 2010), we aimed to test whether at higher levels of CR (as measured by IQ or education), lower levels of functional connectivity between functional brain networks is associated with relatively milder memory impairment in MCI.
To assess functional network changes that underlie memory impairment, we measured resting-state functional connectivity between 2 major functional networks including the dorsal attention network (DAN) and the default mode network (DMN). Previous resting-state fMRI studies have revealed an anti-correlation between the resting-state signal of the DMN and the DAN (Carbonell et al., 2014, Chai et al., 2014, Fox et al., 2005), where greater DMN–DAN anti-correlation was associated with higher cognitive control and better working memory performance (Anticevic et al., 2012, Chai et al., 2014, Hampson et al., 2010, Kelly et al., 2008). The resting-state DMN-DAN anti-correlation is reminiscent of observations from task-related fMRI studies showing a task activation of the DAN concomitant with a deactivation of the DMN to be associated with higher episodic memory performance in cognitively normal individuals (Kim et al., 2010, Kragel and Polyn, 2015, Landsiedel and Gilbert, 2015). Here, we hypothesized that in amnestic MCI–a clinical syndrome characterized by memory impairment–lower levels of the DMN-DAN anti-correlation are associated with low memory performance and that such an association can be buffered by CR. Although the anti-correlation of the DAN and DMN itself was not pathologically altered before the stage of AD dementia in previous studies (Wang et al., 2007, Zhu et al., 2016), we hypothesized that the level of the anti-correlation of the DMN-DAN is predictive of memory impairment in the diseased brain (MCI) at low levels of CR. The rationale is that the reciprocal activation of the DAN and DMN has a central regulating role within the ensemble of functional networks underlying cognition (Cole et al., 2014), which becomes especially crucial to maintain successful cognitive processes including memory at the stage of MCI and low CR.
To assess the DAN-DMN anti-correlation, we applied functional connectivity analysis of the DMN-DAN anti-correlation in 2 independent samples of amnestic MCI patients for cross-validation of the results. Via sliding time window functional connectivity analysis, we identified episodes of peak anti-correlations between the networks during resting-state fMRI (Allen et al., 2014, Chang and Glover, 2010, Hutchison et al., 2013a, Sadaghiani et al., 2015, Zalesky et al., 2014). Subsequently, we tested the hypothesis that higher levels of CR proxies (IQ or years of education) moderate the relationship lower levels of peak DMN-DAN anti-correlation and memory impairment in MCI. We included patients with the clinical syndrome of amnestic MCI patients, both single- and multiple-domain amnestic types, regardless of underlying etiology. To assess whether presence of AD pathology (increased levels of amyloid PET) or genetic risk of AD (ApoE ε4 genotype) influence the effect of CR proxies on the relationship between DMN-DAN anti-correlation and memory impairment, we controlled for both amyloid PET levels and APOE genotype.
Section snippets
Study sample
A total of 76 patients with amnestic MCI (MCI) and 36 healthy older adults (HC) were included from the Alzheimer's Disease Neuroimaging Initiative (ADNI) (Petersen et al., 2010). For cross-validation purposes, we included a sample of 93 patients with MCI, who were recruited at the memory clinic of the Institute for Stroke and Dementia Research (ISD), Klinikum der Universitaet Muenchen, Munich (henceforth referred to as the ISD sample).
In the ADNI study, MCI was diagnosed according to the
Sample characteristics
Demographics, cognitive measures, and the a(p)DMN-DAN correlation coefficients for the ADNI and ISD samples are depicted in Table 1. In the ADNI sample, MCI patients were significantly younger than the HC subjects [t(110) = 2.80, p = 0.006].
Comparison of peak a(p)DMN-DAN anticorrelation between HC and MCI subjects
In the ADNI sample, we tested whether MCI patients differed from HC regarding the a(p)DMN-DAN anti-correlation or within network FC measures. We found no differences in the a(p)DMN-DAN anti-correlation between HC and MCI. There were also no observed
Discussion
The major finding of the present study was that in amnestic MCI, higher levels of CR proxies alleviated the association between lower aDMN-DAN anti-correlation and lower episodic memory performance. This effect was consistently detected in 2 independent samples of MCI patients, using different proxy measures for CR (education or IQ). Our findings suggest that higher levels of CR are associated with protective effects at the functional network level in MCI and allow maintenance of relatively
Conclusions
In summary, the present study suggests that higher levels of education and IQ confer protective effects in MCI, moderating the association between lower levels of aDMN-DAN anti-correlation and memory. The results provide a crucial support for the CR hypothesis in that we show a cognitive benefit through altered association between functional network changes and cognitive outcome. Based on these results, questions on the search for compensatory brain mechanisms and therapeutic interventions to
Disclosure statement
The authors have no actual or potential conflicts of interest.
Acknowledgements
The research was funded by grants (to M. E.) of the LMUexcellent Initiative and the European Commission (PCIG12-GA-2012-334259), Alzheimer's Forschung Initiative.
References (93)
- et al.
Functional-anatomic fractionation of the brain's default network
Neuron
(2010) - et al.
The role of default network deactivation in cognition and disease
Trends Cognitive Sciences
(2012) - et al.
Relationships between years of education and gray matter volume, metabolism and functional connectivity in healthy elders
Neuroimage
(2013) A fast diffeomorphic image registration algorithm
Neuroimage
(2007)- et al.
Unified segmentation
Neuroimage
(2005) - et al.
Efficiency, capacity, compensation, maintenance, plasticity: emerging concepts in cognitive reserve
Trends Cognitive Sciences
(2013) - et al.
Cognitive reserve impacts on interindividual variability in resting-state cerebral metabolism in normal aging
Neuroimage
(2012) - et al.
Cognitive reserve modulates task-induced activations and deactivations in healthy elders, amnestic mild cognitive impairment and mild Alzheimer's disease
Cortex
(2010) - et al.
White matter hyperintensities and cognition: testing the reserve hypothesis
Neurobiol. Aging
(2011) - et al.
Quantification of the impact of a confounding variable on functional connectivity confirms anti-correlated networks in the resting-state
Neuroimage
(2014)