Influence of genetic and cardiometabolic risk factors in Alzheimer’s disease

Alzheimer's disease (AD) is a multifactorial neurodegenerative disorder. Cardiometabolic and genetic risk factors play an important role in the trajectory of AD. Cardiometabolic risk factors including diabetes, mid-life obesity, mid-life hypertension and elevated cholesterol have been linked with cognitive decline in AD subjects. These potential risk factors associated with cerebral metabolic changes which fuel AD pathogenesis have been suggested to be the reason for the disappointing clinical trial results. In appreciation of the risks involved, using search engines such as PubMed, Scopus, MEDLINE and Google Scholar, a relevant literature search on cardiometabolic and genetic risk factors in AD was conducted. We discuss the role of genetic as well as established cardiovascular risk factors in the neuropathology of AD. Moreover, we show new evidence of genetic interaction between several genes potentially involved in different pathways related to both neurodegenerative process and cardiovascular damage.


Introduction
World Alzheimer Report 2018 estimated that 50 million people were suffering from dementia worldwide in 2018, and this number is expected to reach 152 million in 2050. Alzheimer's disease (AD) is the most common type of dementia which accounts for 50-70% of all dementia cases. In 2019, AD was reported as the sixth leading cause of death among all age groups (Gaugler et al., 2019). The global cost of dementia is estimated to be US $1 trillion, which is forecast to double by 2030.
AD is characterized by the presence of extracellular neuritic (senile) plaques formed by the abnormal accumulation of amyloid-β (Aβ) peptide, and intracellular neurofibrillary tangles (NFTs) composed of hyperphosphorylated tau protein. During decades, these hallmarks of pathology, together with a huge body of preclinical evidence, led to the formulation of the amyloid cascade hypothesis that poses the brain deposition of insoluble Aβ peptide as a primary event of the disease, which in turns leads to NFT and progressive neuronal damage (Glenner and Wong, 1984;Hardy and Selkoe, 2002;Karran et al. 2011;Scheltens et al., 2016). Apart from the well-established genetic determinants or risk factors of AD such as mutations of presinilin pathway proteins or apolipoprotein E (ApoE), respectively, recent genome wide association studies (GWAS) showed over 20 genetic risk loci associated with the development of AD (Harold et al., 2009;Seshadri et al., 2010;Naj et al., 2011;Lambert et al., 2013;Tosto and Reitz, 2013). Moreover, many clinical experiences suggested a possible role of other risk factors including arterial hypertension, diabetes mellitus, obesity, smoking, alcohol, hypercholesterolemia, and hyperhomocysteinemia in AD development (Femminella et al., 2020). A complex interaction among human genome, cardiometabolic risk factors and neurodegenerative disease development is an emerging field of research with interesting preliminary data. Inflammation emerged as possible common feature of AD and cardiometabolic diseases as shown in animal model and genome studies (Pasqualetti et al., 2015).
The long preclinical course of this disease and the frustrating failure of any clinical intervention to reverse AD disease progression support the exploration of new pathophysiological hypotheses. In this review, we discuss the most recent advances in AD research focusing on genetic determinants of disease and the role of cardiovascular risks in AD development. Moreover, we report the new evidence that shows the genetic interaction between several genes potentially involved in different pathways related to both the neurodegenerative process and cardiovascular damage.
To identify relevant articles on genetic and cardiovascular risk factors related to AD development and progression, PubMed, Scopus, MEDLINE and Google Scholar databases were searched for studies published before June 2020. Searches on cardiometabolic and genetic risk factors of AD were restricted to articles published in the English language.

Human genome, AD and cardiovascular traits
The epidemiological link between Alzheimer's disease (AD) and cardiovascular traits has been well documented in several studies. A pathological overlap between AD and cardiovascular diseases, characterized by an increased prevalence of vascular lesions (especially white matter lesions and lacunes), was first observed along with the finding that cerebral vascular alterations may lower the threshold for the clinical expression of dementia in AD pathology (Snowdon et al., 1997). However, further studies suggested that the relationship between AD and cardiovascular traits is multifaceted and may share common physiopathological pathways (O'Brien and Markus, 2014).
Cardiovascular and lifestyle-associated risk factors are increasingly recognized as important for AD pathogenesis (Samieri et al., 2018;Broce et al., 2019). In settings, the Framingham Heart Study documented that subjects with more cardiovascular risk factors or at least one ApoE ε4 allele had a greater risk of developing dementia (Peloso et al., 2020).
