Is Alzheimer’s Disease a Systemic Disease?

Abstract

Although Alzheimer’s Disease (AD) is the most common neurodegenerative disease, the etiology of AD is not well understood. In some cases, genetic factors explain AD risk, but a high percentage of late-onset AD is unexplained. The fact that AD is associated with a number of physical and systemic manifestations suggests that AD is a multifactorial disease that affects both the CNS and periphery. Interestingly, a common feature of many systemic processes linked to AD is involvement in energy metabolism. The goals of this review are to 1) explore the evidence that peripheral processes contribute to AD risk, 2) explore ways that AD modulates whole-body changes, and 3) discuss the role of genetics, mitochondria, and vascular mechanisms as underlying factors that could mediate both central and peripheral manifestations of AD. Despite efforts to strictly define AD as a homogeneous CNS disease, there may be no single etiologic pathway leading to the syndrome of AD dementia. Rather, the neurodegenerative process may involve some degree of baseline genetic risk that is modified by external risk factors. Continued research into the diverse but related processes linked to AD risk is necessary for successful development of disease –modifying therapies.

1. Introduction

Alzheimer’s disease (AD) is the most common form of dementia, affecting nearly 10% of individuals over the age of 65 and nearly 50% of those over the age of 85.[1] Increasing longevity in the population combined with the high incidence of AD in older adults will only exacerbate the societal and economic impact of AD in coming years. The neuropathological hallmarks of AD include amyloid plaques and neurofibrillary tangles which are present on microscopic examination of the brain. These neuropathological changes are accompanied by accelerated atrophy in the brain’s gray matter cortex, reflecting loss of neurons, in areas such as the hippocampus and parietal lobes. Ultimately, both gray and white matter abnormalities are observed.[2] The earliest clinical features of AD include short term memory impairment and executive dysfunction corresponding to neurodegeneration in areas that mediate these functions.[3]

AD is classically viewed as a primary neurodegenerative process. In its terminal phases, however, it is well-known that AD patients have physical decline and thus the AD process quite clearly is associated with systemic manifestations that extend beyond the CNS. This physical decline is undoubtedly driven to some extent by the progressive functional and behavioral decline associated with the CNS degeneration.[4] On the other hand, physical decline is observable to varying degrees in the earliest stages of the disease, prior to the presence of significant functional and behavioral decline that clearly underlies some of the physical manifestations seen late in the disease.[5]

The presence of physical or systemic manifestations of AD early in the disease, or even before the onset of clinically recognizable symptoms, suggests that physical decline may not simply represent a secondary result of the CNS pathological process. In fact, studies have long suggested that abnormalities in metabolic and biochemical processes described in AD brains are also present in peripheral cells such as skin fibroblasts derived from AD patients.[6, 7] Individuals with AD also have mitochondrial dysfunction evident in both the CNS and periphery [8] (for example, lymphocytes, [9]) suggesting that pathological processes may co-exist in both brain and non-neural tissues. There remains uncertainty regarding the causal relationship between these variables. To what extent the systemic changes represent an effect of a CNS process, contribute causally to the CNS disease, (i.e., reverse causation) or reflect a biological process that is present in both body and brain remains unclear.

This review sets out to examine these systemic manifestations of AD through the lens of three hypotheses asserting different cause and effect relationships: 1) systemic processes drive CNS dysfunction (for instance, as risk factors for brain dysfunction and AD), 2) AD brain processes drive systemic manifestations (i.e., downstream effects), and 3) a common underlying biological process is present both peripherally and in the CNS suggesting a systemic etiological process. We will also review the concept that AD is perhaps a multifactorial disease that affects both CNS and systemic processes.

2. Alzheimer’s disease and whole body changes: the chicken or the egg?

Association studies clearly demonstrate that patients with AD have a number of systemic (i.e., non-CNS) manifestations that accompany the CNS dysfunction that defines AD. Risk factor studies suggest that the neurodegenerative process may be instigated or exacerbated to some degree by peripheral processes. On the other hand, some of the peripheral manifestations may be the downstream result of AD processes, mediated by dysfunctional CNS control of peripheral processes or through behavior changes (i.e., reduced physical activity, forgetting to eat, etc.) that result in systemic manifestations. The primary goal of this review is to summarize evidence in favor of these possibilities and examine a third possibility that a systemic underlying factor may be common to both CNS and peripheral dysfunction associated with the clinical AD syndrome.

