Functional regions within the brain become less distinct and interconnected in the elderly over time, especially in those networks related to attention span and cognition.
The finding, published by researchers at Duke-NUS Medical School in The Journal of Neuroscience, adds to current understanding of longitudinal decline in brain network integrity associated with aging.
“We currently live in a rapidly aging society,” said the study’s corresponding author, Associate Professor Juan Helen Zhou, a neuroscientist from the faculty of Duke-NUS’ Neuroscience and Behavioural Disorders program.
“Compared to cross-sectional studies, it is vital to understand brain changes over time that underlie both healthy and pathologic aging, in order to inform efforts to slow down cognitive aging.”
The human brain contains functionally segregated neuronal networks with dense internal connections and sparse inter-connectivity. aging is thought to be associated with reduced functional specialisation and segregation of these brain networks.
Joint senior authors Associate Professor Zhou and Professor Michael Chee, Director of Duke-NUS’ Centre for Cognitive Neuroscience, led the research team, collecting data from neuropsychological assessments and functional magnetic resonance imaging (fMRI) brain scans from a cohort of 57 healthy young adults and 72 healthy elderly Singaporeans.
Each elderly participant was scanned two to three times during a period of up to four years.
The neuropsychological assessments tested participants’ ability to process information quickly, focus their attention, remember verbal and visuospatial information, and plan and execute tasks. The fMRI scans measured how brain regions are functionally connected based on low-frequency blood oxygenation level fluctuations over time.
Participants were asked to relax with their eyes open and remain still as these were performed.
Dr. Joanna Chong, first author of the paper and a Ph.D. graduate from Associate Professor Zhou’s lab at Duke-NUS, developed approaches to convert the fMRI images into graphic representations that depict the inter- and intra-network connectedness of each individual’s brain.
She then compared differences in brain functional networks between the young and elderly participants, and in the elderly over time.
The team tracked changes in brain functional networks that affected specific cognitive abilities, such as goal-oriented thought and action, and choosing where to focus attention.
As one ages, these networks associated with cognition are less efficient in information transfer, more vulnerable to disturbance, and less distinctive.
“Overall, our research advances understanding of brain network changes over time, underlying cognitive decline in healthy aging,” said Associate Professor Zhou.
“This can facilitate future work to identify elderly individuals at risk of aging-related disorders or to identify strategies that can preserve cognitive function.”
Commenting on the study, Professor Patrick Casey, senior vice dean for research at Duke-NUS, stated, “aging is a significant risk factor for a variety of chronic diseases in people, including neurodegenerative and cerebrovascular diseases. Governments worldwide are concerned about the public health implications of increasingly aging populations.
Basic research such as this plays a vital role in informing efforts to help us stay healthy longer as we live longer lives.”
The researchers aim to next examine how various factors, such as genetic and cardiovascular risks, might influence aging-related changes in brain networks. By studying a larger group of healthy young, middle-aged and older adults, they hope to develop better ways to predict cognitive decline.
Changes in Cognition with Normal Aging
Studies of cognition across the life span are subject to several biases, some of which apply in general and some that are specific to study design.
General biases include recruitment bias and misclassification bias. Recruiting subjects for any clinical research study may be biased by which subjects are willing to enroll (recruitment bias).
For example, those who are too ill or have more limited social and financial support may find it hard to participate. Recruitment bias tends to underestimate the degree of cognitive decline seen with aging because only the healthiest or most advantaged are included in the study.
Misclassification bias is when a subject is misclassified in a research study, such as classifying a subject as normal when they are not. For example, by misclassifying a subject as normal when they actually have early signs and symptoms of a degenerative dementia, this subject’s cognitive test scores would overestimate the degree of cognitive decline attributed to normal aging and add a misclassification bias to the study.
Study design biases include cohort bias, practice effect (learning) bias, and attrition (survival) bias. Cohort bias occurs in cross-sectional study designs that compare groups of subjects (cohorts) in specific age groups on their performance on cognitive tests.
The cohort bias is the difference between groups that is not due to aging but is due to other differences, often unmeasured, between the age cohorts.
For example, when comparing a cohort of subjects that were born in the 1990s to subjects born in the 1940s, the two cohorts might differ significantly in nutritional variables, childhood educational experiences, exposure to environmental toxins or social stressors, knowledge of new technology, and other unmeasured variables.