Moreover, aging, ApoE ε4 allele, the presence of risk alleles from GWAS, and cardiovascular traits increased the 10-year absolute risk of all-cause dementia (Juul Rasmussen et al., 2020). This strong association between cardiometabolic traits and dementia could be firstly interpreted as a causative relationship if we included vascular dementia independently from genetic background. However, specific clinical studies also showed a strong correlation between cardiovascular or metabolic factors and AD (Bone et al., 2021). This association led to the formulation of the hypothesis of shared genetic influence between AD and cardiometabolic traits (Bone et al., 2021). Apart from the ApoE ε4 allele, few data are available on the possible genetic determinants shared by cardiovascular traits and AD. In this regard, genome wide association (GWA) studies highlighted the importance of specific loci associated to inflammatory responses or metabolic pathways which may also play a role in neurodegeneration (Broce et al., 2019). Bone et al. reported a GWA study aimed at detecting genetic loci potentially shared by AD and cardiometabolic traits. The authors documented three pleiotropic genetic regions previously associated with AD. Four pleiotropic loci that were novel for both AD and cardiometabolic factors were also reported (Bone et al., 2021). The possible pleiotropic genes discovered by single-tissue-colocalization analysis highlighted the roles of genes related to blood pressure and immune response in both AD and cardiometabolic functions. For example, by using Genotype Tissue Expression data they discovered more possible pleiotropic genes, including angiotensin converting enzyme (ACE). Other evidence suggested a genetic link between AD and type 2 diabetes mellitus (T2DM). A GWA study analysing AD and T2DM found that both diseases shared 395 SNPs involved in immunity, cell signalling, cellular processes, and neuronal plasticity (Hao et al., 2015).
Broce et al. reported data from a large GWA study aimed at estimating potential pleiotropy underlining the development of both AD and cardiovascular alterations. The study was aimed to discover SNPs associated with AD and one or more cardiovascular traits, such as body mass index (BMI), T2DM, coronary artery disease, waist/hip ratio, total cholesterol, triglycerides, low-density (LDL) and high-density (HDL) lipoprotein. Beyond ApoE ε4, the authors showed that the polygenic component of AD is enriched for lipid-associated risk factors. They highlighted a subset of cardiovascular-associated genes that strongly increase the risk for AD (Broce et al., 2019). Another study, analysing elderly participants in the Cardiovascular Health Study, showed that variants of the IL-1 gene were associated with the baseline cognitive status, supporting the link between dementia and proinflammatory status (Benke et al., 2011).

AD genetic factors
While the neuropathological changes in AD are well described, the aetiology and the underlying mechanisms are still poorly understood, even though it is now well documented that genetic mutations play a significant role in AD. Apolipoprotein E is a polymorphic protein implicated as a cholesterol carrier that supports lipid transport and injury repair in the brain (Liu et al., 2013). It primarily plays an important role in the microglia and astrocytes, which require de novo cholesterol synthesis. ApoE polymorphic alleles are the main genetic determinants of AD risk. In humans, the ApoE gene has three allelic variants, namely ε2, ε3, and ε4. The ε4 allele has been identified as a risk factor for AD for over 20 years (Corder et al., 1993) and carriers have a relative risk (RR) of ~3.00 compared with those carrying the more common ε3 allele, whereas the ε2 allele decreases AD risk (Yamazaki et al., 2019). Carrying ApoE ε4 has been consistently shown to be associated with a significant increase in Aβ deposition or a greater proportion of amyloid-positive individuals in normal elderly subjects.
The age at which 10-20% of the participants with normal cognition become amyloid-positive is approximately 40-50 years for ApoE ε4 carriers (Mishra et al., 2018). The effect of ApoE on amyloid deposition is dose-dependent, such that Pittsburgh compound B (PiB) uptake increases progressively with each additional ε4 allele Ossenkoppele et al., 2015). This effect is probably global, as it targeted the frontal cortex in one study but posterior regions in another study.
Moreover, the presence of the ApoE ε4 allele increases the prevalence of conversion from amyloid-negative to amyloid-positive and decreases the age of predicted amyloid-positivity by about 10-20 years in non-carriers versus carriers (Mishra et al., 2018). ApoE ε4 has also been shown to modify the relationship between amyloid deposition and cognitive function, increased Aβ load being associated with decreased performance (Yamazaki et al., 2016). It reduces glucose metabolism in normal ageing (Jagust et al., 2012). ApoE receptors play an important role in modulating APP trafficking and Aβ production. It has also been shown that there are isoform-dependent differences in Aβ deposition (ApoE ε4 > > ApoE ε3 > ApoE ε2) (Morris et al., 2010;Gonneaud et al., 2016). It has been suggested that ApoE plays a role in Aβ clearance (Christensen et al., 2010). ApoE ε3 has been shown to inhibit abnormal hyperphosphorylation and destabilization of the neuronal cytoskeleton in AD, whereas the C terminal truncated form of ApoE4 is neurotoxic and stimulates tau phosphorylation, leading to pre-NFTs. ApoE plays an important role in promoting synapse formation and synaptic plasticity (Yamazaki et al., 2016). ApoE ε4 is a cofactor that enhances the toxicity of oligomeric Aβ by directing it to synapses, thus providing a link between the ApoE ε4 genotype and synapse loss. ApoE modulates inflammatory