2.1 Systemic Processes Contribute to AD

A large number of apparent risk and protective factors have been identified that appear to influence an individual’s long term risk of developing AD. Some risk components are likely genetic in nature, although additional factors likely involve processes that originate outside of the CNS, such as diabetes, obesity, and physical inactivity. The precise mechanisms of their influence on AD risk likely include broad systemic effects that presumably transfer to the brain and either influence the initiation of disease processes or exacerbate the CNS dysfunction underlying the disease. Here, we review evidence that systemic factors may contribute to the initiation or exacerbation of AD.

2.1.1. Physical Activity

Physical exercise results in broad physiologic adaptations including improvements in cardiovascular fitness, vascular health, metabolic profile (reduced body fat, increased insulin sensitivity) and body composition (increased lean mass and bone density).[10] Increasing evidence suggests that exercise not only improves general health but positively impacts brain health through a number of potential mechanisms. Increased physical activity decreases AD risk [11] and has been postulated to have a trophic effect on the brain, particularly the hippocampus. For instance, exercise is associated with increased brain-derived neurotrophic factor (BDNF) [12] and other important neurochemicals, [13] supporting a role of exercise in brain growth and survival. Exercise appears to stimulate neurogenesis[14] as evidenced by increased counts of new neurons in adult animals on an exercise regimen. Broadly, exercise may modulate vascular risk factors (atherosclerosis,[15] heart disease,[16] stroke,[17] diabetes [18–23]) that place an individual at risk for dementia, vascular dementia, and AD. More specifically, studies have shown increased inflammation in AD,[24] and exercise decreases systemic inflammatory markers.[25]

Numerous clinical studies suggest a relationship between physical activity and risk of dementia in late life. Cross-sectional studies suggest physical activity is positively associated with cognition, particularly executive and visuospatial function.[26–30]. Multiple longitudinal studies report a relationship between self-reported exercise and cognitive decline, [31–36] and overall physical activity in midlife or later life is associated with a reduced risk of developing AD in late-life.[37, 38] These longitudinal studies suggest that systemic benefits of physical activity may modulate positive cognitive outcomes. This line of thought is further supported by results from intervention studies that have shown improvements in cognitive outcomes following exercise. [39–43] We and others have shown that physical activity and fitness levels are associated with larger brain volume.[42, 44, 45] This relationship may be mediated by exercise effects on neurotrophic factors such as BDNF. Serum levels of BDNF are positively correlated with hippocampal volume[46] and exercise acutely increases hippocampal levels of BDNF in animals [47] while blocking BDNF function ameliorates exercise-induced improvement in cognitive function.[48]

Most of the data supporting a link between exercise and brain health comes from studies of aerobic exercise – walking is the most common form of physical activity for older adults – and little data exists on the role of resistance exercise (i.e., weight lifting) in promoting brain health. There is evidence that resistance exercise may be important in preventing age- and AD-related cognitive decline. For instance, several small studies have found that resistance training is associated with modest cognitive benefits in those with [49] and without cognitive impairment.[43] Association studies suggest that reduced muscle strength is a risk factor for developing AD [50] and that lower strength is associated with greater cognitive decline.[51] These association studies however, cannot assess the cause and effect relationship of muscle strength with cognitive decline and it remains unclear if the declines in muscle strength causally influence AD processes or are a consequence of the early disease process. In fact, AD is associated with measurable changes in body composition, including lean mass and bone density, [52–54] suggesting that body composition changes are an early systemic manifestation of AD rather than factors that exacerbate or initiate the disease process. This will be reviewed in more detail below. Nevertheless, there is strong evidence that the physiologic adaptations (increased cardiorespiratory fitness, metabolic profile, increased muscle mass) from exercise and physical activity may result in beneficial brain effects that result in a lower risk of AD and dementia.

