Growing evidence suggests that human cognitive capacity suffers continuous decline during the lifespan, starting in early adulthood. This decline is likely to reflect age-related changes in brain neurotransmission and structure, which are not uniform (frontal and temporal cortices being among the most affected)24
. The aging decline of neurotransmitter systems selectively impairs regional brain function (i.e decreases in dopamine neurotransmission are linked to decline in frontal function)25,26
and the deterioration of the axonal myelin sheath with age degrades the long-range connectivity of specific neural networks (reviewed by Madden27
). However there is also evidence that some networks might be preserved and it is speculated that these may help to compensate for age effects in brain function 28,29
. Similarly age related decreases in cognitive performance are not uniform, attention, memory and executive function (including working memory) are the most affected30
whereas some cognitive functions are relatively preserved (language, decision making)31,32
. Thus a better characterization of the effects of aging on brain functional networks is of value to help understand better the unique pattern of cognitive decline associated with aging and potential mechanisms for compensation.
Here we document aging and gender effects on short- and long-range FCD in 913 healthy subjects from the image repository “1000 Functional Connectomes” using a data-driven approach at 3-mm isotropic resolution. In addition to the previously described age-associated decreases in connectivity of DMN we also showed FCD-decreases in DAN regions as well as FCD-increases in somatosensory and motor cortex, amygdala and thalamus. Overall aging effects were more pronounced for long-range FCD than for short-range FCD. We also document significant gender differences in network and hub connectivity but showed no significant age by gender interaction effects on long- or short-range FCD.
The main hubs of long- and short-range connectivity were located in posterior ventral parietal/posterior cingulate (main short-range hub and prominent long-range hub) and occipital cortices (main long-range hub). This finding is consistent with previous MRI studies based on graph theory that reported that PC/VP and occipital cortices house prominent hubs of global8,12
functional connectivity, as well as with PET studies showing high metabolic rate of glucose in these posterior brain regions33,34
. The PC/VP hub is interconnected with all DMN regions17
that show synchronous signal fluctuations and high metabolism in resting conditions, and negative fMRI responses during cognition17,35,36
. The occipital cortex has the highest neuronal density among cortical regions37
and is essential for visual processing38
. Recently we have shown that the occipital FCD hubs are highly interconnected with an occipital-parietal network that shows overlap with the posterior dorsal attention network17
In average, the strength of long-range FCD hubs in the DMN decreased 6 ± 1% per decade of life and that of short-range connectivity 1.6 ± 0.3% per decade of life. The decreases were even more pronounced at the location of PC/VP (long-range FCD: 12 ± 2% per decade of life; short-range FCD: 4 ± 2% per decade of life). The sensitivity of the DMN to aging effects is consistent with reduced resting-state activity39
and disruption of cortical networks caused by amyloid deposition9
and potential gliosis40
in precuneus, retrosplenial and posterior cingulate cortices in older adults, and suggest that resting-state MRI acquisition could serve as a biomarker of aging in the human brain41
. Moreover, decreases in DMN connectivity with aging have been associated with impaired performance in working memory42
. However, the aging effects on long-range FCD hubs in the posterior cingulate differ from those on glucose metabolism, which have noted relative resilience of this brain region to the effects of aging and have contrasted this with the higher sensitivity of metabolic decreases in AD43
FCD-decreases with age were also observed in DAN (long-range: 3.4 ± 0.8% per decade of life; short-range: 1.7 ± 0.3% per decade of life) that includes prefrontal cortex, anterior cingulate (ACC) and posterior parietal cortices. This finding is consistent with the age-related decline in glucose metabolism documented by positron emission tomography studies, which shows that the most prominent decreases in metabolic activity in the healthy brain occur in prefrontal cortex (PFC) and in ACC25,43,44
. Decreases in FCD in the DAN could underlie the impairments in sustained attention, which is one of the most affected cognitive functions that occurs with aging45
. In healthy individuals we have shown that the age related decline in striatal dopamine D2 receptors is associated with age-related attention deficits, which in turn are associated with decreased activity in PFC and ACC46