and immune responses in an isoform dependent manner, and plays a significant role in neurotoxicity, lipid metabolism, mitochondrial dysfunction, and blood-brain barrier permeability (Yu et al., 2014). Triggering receptor expressed on myeloid cells 2 (TREM2) is transmembrane receptor of the immunoglobulin superfamily expressed on the plasma membrane of myeloid cells and microglia (Wolfe et al., 2018). Some experiences have shown that TREM2 binds to ApoE using TREM2-Fc fusion pulldown and that genetic variant of TREM2 may also increase the risk of AD (Wolfe et al., 2018). Genome wide association studies have identified over 20 risk loci associated with the development of AD (Kunkle et al., 2019). These loci/genes, including CLU (clusterin), ABCA7 (ATP Binding Cassette Subfamily A Member 7), MSA46A (membrane-spanning 4 A family), CR1 (Complement receptor 1), BIN1 (Bridging Integrator 1), CD33 (sialoadhesin molecule and a member of the immunoglobulin supergene family), PICALM (Phosphatidylinositol Binding Clathrin Assembly Protein), HLA-DRB5 (Major Histocompatibility Complex, Class II, DR Beta 5), EPHA1 (EPH Receptor A1), MEF2C (Myocyte Enhancer Factor 2 C), SORL1 (Sortilin Related Receptor 1) and INNP5D (Inositol Polyphosphate-5-Phosphatase D) (Harold et al., 2009;Seshadri et al., 2010;Hollingworth et al., 2011;Naj et al., 2011;Lambert et al., 2013;Tosto and Reitz, 2013) are listed in Table 1. Along with ApoE ε4 and TREM2 variants, apart from demonstrating the increased risk, these findings shed light on the different pathways which could be involved in the disease pathogenesis. What is apparent from GWAs is that these susceptibility loci converge on three main pathways: immune and complement systems/inflammatory response, endocytosis, cholesterol and lipid metabolism are strongly implicated in the disease pathogenesis (Guerreiro and Hardy, 2011) thus suggesting a possible overlap between the genetic background of cardiometabolic diseases and AD.
A recent GWA study identified 29 risk loci, implicating 215 potential causative genes, strongly expressed in immune-related tissues and cell types (spleen, liver and microglia) (Jansen et al., 2019). Moreover, another recent large GWA meta-analysis of clinically diagnosed LOAD including 94,437 individuals, confirmed the previously identified LOAD risk loci and showed five new genome-wide loci (IQCK, ACE, ADAM10, ADAMTS1 and WWOX) (Kunkle et al., 2019). ADAM10 and ACE were also identified in a GWA familial-proxy study of AD or dementia (Kunkle et al., 2019). More recently, Bellenguez et al. reported 42 new loci potentially associated to AD by a GWA study (Table 1). In particular, they identified 31 genes that were suggestive of new genetically associated processes, including the tumor necrosis factor alpha (TNF-α) pathway through the linear ubiquitin chain assembly complex (Bellenguez et al., 2022).
Genetic data have shed light on the AD disease process in several ways, suggesting that inflammation plays a primary role in its development, in contrast to the view that immune and inflammatory related processes are secondary to disease occurrence. The two genes (clusterin/ CLU and CR1) that code for proteins acting as regulators of the complement system are risk factors for the development of sporadic AD (Schjeide et al., 2011;Foster et al., 2019). Clusterin is a lipoprotein expressed in most mammalian tissues, which can interact with a variety of molecules. It is involved in several physiological processes, including inhibition of the complement system. Clusterin is a heterodimer of two 40 kD glycoprotein subunits found in many body fluids and tissues (Foster et al., 2019). Clusterin can modulate the MAC (membrane attack complex) and prevent the inflammatory response associated with complement activation following protein aggregation (Schjeide et al., 2011). Clusterin has been identified in secretory and intracellular forms and it has been attributed to several processes such as sperm maturation, complement inhibition, apoptosis, cancer promotion, and AD (Trougakos et al., 2009). More importantly, clusterin was found to bind soluble Aβ through a specific, reversible and high-affinity interaction in cerebrospinal fluid (Ghiso et al., 1993;Foster et al., 2019) to form complexes that are able to cross the blood-brain barrier by a high-affinity receptor-mediated process involving transcytosis (Zlokovic et al., 1996). AD individuals with the ε4 allele of ApoE have low brain levels of ApoE and increased levels of clusterin, possibly as an adaptive response as clusterin modulates Aβ clearance from the brain in concert with ApoE (Foster et al., 2019). Interestingly, increased levels of clusterin have been documented, not only in the AD brain, but also in other neurodegenerative diseases, including ALS (Grewal et al., 1999), multiple sclerosis (Ingram et al., 2014), transmissible spongiform encephalopathies (Sasaki et al., 2002), and Huntington's disease (Labadorf et al., 2015). In AD, clusterin has been found in amyloid senile plaques, while in alpha-synucleinopathies, clusterin has been documented in cortical Lewy bodies (LBs) (Sasaki et al., 2002). Conflicting results have been reported on the possible anti-amyloidogenic role of clusterin and some hypotheses have been proposed (Foster et al., 2019). The duality of clusterin features on amyloid toxicity was thought to depend on the molar ratio of clusterin and amyloid (Yerbury et al., 2007). Clusterin may prevent amyloid aggregation and toxicity only when there is an excess of amyloid while, on the contrary, there is an increase in amyloid formation (Yerbury et al., 2007;Foster et al., 2019). However, the specific role of clusterin on AD pathology is controversial by complexities for its biogenesis, the role of extra-vs. intracellular clusterin, and the number of functions (Bradley, 2020).