2.1.2. Type 2 Diabetes

Another peripheral process linked to risk and progression of AD is impaired glucose metabolism. Numerous epidemiologic studies have shown that diabetes and insulin resistance are strong risk factors for cognitive decline and AD, [55–60] and we and others have shown that impaired glucose metabolism is associated with increased progression from mild cognitive impairment to AD. [61, 62] Moreover, clinical studies using FDG-PET have demonstrated that decreased glucose metabolism occurs very early in AD brain and is predictive of disease diagnosis.[63, 64]

Several potential mechanisms may link insulin resistance and brain function. Insulin can cross the blood brain barrier, [65] where it likely modulates several processes including neurotransmission[66–68] cell survival, [69] and amyloid trafficking. [70] Insulin signaling deficits are present in AD brain post-mortem, [71, 72] although the temporal relationship between peripheral insulin resistance and CNS insulin resistance is uncertain. Elevated peripheral insulin, which precedes and often accompanies diabetes, is associated with an increased risk for dementia, [73] although high insulin levels may actually be protective in AD by compensating for impaired insulin signaling, known to occur in cognitively-impaired individuals. [74, 75] Finally, increased glycated hemoglobin (HbA1c), impaired fasting glucose, impaired glucose tolerance, and homeostatic model assessment of insulin resistance (HOMA-IR) have all been associated with impaired memory performance or longitudinal cognitive decline in nondemented adults.[76–79]

Although it is well-accepted that diabetes increases AD risk, the relationship between insulin resistance and amyloid pathology is somewhat more controversial. As mentioned, insulin can increase amyloid efflux from the cell, [70] and insulin and amyloid compete for receptor binding and are even degraded by a common enzyme.[80, 81] However, using advanced imaging techniques, one very recent study has indicated no relationship between insulin resistance and amyloid during life, [82] and supports previous work that showed diabetic individuals did not have more brain amyloid at autopsy compared to controls.[83] It is thus possible that individuals with metabolic impairment are simply more vulnerable to the effects of amyloid aggregation than healthy individuals, or perhaps that aggregation of a protein other than amyloid-beta affects these individuals. In fact, autopsy data indicate that amylin, a peptide produced in the pancreas and co-secreted with insulin, aggregates in the brain of individuals with both AD and vascular dementia independent of amyloid-beta deposition.[84] It has also been shown in animal models that the receptor for amylin may modulate amyloid-beta’s effects on long-term potentiation.[85] However, further studies are needed to determine whether amylin plays a role in cognitive decline and AD.

2.1.3. Obesity and lipid metabolism

Obesity is an important risk factor for dementia and AD. The relationship between obesity and dementia risk seems to peak at mid-life [86, 87] and in a meta-analysis was shown to occur independently of diabetes diagnosis.[88] Higher BMI in midlife is associated with structural brain changes [89, 90] and increased risk of cognitive decline and AD in late life. [91, 92] Thus, it is possible that dysregulation of systemic metabolic processes related to obesity and lipid metabolism may also affect late life AD risk.

The relationship between midlife obesity and AD risk may be modulated indirectly through vascular mechanisms (discussed below). However, a more direct potential link between obesity and cognitive decline involves lipids. Dyslipidemia occurs frequently in obese individuals,[93] and is characterized by increased levels of low density lipoprotein (LDL). In culture, oxidized LDL is associated with increased formation of “lipid rafts,” [94, 95] which are groups of molecules that change the fluidity of the plasma membrane and are integral to cell signaling. Interestingly, the processing of amyloid precursor protein to form amyloid-beta depends upon dynamic interactions with these microdomains [96]. In fact, recent work has shown that palmitoylation of APP increases amyloid processing through targeting of APP to lipid rafts.[97]

Lipid raft formation directly affects recruitment of signaling proteins and receptors to particular regions of the membrane. Cholesterol, sphingomyelin, and ceramide are important components of lipid rafts, and reduced sphingomyelin and increased ceramide levels have been observed in AD plasma [98]. Very recently, subjects in the middle and highest tertiles of ceramide at baseline had a 10 and 7.6-fold increased risk of AD, respectively.[99] However, another study indicated that ceramide only predicted cognitive decline and neuronal loss in subjects with MCI. [100] Thus, the effect of disease stage on the relationship between ceramide and cognitive function is not well understood. Ceramide has been shown to increase amyloid-beta generation,[101] Apolipoprotein (APOE) binding,[102] and APOE secretion, [103] and sphingomyelinase (which generates ceramide) may be involved in cell death.[104, 105] Studies have also shown that ceramide can play a role in insulin resistance and mitochondrial dysfunction (reviewed in [106]), making this molecule an intriguing player in multiple systemic processes that have been linked to AD.