These other factors may influence test performance over and above normal cognitive aging.
Longitudinal studies examine how an individual person performs on cognitive tests over time to understand how aging affects cognition.
One limitation of longitudinal studies is the practice effect (or learning) bias.
By testing subjects on similar test batteries over time, there is the potential for improvement in test performance due to a practice effect.
A second bias found in longitudinal studies is attrition or survival bias.
If there is selective attrition of subjects over time, the remaining subjects’ results may not be generalizable to other older adults.
For example, if a subgroup of patients selectively remains in the study (e.g., the healthiest or the best educated), their change in cognition may not accurately reflect the change in cognition with normal aging for many older adults. Both learning bias and attrition bias tend to underestimate the degree of cognitive decline seen with age.
The most common terminology, used to describe which cognitive abilities change with age and which do not, divides cognitive abilities into crystallized abilities and fluid abilities.
Crystallized abilities are the cumulative skills and memories that result from cognitive processing that occurred in the past, typically in the form of acquired knowledge.
Tests of general knowledge (e.g., reading comprehension, math, science), historical information, and vocabulary would reflect crystallized abilities.
Fluid abilities require cognitive processing at the time of assessment and reflect manipulation and transformation of information to complete the test.
Tests of fluid abilities require the subject to attend to one’s environment and process new information quickly to solve problems.
Multiple cross-sectional studies have shown that there is an improvement in crystallized abilities until approximately age 60 followed by a plateau until age 80, and there is steady decline in fluid abilities from age 20 to age 80 (see Fig. 1).
For example, there is a nearly linear decline in processing speed, a fluid ability, with a −0.02 standard deviation decline per year in one very large study.4
Cognitive abilities can be divided into several specific cognitive domains including attention, memory, executive cognitive function, language, and visuospatial abilities.
Each of these domains has measurable declines with age.6
For each of these domains, a subject must first perceive the stimulus, process the information, and then respond.
Both sensory perception and processing speed decline with age, thus impacting test performance in many cognitive domains.
For example, auditory acuity begins to decline after age 30, and up to 70% of subjects age 80 have measurable hearing loss.
Also, speech discrimination and sound localization decrease in advance age.
In addition to these change in sensory perception, there is a clear decline in processing speed in advancing age with older adults performing these activities more slowly than younger adults.4
This slowing of processing speed causes worse test performance on many types of tasks that involve a timed response.
The most noticeable changes in attention that occur with age are declines in performance on complex attentional tasks such as selective or divided attention.6
Selective attention is the ability to focus on specific information in an environment while at the same time ignoring irrelevant information.
Divided attention is the ability to focus on multiple tasks simultaneously, such as walking an obstacle course and answering questions.
Normal subject performance declines progressively with age on these more complex attentional tasks. However, simple attention tasks such as digit span are maintained in normal subjects up to age 80.6
Some aspects of memory are stable with normal aging, but there are consistent declines in new learning abilities with increasing age and some decline in retrieval of newly learned material.6
Immediate or “sensory memory” is stable with age, but tests that require subjects to exceed normal primary storage capacity (e.g., six to seven items) are more difficult for older adults.
Historical memories for public events and autobiographical memories (episodic memory) are relatively stable with advanced age, but the accuracy of source memory (i.e., accurately knowing the source of the known information) declines with age, as does the level of detail of recalled episodic memories. New learning, as measured by delayed free recall, also declines with age. Learning is further compromised in older adults if the test requires mental manipulation of the material to be learned (working memory) or if subjects must perform more than one activity while learning (divided attention).
Working memory requires active manipulation of material to be learned and declines with age.
Retention of newly learned information is relatively stable with advancing age, but retrieval of information may require more cueing or a recognition format to remain stable in advanced age groups.
Prospective memory, specifically remembering to perform intended action in the future (e.g., taking medication after breakfast), declines with age.
Procedural memories, such as remembering how to play the piano or ride a bike, are preserved with age.
Executive cognitive function involves decision making, problem solving, planning and sequencing of responses, and multitasking.
Each of these areas of executive cognitive function declines with advancing age.6 Executive cognitive function is particularly important for novel tasks for which a set of habitual responses is not necessarily the most appropriate response and depends critically on the prefrontal cortex.