CR1 is a polymorphic protein that also acts as a negative regulator of the complement system by inhibiting both the classical and alternative pathways (Schjeide et al., 2011). Other genes have also been suggested to have a role in the pathogenesis of AD. BIN1, members of the MS4A gene family, CD33, ABCA7, CD2AP and EPHA1 may also potentially be related to the immune and inflammatory events seen in AD. BIN1 knockout mosaic mice have been reported to show reduced inflammation with aging (Pasqualetti et al., 2015).
Endocytosis is crucial for several cellular functions. Eukaryotic cells use multiple pathways for the endocytic entry of molecules (mostly lipids and proteins) through the plasma membrane. Endocytic pathway activation has been shown to be a prominent and early feature of neurons in vulnerable regions of the brain in sporadic AD. Both BIN1 and PICALM are directly involved in clathrin-mediated endocytosis, suggesting a primary role of this pathway in AD. BIN1 is also implicated in membrane dynamics, such as vesicle fusion and trafficking, specialized membrane organization and actin organization (Harold et al., 2009). PICALM recruits clathrin and AP-2 (adaptor protein 2) to the plasma membrane and, along with AP-2, recognizes target proteins. PICALM is also essential in the fusion of synaptic vesicles to the presynaptic membrane by directing the trafficking of VAMP2 (vesicle-associated membrane protein 2). Changes in the expression of either one or both of these genes may modulate neuronal function and behavior. ABCA7 is part of the ABC transporter superfamily, and this family of proteins is known to have roles in transporting substrates across cell membranes. ABCA7 is highly expressed in brain, particularly in hippocampal CA1 neurons and in microglia (Harold et al., 2009;Hollingworth et al., 2011). It is known to be involved in the efflux of lipids from cells to lipoprotein particles and, in this way, may interact with the effects of ApoE and clusterin in AD. It has also been reported to regulate APP processing, inhibiting Aβ secretion in cultured cells overexpressing APP.
In a recent GWA study, from the International Genomics of Alzheimer's Project Consortium, including over 7 million genotypes from around 17,000 Alzheimer's disease cases and 37,000 controls, (Baker et al., 2019) found three novel loci PPARGC1A, RORA and ZNF423 significantly associated to AD. PPARGC1A and RORA are genes related to circadian rhythm (an early disturbances of AD) (Baker et al., 2019). PPARGC1A is also associated with energy metabolism and the generation of amyloid-βplaques, while RORA is implicated in other metabolic functions. The ZNF423 gene encodes for an AD-specific protein network involved in DNA damage repair (Baker et al., 2019).
Although the effect of the single locus might be small, a polygenic risk score (PRS) can help evaluate the combined effects of gene variants. To date, several studies have used this polygenic approach to estimate the risk of AD progression and to evaluate the association of AD genetic risk with endophenotypes of the disease. Harrison et al. have demonstrated the association between a PRS and hippocampal thinning in healthy individuals (Harrison et al., 2016). Other studies have shown that an association exists between PRS and CSF biomarkers and disease progression (Martiskainen et al., 2015), as well as between PRS and plasma inflammatory biomarkers (Morgan et al., 2017). A PRS can improve the diagnostic accuracy of ApoE alone in identifying AD cases (Chaudhury et al., 2018), can predict the age of AD onset (Desikan et al., 2017) and generally improve risk prediction in healthy older adults (Escott-Price et al., 2015;Chouraki et al., 2016). Recently, a PRS has been demonstrated to predict cognitive decline and neurodegeneration in subjects at risk for AD (Tan et al., 2017). Moreover, other researches focused on gene expressions in AD that may reflect epigenetic modifications (Uddin and Singh, 2017). In this regard, a recent transcriptome-wide meta-analysis study of blood-based microarray gene expression profiles as well as neuroimaging and cerebrospinal fluid (CSF) endophenotype analysis showed that five genes (CREB5, CD46, TMBIM6, IRAK3, and RPAIN) were significantly dysregulated in LOAD. The most significantly altered gene, CREB5 (cAMP response element-binding protein family that play important roles in synaptic strengthening and memory formation), was also associated with brain atrophy and increased amyloid-β(Aβ) accumulation (Nho et al., 2020).

Cardiometabolic risk factors
Apart from the genetic factors, a considerable number of studies have demonstrated many other potentially modifiable risk factors for AD, amongst which cardiometabolic factors have been regarded as crucial. Cardiometabolic risk factors including diabetes, obesity, high levels of low-density lipoprotein (LDL) cholesterol and low levels of high-density lipoprotein (HDL) cholesterol and high blood pressure have been linked with the development of AD.