Although a number of studies suggest that obesity is a risk factor for dementia and AD, this risk effect appears to be modified, and perhaps reversed, by age. Studies in older adults, rather than middle aged adults, suggest that a higher BMI may be associated with a lower risk of cognitive decline, dementia, and AD. [107–109] As discussed below in more detail, these observations suggest that obesity may influence long term risk of AD but that the early or preclinical stages of AD may be associated with weight loss.

2.1.4. Inflammation and Cytokines

A number of studies suggest inflammatory processes may play a role in AD. Although increased markers of inflammation are observed in AD brain postmortem, [110] these likely reflect local neurodegenerative processes occurring in brain. Thus, studies have examined if plasma biomarkers can predict cognitive decline and AD risk. Tumor necrosis factor alpha (TNF-α) has been associated with cognitive decline, [111] and both TNF-α and another inflammatory molecule, IL-1β, have been associated with increased AD risk. [112] Inflammatory processes are actually a common link between many AD risk factors discussed in this section: TNF-α, for instance, has been shown to be increased in T2D and obesity, and decreased with exercise,[113] although not all studies are consistent. Interestingly, a very recent large study that analyzed data from 3 independent cohorts found that only 4 plasma analytes (APOE, B-type natriuretic peptide, C-reactive protein, and pancreatic polypeptide) were linked with MCI or AD diagnosis.[114] One of these analytes, C-reactive protein, is known to be particularly responsive to inflammatory processes.

Although some studies have linked peripheral inflammatory markers to AD, clinical trials to reduce inflammation have produced contentious findings. For instance, in cross-sectional and population-based cohort studies, use of anti-inflammatory drugs is associated with reduced AD risk.[115–117] However, clinical trials have failed to show that anti-inflammatory therapy prevents AD [118] or improves cognitive function in either AD subjects or individuals with family history of AD. [119, 120] Safety concerns were often noted, and one trial was halted.[121] In summary, although peripheral inflammation has been observed in AD, the degree to which inflammation drives brain changes is unclear and at the present time there is little clinical evidence that inflammation is an efficacious target for AD prevention or treatment.

2.2. AD Drives Systemic Changes

It is well known that the brain mediates a variety of peripheral processes through mechanisms such as the autonomic nervous system and motor circuits. For instance, the CNS modulates bone health through autonomic output from the hypothalamus,[122] a central regulator of a number of peripheral metabolic processes. Additionally, as AD brain dysfunction progresses the associated functional declines result in reduced levels of physical activity which may in turn result in body composition changes (i.e. increased fat mass and reductions in lean mass and bone density). Thus, CNS processes can influence peripheral processes both directly (i.e. autonomic output) or indirectly through behavioral and functional changes (i.e., reduced physical activity, forgetting to eat). This section reviews evidence of peripheral effects that may be mediated or modulated by the AD neurodegenerative process.

2.2.1. Body Composition

As discussed, obesity is a clear risk factor, particularly in midlife, for the future development of AD. However, there is substantial evidence that changes in body composition, such as weight loss, occur in the earliest or even preclinical stages of AD. These findings suggest that as early CNS manifestations of AD occur in the brain (i.e., plaques and tangles) there are co-occurring systemic changes associated with the onset of disease.

As noted above, while higher body mass index (BMI) is associated with increased dementia risk, [91, 92] this risk appears to attenuate or reverse in older adults where higher BMI is associated with a lower risk of cognitive decline, dementia, and AD. [107–109] In fact, studies using sensitive measures of body composition suggest that changes in lean mass (i.e. muscle mass) and bone density may be among the earliest manifestations of the AD clinical syndrome. We have found that individuals in the early stages of AD have reduced lean mass [52] and lower bone density [53] than nondemented controls. Bone density has been correlated specifically with measures of hypothalamic atrophy in AD, while reductions in lean mass and bone density were both associated with greater whole brain atrophy and cognitive decline. [54] Reductions in lean mass may be attributed in part to functional impairment in individuals with AD, which can result in feeding difficulties, especially in subjects with aged caregivers, [4] although our studies included individuals in the earliest stages of AD when these types of difficulties are not overtly present.