Performance on tests that are novel, complex, or timed steadily declines with advancing age, as does performance on tests that require inhibiting some responses but not others or involve distinguishing between relevant and irrelevant information.
In addition, concept formation, abstraction, and mental flexibility decline with age, especially in subjects older than age 70.6
Speech and language function remains largely intact with advancing age.6
Vocabulary, verbal reasoning, and speech comprehension in normal conversation all remain stable into advanced age.
Speech comprehension in the setting of background noise and ambiguous speech content declines with age.
Speech comprehension involves both the peripheral nervous system’s sensitivity for perception and the central nervous system’s speech-specific cognitive abilities.8
These central nervous system cognitive abilities are especially important under less favorable listening conditions and are sensitive to changes with age.
Recent work suggests that aging-related changes in left frontal lobe structures correlate with performance on a speech-in-noise test.9
Verbal fluency, verbal retrieval, and some confrontational naming tasks show some decline with age.
Critchley observed that in advanced age, older adults were less verbose, more repetitive, and less specific in word choice in spontaneous speech when compared with young adults.10
There are age-related declines in aspects of visuospatial processing and constructional praxis.6
Visual recognition of objects, shapes, gestures, and conventional signs remains stable into advanced age.
However, visuoperceptual judgment and ability to perceive spatial orientation decline with age.
A person’s ability to copy a simple figure is not affected by age, but ability to copy a complex design (e.g., Rey figure) declines with age.
On standard IQ measures such as block design and object assembly, much of the declines with age are due to time, but when time is factored out, there are still declines in test performance with increasing age. On free drawing tasks, pictures drawn by older adults become more simplified and less articulated with age.
Age-Related Changes in Brain Structure and Function
The size of the brain decreases with age.3
In the past 20 years, our ability to quantify this atrophy has improved using structural brain imaging technology (computed tomography and magnetic resonance imaging [MRI]).
The brain is often divided into gray matter and white matter based on the brain’s appearance at autopsy.
Gray matter is used to describe the cerebral and cerebellar cortex and subcortical nuclei, each of which contains a predominance of cell bodies and dendrites.
White matter refers to regions of the brain with a predominance of myelinated axons that connect gray matter structures.
Not all brain areas develop atrophy equally with aging, but both gray and white matter regions are affected with aging.
The temporal lobes, especially the medial temporal lobe, which includes the hippocampus, also show moderate declines in volume with aging.
White matter volumes decline with age also.13
The greatest white matter volume losses are seen in the frontal lobe white matter and in major white matter tracts such as the corpus callosum.12
In addition to age-related decreases in volume of the white matter, there is evidence of a decline in white matter tract integrity with age, using MRI diffusion tensor imaging.14
It has been assumed that gray matter volume loss was due to neuronal loss, but with improvements in neuron-counting techniques, it is now clear that this is not the case.
Cortical neuronal loss is most notable in the dorsal lateral prefrontal cortex and hippocampus, and greater subcortical neuronal loss can be seen in the substantia nigra and cerebellum.
Age-related neurodegenerative diseases such as AD are associated with much greater loss of neurons, especially in the hippocampus and entorhinal cortex.16
In normal aging, a substantial number of neurons change in structure but do not die.
These aging-related structural changes to neurons include a decrease in the number and length of dendrites, loss of dendritic spines, a decrease in the number of axons, an increase in axons with segmental demyelination, and a significant loss of synapses.15 Synaptic loss is a key structural marker of aging in the nervous system.17 18
The number of neuronal synapses can now be measured using immunohistochemistry techniques that label presynaptic proteins, such as synaptophysin.
Using synaptophysin antibodies to quantify presynaptic terminals in the superior, prefrontal gyrus, a steady decline in synaptic number can be seen across the life span.
Results from dementia research suggest that symptomatic dementia occurs when there is a 40% or greater loss of neocortical synapses as compared with normal adults.17
Using the rate of change in cortical synapses seen with normal aging and the 40% synaptic loss threshold, Terry and Katzman predicted that dementia due to aging (senility) would occur at approximately age 130 without requiring the development of a disease state such as AD.19
They also discussed the concept of cognitive reserve in terms of cortical synaptic density and discussed how synaptic reserve, aging, and the development of a neurodegenerative disease could all impact when a person would cross the symptomatic threshold of 40% loss of cortical synapses and develop signs and symptoms of dementia.19 20 21
For example, those with a synaptic density deficiency at birth (e.g., low synaptic reserve due to neonatal hypoxic brain injury) would cross the 40% synaptic threshold earlier in life with the same rate of synaptic loss with aging.