Finnish Geriatric Intervention Study to Prevent Cognitive Impairment and Disability (FINGER) study investigated the effects of a 2-year comprehensive intervention targeting several lifestyle and vascular risk factors simultaneously in 1260 people. This study showed an overall 25% improvement in Neuropsychological Test Battery (NTB) scores in the intervention group compared to the control group. For executive function, test scores were 83% higher in the intervention group, and processing speed was 150% higher (Ngandu et al., 2015). This study clearly highlights the role of cardiometabolic risk factors on cognition.
Ultrasonography has allowed for the measurement of atherosclerosis in the carotid artery by measuring intima media thickness (IMT) and carotid plaques. IMT and carotid plaques are more prevalent in AD patients and both are related to increased cognitive decline (de Bruijn and Ikram, 2014). Atherosclerotic plaque is also a pre-clinical large vessel disease, and the calcification volume can be assessed through CT scans. A larger calcification volume is associated with smaller brain tissue volumes and decreased microstructural integrity of the white matter, both of which are associated with increased risk of AD (Bos et al., 2012). Cerebral small vessel disease, such as white matter lesions and lacunar infarcts are related to AD or cognitive impairment (van Dijk et al., 2008;Blum et al., 2012). The Cardiovascular Health Study demonstrated strong associations with increased risk of AD with peripheral artery disease and atherosclerosis (Newman et al., 2005). It is important to note that cardiovascular diseases play an important role in the aetiology of AD, but needs further research to unravel the mechanisms. Cardiovascular disease and cognitive decline could be linked through cerebral hypoxia and silent brain lesions (Qiu and Fratiglioni, 2015).

Diabetes
AD and diabetes mellitus are two metabolic disorders characterized by increased incidence and prevalence with aging. In the last few decades, emerging epidemiological, preclinical and clinical evidence has established a possible connection between AD and type 2 diabetes mellitus (T2DM) (Li et al., 2015). A greater prevalence of T2DM in AD compared to non-AD healthy controls (34.6 vs. 18.1%; p < 0.05)  and a higher incidence of AD among T2DM patients (Hölscher, 2011) were revealed by longitudinal population-based studies. A recent meta-analysis carried out in almost 400 clinical studies, confirmed that diabetes is a main target for AD prevention (Nho et al., 2020). These findings led to the recognition of T2DM as a tangible risk factor for the development of AD.
Studies have shown that insulin could prevent Aβ accumulation and plaque formation by promoting brain Aβ clearance (Watson et al., 2003). Insulin receptors (IR) are present in copious amounts in the brain, where they are particularly present in synapses on both astrocytes and neurons and the hippocampus, hypothalamus, cerebral cortex and olfactory bulb (Abbott et al., 1999;Kandimalla et al., 2016). Although insulin and insulin receptors are abundant in the brain, they are selectively distributed (Craft and Watson, 2004). The insulin receptor signalling pathways (IRSPs) are involved in synaptic plasticity (Chiu et al., 2008), neuroprotection, growth, energy metabolism and longevity (van der Heide et al., 2006). One mechanism that could explain the underlying brain insulin resistance in AD is neurotoxicity mediated by oligomeric amyloid-β. Insulin degrading enzyme (IDE) was one of the principal regulators of Aβ levels in neuronal cells. IDE is responsible for degradation of insulin, Aβ and amylin, while declined activity of this enzyme could break the homeostasis, shifting the equilibrium toward Aβ deposition (Farris et al., 2003). Insulin resistance, defined clinically as the inability of a known quantity of exogenous or endogenous insulin to increase glucose uptake and utilization in an individual as much as it does in a normal population (Lebovitz, 2001), also induces Aβ oligomerization and activates the downstream c-Jun NH2-terminal kinase (JNK) pathway (Yoon et al., 2012). One preclinical study showed that diet-induced insulin resistance was associated with decreased IDE levels and increased amyloidosis (Ho et al., 2004). Another crucial finding was that Aβ oligomers could act on hypothalamus and trigger peripheral insulin resistance via NF-κB signaling (Clarke et al., 2015). Insulin resistance, together with hyperinsulinemia, therefore, were believed to promote neurodegeneration and facilitate the onset of AD (Kandimalla et al., 2016).
Another pathogenic event in both AD and T2DM is the formation of NFTs consisting of hyper-phosphorylated microtubule-associated protein tau. Several studies have implicated activation of insulin receptors and insulin resistance in tau phosphorylation and aggregation. Various kinases and phosphatases, including JNK, phosphatidylinositide 3-kinases (PI3K), protein kinase B (PKB/Akt) and glycogen synthase kinase-3 (GSK-3) have been identified. In T2DM patients, chronic hyperglycaemia and IR are thought to induce JNK activity and, as a result, cause oxidative stress and pancreatic β-cell apoptosis (Kaneto et al., 2007). Induction of the JNK pathway and PI3K/PKB/GSK-3 pathway appears to be crucial in AD development because it promotes abnormal tau hyperphosphorylation (Thakur et al., 2007). In particular, GSK-3β is widely considered to be the primary kinase responsible for the phosphorylation of tau and is modulated by insulin via the PKB/Akt pathway (Li and Hölscher, 2007). GSK-3β contributes to the pathology of AD-type neurodegeneration, since it is important for the formation of long-term memory and the maintenance of synaptic plasticity.