Although functional impairments occurring in later stages of AD may contribute to the observed decrease in body weight, other research suggests that weight loss occurs before significant functional decline has begun. For instance, weight loss has been shown to occur prior to development of AD.[5] In a prospective study that followed following elderly individuals for 20 years, individuals who were diagnosed with AD at follow-up had lost more weight since baseline than individuals with normal cognition. In this case, weight loss occurred prior to AD diagnosis, indicating that this effect was likely not due to functional or behavioral changes impacting nutrition.[123] Lower BMI is associated with faster cognitive decline over one year in individuals with MCI,[107] consistent with studies suggesting that weight loss in elderly individuals may be an early systemic manifestation of the AD process. [124–128] An inverse relationship has been observed between BMI and AD biomarkers in both normal and cognitively-impaired subjects, with the relationship most strongly evident in individuals with MCI.[129] MCI is a heterogeneous pathological state, suggesting that individuals with MCI who are normal or low weight (BMI 18.5–25 kg/m²) are more likely to have amyloid-based cognitive impairment compared to those who are overweight (BMI > 25 kg/m²).[129]

Interestingly, a relationship between BMI and AD neuropathology has also been observed in cognitively-normal elderly subjects. We and others have shown that neuropathological changes of AD found at autopsy are associated with low and declining body mass index (BMI).[129, 130] Given that elevated BMI at midlife is a risk factor for AD, it is possible that low BMI in late life is somehow a consequence of the disease process. However, this relationship may also suggest that there are multiple etiologies leading to AD. For instance, it is possible that individuals with elevated BMI may exhibit more vascular-related pathology and lower amyloid neuropathology for a given level of cognitive function, while subjects with lower BMI may be more apt to exhibit more classic AD neuropathology in the absence of vascular disease.

2.2.2. Physical function and fitness

Motor slowing has been observed in subjects with early AD,[131] indicating that motor dysfunction may be a very early manifestation of disease. Cognitively-impaired subjects have been shown to exhibit greater decline in strength and performance on tests of physical function compared to controls,[132] and it is reported that cognitive decline predicts decline in upper muscle strength.[133] Along these lines, we have found that AD subjects also exhibit reduced VO₂^peak (independent of dementia severity or physical function decline), and this reduction is correlated with brain atrophy. [44, 45] Similarly, our imaging studies suggest that decreased aerobic fitness is associated with hippocampal atrophy in early AD.[134] It is possible that declining functional capacity, especially in later stages of the disease, has a detrimental effect on physical fitness in these individuals. However, mild decreases in VO₂^peak have been observed by us even in the earliest clinical stages of the disease prior to the emergence of clinically significant physical function impairment. This suggests that motor dysfunction may not simply be the result of progressive CNS dysfunction but an early systemic manifestation of the disease.

2.3. Common Peripheral and CNS Etiological Processes

Data from the prior sections suggests that systemic processes can influence AD risk while AD-related CNS dysfunction can influence whole-body health. In this section, we will review evidence suggesting that abnormal physiologic processes may underlie both whole body and brain health. For instance, mitochondrial dysfunction is one underlying factor that precedes both CNS neuropathological manifestations and peripheral metabolic dysfunction often observed in AD. Thus, a common underlying etiological process may mediate some of the apparent co-occurring decline in both the body and the brain that is associated with the clinical syndrome of AD.

2.3.1. Genetic factors

Genetic factors are likely to play a common underlying role in peripheral and central dysfunction associated with AD, although the functional actions of many of the risk genes for AD are not fully known. The most commonly identified gene that increases sporadic AD risk in roughly 40% of individuals is Apolipoprotein ε4 (APOE ε4). APOE is involved in lipid transport and cholesterol metabolism within the cell.[135] Interestingly, differences in regional white and gray matter are detectable in APOE ε4 carriers from infancy.[136] Although the precise mechanisms are still not well understood, we have previously discussed evidence that aberrant lipid metabolism may play a role in AD risk. We and others have found that the APOE ε4 allele is associated with decreased cognition, gray matter volume in memory areas (the hippocampus), white matter tract integrity, and increased magnetic resonance imaging markers for cardiovascular disease [137–139]. Furthermore, decreases in cerebral glucose metabolism are a known biomarker for AD, [140] and cognitively-normal, middle-aged APOE ε4 carriers have AD-like changes in cerebral glucose metabolism, [141, 142] with a possible gene-dose effect. [143]