Similarly, the development of a neurodegenerative disease such as AD would accelerate the rate of synaptic loss.
The age of dementia symptom onset would be determined by a combination of how close the person was to the 40% synaptic threshold at the time of disease onset and how quickly synapses were lost due to disease.
Alternatively, if there was a preventative lifestyle or treatment that slowed the rate of cortical synaptic loss caused by aging, then the 40% synaptic threshold would be reached later in life and this person would have greater synaptic reserve to compensate for degenerative disease-associated synaptic loss (see Fig. 2).
Age-related changes in the structure and function of synapses and changes in neuronal networks correlate with cognitive changes with aging.
Morrison and Baxter reviewed the aging changes that occur in the cortical synapse in the dorsal lateral prefrontal cortex, an area important in working memory and executive function, and the hippocampus, an area vital for learning and memory.22
They summarized the morphological and functional changes that occur at these synapses and how these changes may correlate with changes in cognitive function.
For example, in the dorsal lateral prefrontal cortex, there is a 46% loss of one subtype of cortical neuron dendritic spines (i.e., thin spines).
These spines are the most plastic spines and their loss causes a loss of dynamic plasticity in the ever-changing circuits that are important for cognitive flexibility, working memory, and executive cognitive function.
Interestingly, there is relative stability of a second type of dendritic spine (i.e., mushroom spine) in the dorsal lateral prefrontal cortex, which mediates more stable circuits that may be related to maintenance of crystallized cognitive abilities (i.e., experiential expertise). Resting state functional MRI imaging (rs-fMRI) can identify functional connectivity across distinct brain regions.14
A series of intrinsic connectivity networks have been identified, including networks important for memory, organization, and coordination of neuronal activity, priming of the brain for coordinated responses, and the default mode network (DMN), which is active in the absence of a task.
The DMN is thought to be important for memory consolidation.
Connectivity and network integrity appear to decrease in normal aging.14
In neurodegenerative diseases, such as AD, these declines and disruptions are accelerated, especially in the DMN, and can bias rs-fMRI studies of normal aging that include subjects with preclinical AD.23
Correlation of structural changes in the brain and measured age-related, cognitive changes have been modest and at times inconsistent, but inclusion of functional measures such as blood flow, glucose metabolism, and rs-fMRI or the combination of functional and structural measures can provide stronger correlations.14 24 25
Age-Associated Diseases and Cognition
A variety of factors can cause cumulative damage to the brain with age and produce cognitive impairments.
These factors include damage to the brain due to cerebral ischemia, head trauma, toxins such as alcohol, excess stress hormones, or the development of a degenerative dementia such as AD.
Degenerative dementias are the most common cause of significant late-life cognitive decline, but a combination of factors is common.
Community-based autopsy series of patients who died with dementia found that the most common cause of dementia was AD, followed by vascular dementia, and then dementia with Lewy bodies.26
However, mixed dementia or dementia caused by more than one pathology was very common.27
These same pathologic changes are very common in older adults without dementia.
In a large clinical-pathologic study of older adults without dementia combining participants from the Rush Memory and Aging Project and the Religious Orders Study, 100% had neurofibrillary tangles, 82% had amyloid plaques, 29% had macroscopic infarcts, 25% microscopic infarcts, and 6% had neocortical Lewy bodies.28
Because of the very common overlap of disease-associated pathology and cognitive decline in the elderly population, it is difficult to separate disease-related declines in cognition from those due the normal aging.
A recent larger study from the same longitudinal studies found that faster rates of cognitive decline were associated with AD pathology, macroscopic infarcts, and neocortical Lewy bodies, but the combination of all of these pathologies explained only 41% of the variation in rate of decline in this sample of older adults without dementia.29
Thus, these late-life diseases cause an acceleration of cognitive decline that results in the development of dementia in many patients, but some older adults without dementia do have cognitive decline not caused by these pathologic changes.
AD is the most common cause of cognitive decline in older adults.
The prevalence of clinically diagnosed AD increases exponentially with age.