Soluble Aβ oligomers were shown to significantly lower response of insulin receptors to insulin, which also led to neurotoxicity in AD (Zhao et al., 2008). Additionally, advanced glycation end-products (AGEs) may also play a pivotal role in the relationship between diabetes and AD-type dementia. AGEs are proteins or lipids that become glycated after exposure to sugars. They are aging factors, being prevalent in many degenerative diseases, such as diabetes and AD (Goldin et al., 2006). The formation of AGEs is a part of normal metabolism, but if excessively high levels of AGEs are reached in tissues and the circulation, they can become pathogenic. The pathologic effects of AGEs are related to their ability to promote oxidative stress and inflammation by binding with cell surface receptors or cross-linking with body proteins, altering their structure and function. It is important to note that AGEs have been primarily found in the cytosol of neurons, especially in the hippocampus and parahippocampal gyrus (Fournet et al., 2018). Levels of AGEs and receptor for advanced glycation end-products (RAGE) are both significantly higher in AD patients, which are associated with the formation of amyloid plaques and neurofibrillary tangles (Choi et al., 2014). A recent study, aimed at investigating the effect of metformin on hyperphosphorylated tau proteins in diabetic encephalopathy (DE) by regulating autophagy clearance, showed that metformin may decrease tauopathy and improve cognitive impairment in an animal model, by modulating autophagy through the 5' AMP-activated protein kinase (AMPK) dependent pathway (Chen et al., 2019).
However, even if several preclinical and clinical studies suggested a tight link between diabetes or insulin signal alterations and AD, we have also to consider some conflicting results from epidemiological experiences (Li et al., 2015). In this regard, the Rotterdam study examined over 6000 patients without dementia, around 11% of whom were diagnosed with T2DM. Subjects were then monitored for one year. Diabetes nearly doubled the risk of developing AD in this study (Ott et al., 1999) and, in the Framingham study, a relationship between diabetes and dementia was found to be more evident in participants older than 75 (Akomolafe et al., 2006). On the other hand, Van den Berg and colleagues found that diabetes was not associated with augmented cognitive decline in participants older than 85 (van den Berg et al., 2010), indicating possible involvement of other cardiometabolic risk factors in cognitive decline. Moreover, a study in Japanese Americans showed no association between diabetes in middle age and dementia (Curb et al., 1999), and a Swedish study in 1301 community dwellers aged 75 years and older found no significant risk of developing AD in diabetes patients (Xu et al., 2004). In this setting, a meta-analysis showed that the aggregate relative risk of AD for people with diabetes was 1.5 (95% CI 1.2-1.8), while for vascular dementia the relative risk was 2.5 (95% CI 2.1-3.0) (Cheng et al., 2012). Given that DM is systemic metabolic disorder closely related with ischemic vascular events and vascular dementia, it is possible that many trials showed a false association between DM and AD for the lack of sensitivity of AD diagnosis. For these reasons, the relationship between diabetes and the major types of dementia remains controversial.

Obesity
Epidemiological studies have demonstrated a link between obesity and an increase in the risk of developing AD and other types of dementia, which is independent of co-morbid conditions such as diabetes (Nguyen et al., 2014). Several population-based studies have attempted to investigate the effect of obesity on AD. A Swedish study reported a 36% increased risk of AD associated with every unit increase in body mass index (BMI) in women at age 70 years (Gustafson et al., 2003). It is interesting to note that obesity developed in middle age seems to be critical for the increased risk for AD. One study, which included both male and female subjects, observed a greater risk in those who have a midlife body mass index > 30 kg/m 2 (Kivipelto et al., 2005). Another study in 2005 demonstrated that overweight individuals (BMI = 25-29.9 kg/m 2 ) aged 40-45 years had an increased risk of developing dementia of 35%, whereas obese individuals (BMI ≥ 30 kg/m 2 ) had an increased risk of dementia of 74% (Whitmer et al., 2005).