A handful of studies using functional magnetic resonance imaging of the default mode network (DMN) have also shown differential oxygen uptake in the brain at rest in young APOE ε4 carriers, indicating differences in brain metabolic function early in life. [144–146] Some have argued that default mode network changes may more closely represent actual brain oxygen consumption [147, 148] and thus mitochondrial function. In fact, cytochrome oxidase activity, a marker of mitochondrial bioenergetics, has been measured directly in the brains of young adult APOE ε4 carriers. These individuals exhibited mitochondrial dysfunction decades before they would likely have any cognitive symptoms.[149] This indicates a potential underlying factor for the effects of genotype and mitochondrial bioenergetics, which will be discussed later. Moreover, genetics may even affect the responsiveness of individuals to interventions that aim to reduce AD risk: the finding that higher leisure-time physical activity decreased AD risk in late life was strongest in individuals who were APOE ε4-positive. [150] However, recent studies that have examined insulin resistance in APOE ε4 carriers have not shown any effect of genotype, [151, 152] suggesting that the relationship between insulin resistance and AD may be through a separate mechanism.

Although APOE ε4 is by far the most widely-recognized genetic risk factor of late-onset AD, other genes have also been linked to AD. To date, 660 candidate genes for AD risk have been identified, although results are inconsistent between studies. The development of genome-wide association studies (GWAS) has greatly improved AD genetic knowledge.[153] GWAS with fewer than 1000 case or control subjects have implicated novel AD risk single nucleotide polymorphisms (SNPs) in GOLPH2, GAB2, and PCDH11X genes[154–157]. More powerful, higher number case and control GWAS have identified or replicated novel AD risk SNPs in BIN1, CLU, CR1 and PICALM genes, [158–165] and a recent meta-analysis of GWAS identified a total of 20 genetic susceptibility loci (11 new genes, in addition to 9 previously-identified risk genes).[157]. However, interpretation of GWAS findings to reveal disease-relevant biological mechanisms remains a challenge because the genetic architecture of AD is incomplete [166]. Even the most robust genetic associations appear to explain only a small portion of the disease burden in the population, with the majority of the heritable component of the disease unexplained.

2.3.1. Mitochondrial Dysfunction

There is a large body of evidence that mitochondrial dysfunction, and perhaps energy failure, plays a central role in AD pathophysiology.[167, 168] Mitochondrial deficits in AD are not isolated to neurons, but occur systemically. [8, 169] Mitochondrial dysfunction likely contributes to insulin resistance [170] and has been shown to occur in pre-diabetic animal models. [171] Very recently, mitochondrial dysfunction at the molecular level was shown to lie upstream of pancreatic beta-cell death and was linked to development of diabetes in mice.[172] Moreover, muscle contraction, as occurs during exercise, has been linked to improvement in mitochondrial energy metabolism and to normalization of insulin signaling.[173] It is possible that mitochondrial dysfunction is one underlying factor that precedes both CNS neuropathological symptoms and contributes to peripheral metabolic dysfunction often observed in AD.

Family history studies also support a role for mitochondrial dysfunction in AD. In contrast to nuclear DNA, mitochondrial DNA is inherited maternally, and both specific mitochondrial haplotypes and maternal family history are linked to AD-related structural, cognitive, CSF, and metabolic biomarkers. [174–177] These findings of increased AD-related change in maternal lines of AD suggest that transmission of risk is preferentially found in maternal inheritance. This provides indirect evidence that mitochondrial function is related to manifestation of AD symptoms. Perhaps more intriguing than genetic risk, however, is that mitochondrial dysfunction accumulates throughout the aging process.[178–180] This suggests a role for both inherited and acquired mitochondrial dysfunction in modulating AD risk. In AD patients, overt markers of mitochondrial dysfunction have been consistently observed. For instance, activity of cytochrome c oxidase, an enzyme in the electron transport chain essential for energy production, is decreased in the brains of AD patients [169, 181–185] and in adult children with a maternal family history of AD.[186] Furthermore, mitochondrial DNA isolated from AD brain exhibits a loss of integrity, such as increased number of deletions and mutations.[187–189]

