Individual differences in brain activity during cognitive tasks are partly driven by factors, such as aging and genetics. One of the key constructs that may explain age-related response to pathology is cognitive reserve (CR). This construct invokes both static and dynamic responses to pathological change, and implies varying levels of cognitive efficiency in the performance of tasks (Box 3
). The concept of CR may explain variability in the relationship between AD pathology and cognitive function, as individuals with high CR may tolerate substantial pathology before showing cognitive loss, whereas those with low CR may decline earlier. We postulate that one model for low CR is the presence of the Apolipoprotein E4 allele.
Box 3. Cognitive reserve
The concept of cognitive reserve (CR) invokes both static and dynamic processes that permit some individuals to remain cognitively intact despite substantial pathology [62
]. For example, individuals with larger temporal lobe volumes may better withstand the effects of Aβ deposition [63
]. Reserve, however, is likely a complex construct: in epidemiological studies it is related to education, occupation, socioeconomic status, social networks and lifelong participation in cognitive and physical activity [64
]. Such factors are difficult to disentangle and, taken together, are likely to have profound effects on the brain across the lifespan that affect cognition through multiple pathways.
We consider reserve to reflect lifelong patterns of behaviors, endogenous factors (including genetics) and exposure to environmental factors that have consequences for how the brain processes information. Furthermore, cognitive reserve may play different roles during pre- and post-amyloid plaque stages (Box 2
). Once Aβ deposition begins, cognitive reserve may mitigate detrimental effects by allowing high CR individuals to cope with more downstream neuronal dysfunction and loss than individuals with low CR. However, in pre-amyloid plaque stages CR could act to diminish Aβ production through better ‘neural efficiency’, a concept that is widely applied in interpreting fMRI data. For example, a polymorphism in the catechol-O
-methyltransferase (COMT) gene affects the rate of dopamine metabolism: individuals with the less active enzyme show greater neural efficiency, demonstrated as less brain activation required for executive tasks [67
]. Similar mechanisms of neural efficiency have been invoked to explain age-related differences in behavioral performance and fMRI activation on cognitive tasks [40
]. Even during development, low CR (in this case, linked to socioeconomic status) is associated with greater bilateral prefrontal recruitment during phonological processing [68
Evidence linking increased neuronal efficiency to reduced Aβ comes from a recent study demonstrating that individuals who participated in more physical exercise (which has been independently associated with improvements in network efficiency [69
]) showed less evidence of brain Aβ [70
]. The Nun study revealed that adolescent women with more complex autobiographical essays (and thus presumably high CR) showed less cognitive decline and Aβ deposition in old age [71
]. Animal data showing that transgenic mice reared in enriched environments deposit less Aβ [72
] are also consistent with this idea. Thus cognitive reserve could act differently in pre- and post-amyloid plaque stages reducing Aβ deposition early and mitigating its effects later.
Whereas high cognitive reserve may result in life-long patterns of efficient neural activation that might mitigate Aβ deposition, genetic factors may underlie neural differences that promote Aβ deposition. The most important known genetic polymorphism in Alzheimer’s disease is the apolipoprotein E gene (APOE), which plays a role in the transport of cholesterol and other lipids, and has 3 polymorphisms (APOE 2/3/4). The APOE4 allele has a dose-dependent effect on risk for AD and decreases the age of AD onset by more than 10 years [20
]. This polymorphism also has effects on neural activity that may ultimately lead to Aβ accumulation through mechanisms involving diminished neural efficiency and reserve.
Animal and human studies have converged to reveal lifelong neuronal deficiencies associated with APOE4. Specifically, mice expressing APOE4 show reduced synaptic plasticity and spine density [21
] and older cognitively normal human carriers show reduced glucose metabolism [23
] as well as impaired DMN deactivation during episodic memory (EM) encoding [24
]. DMN dysfunction has also been shown amongst normal elderly APOE4 carriers confirmed to have low levels of amyloid, suggesting an effect of APOE4 that is independent of fibrillar Aβ [25
Although there is evidence that suggests an impact of APOE4 on AD via Aβ and non-Aβ pathways [26
], it is difficult to separate these effects in human studies when Aβ measurements are unavailable. However, elevated Aβ deposition is uncommon prior to age 50 (in 3 studies, only 6 out of 60 carriers younger than 50 had evidence for detectable quantities of elevated Aβ [27
], while elevated Aβ amongst APOE4 carriers after age 50 is present across multiple studies [27
]). Based on this pattern, a reasonable demarcation into pre-amyloid and post-amyloid plaque stages is possible around age 50 in carriers.
Examination of APOE4 carriers in the pre-amyloid plaque stage reveals evidence for substantial detrimental neuronal effects. For instance, carriers under age 40 have reduced glucose metabolism (measured with fluorodeoxyglucose [FDG] PET) in the multimodal brain regions characteristic of Aβ deposition discussed above [31
] and also show impaired mitochondrial activity (measured in postmortem tissue) [29
]. Furthermore, structural MRI studies have revealed reduced cortical thickness in child carriers [32
] and smaller hippocampal volume in young adult carriers [33
]. Although this primarily occurs in the medial temporal lobe, thinning is also seen in more widespread cortical areas [32
]. These findings require more study as they have not been uniformly confirmed [34
] and their predominant presence in the medial temporal lobe, an area with minimal Aβ, is unexplained. Overall, however, these studies provide human evidence for an early detrimental effect of APOE4 that is unrelated to and likely to precede abnormal Aβ accumulation, suggesting that carriers may demonstrate alterations in brain structure and function consistent with low reserve.
These neural deficits are accompanied by alterations in brain activation during EM processes in cognitively normal APOE4 carriers [35
]. While these results fail to reveal a consistent pattern, a major confound is the difference in cohort age, which results in different proportions of individuals in pre- and post-amyloid plaque stages. Thus, there are little existing data on how brain activation differs between carriers and noncarriers over the lifespan.
In our proposed model, individuals in the pre-amyloid plaque stage would show increased brain activation, whereas individuals in the post-amyloid plaque phase may show additional increases (a secondary compensatory response) and ultimately decreased activation (). To our knowledge, there are 4 fMRI studies of EM encoding in APOE4 carriers safely within the pre-amyloid plaque stage. Although these studies differ in task design, 3/4 studies revealed increased activation in APOE4 carriers [36
] whereas 1 study did not (however, the design and analysis in this study was remarkably different from the 3 other studies, complicating direct comparisons [39
]). In one study that examined APOE4 carriers at different ages, there was an interaction between age and genotype on activation. Specifically, activation during memory encoding across multiple brain regions was increased in young APOE4 carriers (20’s) whereas activation in these same regions was decreased in elderly carriers (70’s) [37
]. This pattern is consistent with the trajectory presented in , suggesting that older APOE4 carriers have lost the ability to maintain high levels of activation and consequently show reduced brain activation. Together with studies using transgenic models, FDG-PET, and structural MRI, these results suggest that synaptic and structural alterations underlie the greater neural responses in APOE4 carriers during the pre-amyloid phase that, we hypothesize, in turn, precipitate Aβ deposition.