The exact mechanism by which obesity is related to AD pathology is still not fully understood. There is a general consensus that the aberrant inflammatory response underlying metabolic syndrome may arise from the dysregulation of the endocrine homeostasis. It has been suggested that subclinical inflammation of adipose tissue may interact with the cerebral inflammatory response, leading to neurodegeneration. Adipocytes can mimic immune cells and produce pro-inflammatory adipokines, cytokines and chemokines including leptin, adiponectin, CRP, MCP-1, MIF, PAI-1, RBP4, CLU and various pro-inflammatory cytokines (Monteiro and Azevedo, 2010;Won et al., 2014). Amyloid precursor protein expression has been shown to be upregulated in adipocytes in obesity (Puig et al., 2017). Several lines of evidence indicate that, in obesity, adipose tissue is infiltrated by immune cells including macrophages, monocytes, natural killer cells and lymphocytes. With the appearance of visceral obesity, there is a dynamic switch in the populations of immune cells, resulting in increasing predominance of M1 macrophages, which release pro-inflammatory cytokines such as IL-6 or TNF-α, and generate a large number of reactive oxygen species (ROS). These could cross the blood-brain barrier (BBB) and trigger central neuroinflammation (Cai, 2013) leading to activation of microglial cells. Microglia, in turn, release potentially neurotoxic substances, such as nitric oxide, pro-inflammatory cytokines, complement proteins, and other inflammatory mediators, resulting in tau phosphorylation and neurodegenerative changes (Eikelenboom et al., 2002). Animal models, of obesity, induced by the consumption of a high-fat diet (HFD), is associated with inflammation, in both peripheral tissues and hypothalamic areas critical for energy homeostasis (Thaler and Schwartz, 2010;McLean et al., 2019;Magnuson et al., 2020). Interestingly, in animals fed with HFD, neuroinflammation could be reduced by administration of the stable analogue of glucagon-like peptide-2 (GLP-2) (Nuzzo et al., 2019).
It is worth noting that obesity and metabolic syndrome, characterized by high fasting glucose levels, have long been regarded as important precursors of T2DM. Moreover, decreased insulin-stimulated glucose transport manifests as insulin resistance and is commonly seen in obesity. It has been suggested that, in the state of peripheral insulin resistance in obesity, insulin may cross the BBB and provoke overproduction of IL-1β, IL-6 and TNF-α. These inflammatory cytokines stimulate trafficking of APP and lead to increased deposition of Aβ. Furthermore, Aβ itself contributes to increased production of proinflammatory cytokines in the manner of a vicious cycle.
As mentioned above, emerging evidence suggests that changes in body weight are associated with AD. Therefore, factors which regulate body weight are likely to influence the development and progression of AD. The adipocyte-derived hormone, leptin, is a major regulator of body weight, mainly by activating hypothalamic neural circuits (McGuire and Ishii, 2016). Besides the hypothalamus, leptin receptors are widely expressed throughout the brain including the hippocampus and neocortex. It has been suggested that hyperleptinemia may be involved in cognitive deficits. In obese subjects, hyperleptinemia may result from different alterations, such as down-regulation of leptin receptors, an increase in circulating C reactive protein (CRP), oxidative stress and overexpression of SOCS3 (Beltowski et al., 2002). CRP attenuates the physiological functions of leptin and contributes to leptin resistance. Moreover, leptin may itself promote the production of CRP, suggesting the existence of feedback between leptinemia and inflammatory response. In vivo, it has been shown that neuronal MyD88-dependent signalling serves as a key regulator of diet-induced leptin and insulin resistance. In obese patients, plasma clusterin levels are increased and associated with BMI, waist circumference, markers of inflammation, and insulin resistance (Won et al., 2014). In mice, skeletal muscle and hepatic gene expression of clusterin are increased after high-fat diet feeding, and whole-body clusterin knockout (KO) mice are insulin sensitive compared to wild-type (WT) mice (Kwon et al., 2014). Interestingly, in T2DM patients with morbid obesity who underwent gastric by-pass, the expression of clusterin was significantly reduced as well as APP mRNA expression (Ghanim et al., 2012). These data suggest a potential role of clusterin on obese patient AD development. The excess of peripheral clusterin production in obese patients may have an influence on CNS tissue favouring after BBB crossing the deposition of amyloid.
Previous research has shown inconsistencies between age, obesity and the risk of developing AD. These mixed results could possibly be explained by the duration of the studies. Indeed, there is a large body of evidence to suggest that obesity or a high fat diet has a detrimental effect on cognitive performance. However, one study showed that, over time, there is an inverse association between BMI and dementia risk. A retrospective cohort study also revealed that, over two decades, being underweight in middle age and old age carries an increased risk of dementia (Qizilbash et al., 2015). The association between BMI and dementia has been suggested to be probably attributable to two different processes: a harmful effect of higher BMI, which is observable in long follow-up, and a reverse-causation effect that makes a higher BMI appear to be protective when the follow-up is short (Kivimäki et al., 2018). Clearly, further investigation is needed to fully comprehend how exactly obesity relates to AD pathology, but it can be concluded that there is an association between obesity and cognitive performance.