Interestingly, alterations in the association of mitochondria with other cellular compartments are also observed in AD and may further contribute to mitochondrial dysfunction. For instance, mitochondria are normally associated with the endoplasmic reticulum (ER) through physical ER connections called mitochondria associated membranes (MAMs). These structures play an important role in communication between the mitochondria and ER, linking them structurally and functionally. These membrane areas have the characteristics of previously discussed lipid rafts, and the activity of enzymes linked to AD, such as γ-secretase, is heavily enriched in MAM regions. [190, 191] MAMs are important for regulating processes such as phospholipid synthesis and calcium levels [192, 193] and may provide a potential link between processes such as apoptosis and synchronization of energy production and energy use via calcium signaling. [190, 193, 194] The function of MAMs is altered in AD, with consequences ranging from mitochondrial dysfunction to altered APP processing.[195] Loss of DNA integrity, dysregulated calcium homeostastis, and a failure to adequately match energy supply and demand due to downregulation of key enzymes may all contribute to mitochondrial damage. This evidence provides an additional potential link between mitochondrial function, lipid metabolism, and AD.

It is known that overproduction of reactive oxygen species (ROS) and subsequent oxidative stress plays a key role in mitochondrial dysfunction. [196] Interestingly, mitochondria themselves are the primary source of ROS production in the cell. [197] Under normal conditions, ROS serve important signaling functions,[198] but damaged mitochondria can overproduce ROS and increase cellular oxidative stress.[197] ROS generation is increased in MCI, [199] and oxidative stress can mediate AD pathophysiology through increased production and secretion of amyloid beta (Aβ). [200, 201] In fact, mitochondrial dysfunction may play a role in the downstream protein aggregation that serves as the neuropathological hallmark of AD. Aβ deposition initially occurs within the neuron, [202] and Aβ trafficking into mitochondria precedes plaque formation. [203] Cell culture studies have shown that functional mitochondria are necessary for Aβ to induce cellular toxicity.[204] Although the temporal relationship between impaired mitochondrial function and amyloid neuropathology in humans is not well understood, studies in transgenic mice suggest that mitochondrial dysfunction occurs prior to Aβ plaque formation. [205] Furthermore, Aβ is postulated to impair mitochondrial protein import, [206] and complex IV dysregulation in triple transgenic AD mice is dependent on Aβ,[207] providing multiple mechanisms for impaired mitochondrial function. These effects may trigger a vicious cycle where damaged mitochondria can generate Aβ and Aβ can then enter mitochondria and exacerbate damage. Interestingly, amyloid is imported into mitochondria by translocase of the outer membrane (TOM) machinery, [208] and TOMM40, which encodes the channel protein subunit of the translocase of the outer mitochondrial membrane (TOMM) complex [209] is in high linkage disequilibrium with APOE.[210] This makes it difficult to discern which gene actually contributes to AD risk.

2.3.2. Vascular mechanisms

The strongest modifiable risk factors for AD are also risk factors for vascular disease, including sedentary behavior, glucose intolerance, and obesity.[211, 212] Vascular pathologies are increasingly recognized as having an important role in late-life dementias given recent observations of the heterogeneity of associated dementia pathologies.[213] Although most studies attempt to describe vascular and AD pathologies as discrete syndromes, the clinical and neuropathological boundaries frequently overlap. Vascular-related injury often coexists with AD neuropathology at autopsy,[214] and the presence of cerebrovascular disease lowers the burden of AD neuropathological changes associated with a given level of cognitive impairment. [215, 216] Many individuals clinically diagnosed with AD actually exhibit a “mixed dementia” with both AD neuropathological changes and abnormalities indicative of vascular damage. [217, 218]