For patients who develop AD, most first demonstrate a subtle decline in memory and new learning, followed by mild changes in executive cognitive function and later changes in language and visuospatial processing.
The onset of cognitive decline is subtle and hard to determine.
Progression is gradual and may be more apparent to family members than the patient.
Clinically, most patients first develop mild cognitive impairment (MCI), which is defined as a syndrome of cognitive complaints, measureable mild declines in cognition, but no change in functional abilities, including instrumental activities of daily living.
MCI can involve one or more cognitive domains, but memory domain-only MCI (i.e., amnestic MCI) is seen most commonly in patients who go on to develop AD.
If cognitive impairments continue to progress and the patient develops evidence of functional impairment caused by these cognitive impairments, then he or she would be diagnosed as having dementia.
If the patient meets the clinical criteria for AD, then he or she would be diagnosed with probable AD. Longitudinal studies suggest that the conversion rate from amnestic MCI to probable AD is ∼15% per year.33
The development of biomarkers for AD has improved our ability to understand the temporal sequence of changes that occur in the brain of someone with AD.
The development of amyloid positron emission tomography (PET) imaging (e.g., Pittsburgh Compound B/PiB) has allowed researchers to detect the presence of amyloid plaque deposition in living subjects.
Studies in subjects with genetic forms of AD demonstrate that amyloid plaques can be detected 15 years before clinical symptom onset and cortical amyloid deposition is the earliest marker of AD pathology.34
PET imaging of glucose metabolism uses fluorodeoxyglucose (FDG-PET) as a marker of neuronal activity and neurodegeneration.
Glucose metabolism, as detected by FDG-PET, declines in the posterior cingulate gyrus and the association cortices of the temporal and parietal lobes closer to the time of measureable cognitive decline.
Additional markers of neurodegeneration include volumetric MRI measurements of the hippocampus and measurements of cerebrospinal fluid levels of the protein tau.
These biomarkers begin to show changes before very mild cognitive symptoms appear and can be measured.
This classification system includes a determination of whether there is evidence of amyloid deposition, neurodegeneration, or both and whether cognition and function are normal or abnormal.
Patients with stage 1 disease have cerebral amyloidosis only; those with stage 2 disease have amyloidosis plus neurodegeneration but no cognitive decline; those with stage 3 disease have amyloidosis, neurodegeneration, subtle cognitive decline, but no functional decline; and those with stage 4 disease have amyloidosis and neurodegeneration, with measurable cognitive and functional decline.
The availability of these new biomarkers and the new classification system has been helpful to define preclinical AD for prevention trials; individuals with preclinical AD have evidence of amyloid deposition on amyloid PET imaging, but normal cognition and function (i.e., stage 1 and 2 AD), and AD biomarkers predict incident cognitive impairment in cognitively normal subjects followed longitudinally.37
Recently, Jack et al examined a large cohort of subjects with normal cognition age 50 to 89 for evidence of amyloid deposition using PET imaging and evidence of neurodegeneration using MRI measurements of the hippocampus and FDG-PET measurements of hypometabolism.38
He divided the cohort into four groups based upon biomarker results, specifically classifying amyloid imaging results into positive (A+) or negative (A−) and classifying combined neurodegeneration biomarker results into positive (N+) or negative (N−).
With this classification, the A−N− group would be classified as having normal biomarkers, the A−N+ group would have suspected non-Alzheimer pathology (SNAP) with neurodegeneration, and the A+N− and A+N+ groups would have evidence of different stages of AD pathology (i.e., stage 1 and 2 AD, respectively). In this study, the population frequency of A−N− (normal biomarkers) was 100% at age 50 and declined to 17% by age 89. The frequency of A+N+ (AD with neurodegeneration) increased to 42% by age 89.
The frequency of A−N+ (SNAP) increased to 24% by age 89. Thus, AD pathology with neurodegeneration and SNAP with neurodegeneration become increasingly more common with age and affected up to 66% of people by age 89, despite normal performance on cognitive testing.
More information: Chong, J., Ng, K., Tandi, J., Wang, C., Poh, J., & Lo, J. et al. (2019). Longitudinal changes in the cerebral cortex functional organization of healthy elderly. The Journal of Neuroscience, 1451-18. DOI: 10.1523/jneurosci.1451-18.2019
Journal information: Journal of Neuroscience
Provided by Duke-NUS Medical School