. Moreover, in healthy controls the dopamine enhancing drug methylphenidate increased activation of the DAN while it increased deactivation of the DMN during performance of a visual attention task47
suggesting that the age related decline in dopamine neurotransmission may contribute to the decreased FCD in DAN,
Conversely, there were concomitant age-related increases in long-range FCD in other networks (somatosensory: 6 ± 1% per decade of life; cerebellum, thalamus, and amygdala: 20 ± 3% per decade of life) that are consistent with increases in brain activation4,5,48
and functional connectivity19
in prefrontal regions for older compared to younger individuals. These results conform well with findings from studies investigating the effects of normal aging on brain glucose metabolism most of which have shown that in addition to the posterior cingulate, the thalamus, cerebellum, striatum and limbic structures (including amygdala) are the least affected by aging (reviewed by Kalpouzos43
). Moreover a longitudinal study reported significant increases in regional cerebral blood flow with age in cerebellum, thalamus and putamen in addition to the significant decreases in prefrontal regions49
To our knowledge there are no prior reports that differentially evaluate the aging effect on short vs. long-range functional connectivity. The higher sensitivity of the long-range FCD, compared to that of the short-range FCD, suggests that long-range connections may be more vulnerable to aging effects than short-range connections. The physiological significance of this is unclear but we speculate that longer fibers may be more vulnerable to degeneration than shorter ones since proteins have to travel longer distances and hence are more vulnerable to energy deficits. While there is no current evidence in the human brain in peripheral neurons the length of the axons has been associated with their vulnerability to degenerative processes as is the case for amyotrophic lateral sclerosis and spastic paraplegias50
. Also aging has been associated, at least for the case of cholinergic fibers, with a decrease in fiber length51
. Similarly, postmortem studies have reported decreases in myelinated fiber length with age in the human brain52
(10% decrease per decade), which is likely to contribute to decreased long-range FCD with aging. However, a limitation on our interpretation for the greater vulnerability of long FCD to aging is the fact that long distance functional connectivity also relies on polysynaptic circuits, which may not necessarily entail longer fibers and since we did not have measures of fiber length (as per diffusion tensor imaging) we can not exclude this as a confound.
In this study we also find significant gender differences in FCD. Specifically short- and long-range FCD hubs were stronger in DMN and weaker in somatosensory network for females than for males. This is consistent with previous studies that documented gender effects in functional and anatomical connectivity19,53
. Higher long-range FCD in DMN in females than males is also in agreement with reports of higher cerebral blood flow and higher baseline glucose metabolism in DMN for females than for males54,55
and could help explain why female AD-patients show greater cognitive impairments than male patients for the equivalent reductions in regional brain metabolism56
. However in this study we failed to find significant gender by age interactions suggesting that aging effects on FCD may be similar for males and females.
The 1000 functional connectomes database includes limited phenotypic characterization of the individuals. Thus it was not possible to ascertain the functional significance of the aging effects on short- or long-range FCD. More specifically, we were unable to evaluate confounds such as vascular risk, hormonal changes, genetic factors and cognitive aptitude that may be related to network properties. Since vascular risk increases with age we speculate that this could contribute to the age-related changes in long-range FCD and that the later in turn may underlie some of the changes in cognitive performance that occur with aging. However studies are needed to demonstrate this. Nonetheless the consistency of the findings from resting FCD patterns16
makes it possible to generate standards that can be used in subsequent studies to compare with patient populations. The cross-sectional study design made possible evaluation of aging effects during the lifespan but limits the interpretability of the results. Note that the accelerated evolution of medical imaging technology poses challenges for longitudinal studies of brain function during the lifespan to ensure that the data is comparable as spatial and temporal resolution of the imaging tools change.
We evaluated aging and gender effects on short- and long-range FCD in 913 healthy subjects. DMN and DAN exhibited age-related decreases that were more pronounced for long-range FCD than for short-range FCD. Age-related increases in FCD were found in somatosensory cortex (in females), limbic regions and thalamus.