Mechanistically, vascular mechanisms may compromise cognitive function and contribute to AD in several ways. For instance, cardiovascular risk factors are linked to white matter lesions in elderly subjects.[219] White matter lesions are prevalent in both aging and AD,[216] and we have shown that cardiorespiratory fitness correlates with longitudinal brain atrophy in AD.[134] Moreover, many conditions comorbid with vascular disease, such as hypertension and diabetes, have been linked to dementia. High blood pressure has been observed in subjects with both AD and vascular dementia,[220] and high blood pressure during midlife is associated with increased late life AD risk.[221] The link between AD and diabetes has been previously discussed, but it is interesting that pancreatic dysfunction is linked to high levels of amylin, which aggregates in a manner similar to amyloid-beta and cause pancreatic beta-cell cytotoxicity.[222] Amylin also aggregates in the brain in both vascular dementia and AD, linking peripheral metabolic dysfunction with vascular disease and cognitive decline.[84] Both AD and vascular dementia have also been linked to decreased regional blood flow.[223] Decreased blood flow to the brain reduces the critical supply of glucose and oxygen to the brain that is necessary to sustain proper neuronal metabolism. Finally, vascular disease may be linked to impaired energy metabolism through mitochondrial function. In cardiac microvascular endothelial cells, high glucose induces apoptosis through induction of FoxO3a, [224] a transcription factor known to regulate mitochondrial gene expression and production of reactive oxygen species.[225] A similar mechanism may also hold true for brain endothelial cells.

Thus, there may be multiple paths to AD dementia, with genetic, mitochondrial, and vascular influences. It is intriguing that diabetics, for instance, have a higher risk for AD diagnosis and yet lower amyloid pathology postmortem.[217] There is thus evidence that vascular risk factors may lower the threshold for additional damage necessary to cause clinical expression of AD symptoms.

3. Additional considerations

Despite efforts to strictly define AD as a homogeneous CNS disease, there may be no single etiologic pathway leading to the syndrome of AD dementia. An individual’s baseline risk is likely determined by inherited nuclear and mitochondrially-encoded genes. Environmental factors, such as midlife obesity, insulin resistance, and inflammatory processes, likely modify this baseline risk. The contributions of genetic vs. environmental factors certainly vary from individual to individual in complex ways. Because environmental “risk modifiers” often interact (for instance, midlife obesity can lead to insulin resistance), environmental factors may play a more central role in the development of dementia in some individuals. These individuals may have lower levels of traditional AD neuropathological burden (i.e. amyloid plaques and neurofibrillary tangles) compared to genetically predisposed individuals who generally have higher levels of AD neuropathology. Vascular mechanisms and processes influencing cellular stress may influence the emergence and progression of the clinical manifestation of the dementia syndrome. Continued research into the diverse but related processes linked to AD risk is necessary, as individuals with genetic risk (and potentially more amyloid pathology) may benefit more from amyloid clearance drugs, whereas individuals who exhibit more vascular-related pathology may preferentially benefit from lifestyle interventions such as diet and exercise.

Limitations and forward progress

Although promising, there are limitations to many studies discussed in this article. Studies that are observational or cross-sectional, for instance, cannot be used to establish causality – this requires well-designed clinical trials. Because AD has a notably long asymptomatic time course, and many of the classical outcome measures (such as neuropsychometric tests) are not sensitive to the earliest signs of the disease, design of randomized clinical trials can be difficult and expensive. However, there has also been exciting progress in the field, notably in the development of advanced imaging techniques. These techniques can be used to determine individuals at risk for AD prior to cognitive impairment. For instance, ligands that can be used to visualize both Aβ and Tau using PET imaging in vivo will allow future randomized clinical trials to investigate lifestyle interventions, such as exercise, or pharmaceutical interventions that may affect brain metabolism, such as intranasal insulin, as potential modifiers of AD neuropathology in the preclinical stages of the disease.

4. Conclusions

Here we present evidence that 1) peripheral processes contribute to AD risk, 2) AD influences whole-body changes, and 3) common processes may modulate both peripheral and CNS manifestations of AD dementia. It is our view that AD is a multifactorial disease that affects both CNS and systemic processes. It is possible that there is no single “smoking gun” that will explain the etiology of AD, but that several interconnected processes, many of which are directly related to energy metabolism, together determine the cognitive trajectory of individuals that are initially at more or less risk from genetic factors.

Figure 1. Schematic of potential risk factors for AD.

We propose that a combination of nuclear and mitochondrially-encoded genes determine one’s baseline risk for AD. This baseline risk can be modified – either increased (+) or decreased (−) by a wide variety of environmental factors, (diet, exercise) socioeconomic factors (i.e. education) and inevitably by the aging process.

Highlights.

• Several diseases and risk factors linked to AD risk involve systemic metabolic dysfunction.

• The cause and effect relationship between these factors remains imprecisely defined.

• Common factors affecting both CNS and systemic processes contribute to AD

• AD likely represents a heterogenous disease with both central and peripheral manifestations