Hypercholesterolemia
Clinical studies indicate that middle-aged individuals with increased cholesterol are more susceptible to AD. This is supported by studies showing that diet-induced hypercholesterolemia increases Aβ accumulation in both white rabbits (Sparks et al., 2000) and transgenic mouse models of AD (Levin-Allerhand et al., 2002). Many studies suggest that cholesterol can influence the promotion of AD. One study performed in Japan tested cholesterol levels of nearly 2600 people between the ages of 40 and 79 who had no signs of AD. They found formation of senile plaques in 86% of people with high cholesterol, whereas they were present in only 62% of people with low cholesterol (Matsuzaki et al., 2011). A more recent study revealed that cholesterol could act as a catalyst and speed up the aggregation of Aβ. It was estimated that the presence of cholesterol caused Aβ plaques to develop 20 times faster than they would have otherwise (Habchi et al., 2018). Therefore, the use of cholesterol-lowering medication, such as statins, has led to the hope of treating or preventing AD. It should be noted that statins do not simply inhibit the synthesis of cholesterol, but also inhibit effects on inflammation (Wolozin et al., 2007). However, the results of work on this topic were inconsistent: some studies found beneficial effects, while others did not (Shepardson et al., 2011).
There are two main types of cholesterol carried by different types of lipoproteins, low-density lipoproteins (LDL) and high-density lipoproteins (HDL). It is important to note that the effect of LDL cholesterol and HDL cholesterol on the risk of developing neurodegenerative disease including AD is not yet well-established. Elevated LDL concentration in mid-life was revealed to increase AD risk in later life. Increased oxidative modification and nitration of LDL were observed in AD-type dementia and hypercholesterolemia compared to control subjects (Dias et al., 2015). On the other hand, higher levels of HDL were associated with a decreased risk of both probable and possible AD and probable AD (Reitz et al., 2010). However, the connection between high LDL or low HDL and AD risk is not definite and awaits further investigation.
Changes in the cholesterol levels have also been found to cause changes in cell membrane properties. Lipid rafts are cholesterol-rich and sphingomyelin-rich membrane microdomains where BACE1 (beta-site amyloid precursor protein cleaving enzyme 1) and γ -secretases are located. As the amount of cholesterol increases in the cell membrane, large portions of which also gather within these lipid rafts. This increase in cholesterol promotes the binding of APP to the lipid rafts, leading to the production of Aβ following cleavage by BACE1 and γ-secretase (Beel et al., 2010). If the level of intracellular cholesterol decreases, it could result in a decrease in the catalytic activity of BACE1 and γ-secretase (Grimm et al., 2008). These results collectively imply that the inhibition of acyl-CoA cholesterol acyltransferase (ACAT) could be a promising therapeutic strategy, since it can reduce the production of cholesterol esters and, in turn, reduce the amount of Aβ produced.

Mid-life hypertension
Hypertension has been identified as a tangible risk factor in the development of all types of dementia, not only AD. Several studies have examined the association between elevated blood pressure in midlife (age 40-64 years) and the progression of dementia (Nelson et al., 2014). Launer and colleagues in the Honolulu Asia Aging Study (HAAS) examined this relationship in 3703 Japanese-American men aged 45-68. Among those untreated for high blood pressure (n = 2111), there was a strong association between midlife hypertension and both AD and vascular disease when 160/95 mmHg was used as the blood pressure (BP) cut off value (Gelber et al., 2012). In another study, 243 brains were examined by Petrovitch et al. (Petrovitch et al., 2000). They found that those with elevated systolic blood pressure (SBP) in midlife showed vasculopathic changes and increase in Aβ plaques in both the neocortex and hippocampus. Patients with a high diastolic blood pressure (DBP) also showed elevated levels of NFTs and Aβ plaques (Launer et al., 2000). In this regard, Walker et al., in a prospective cohort study including 4761 subjects, showed that midlife and late-life hypertension had higher risk for incident dementia compared with those who remained normotensive (Walker et al., 2019). Another study showed that both elevated SBP and DBP in midlife was not only associated with an increase of AD and vascular dementia, but that this association was independent of ApoE genotype (Launer et al., 2000).
In line with clinical findings, many preclinical studies using animal models also showed an association between hypertension and AD. Hypertension in mice was shown to significantly reduce cerebral blood flow, leading to increasing activated microglia, indicating an inflammatory state. In mouse models, inflammatory stress has been shown to develop before the presence of AD pathology, including increased APP cleavage, tau hyperphosphorylation, and subsequent cognitive impairments (Krstic et al., 2012). In a mouse model of hypertension, it has also been demonstrated that oxidative stress is increased in the cerebrovasculature, which could lead to an upregulation of receptor for advanced glycation end products (RAGE) mRNA and its ligands. RAGE was discovered to control the transport of Aβ across the BBB (Deane et al., 2003) and brain endothelial expression of RAGE was increased in both AD mouse models and AD patients (Deane et al., 2009). Gentile and colleagues discovered that, in hypertensive mouse models, hypertension causes deterioration in the BBB and increases Aβ deposition in the hippocampus (Gentile et al., 2009).
Recently published research indicated that older people with higher average blood pressure are more likely to develop NTFs and plaques in the brain compared to their age-matched controls (Arvanitakis et al., 2018). In this longitudinal cohort study of 116 older adults without dementia, higher cardiovascular risk scores during a 20-year period were significantly associated with lower cerebral blood flow to the medial temporal, parietal, and occipital cortices. The association varied during the life span such that cardiovascular risk in midlife, but not in later life, was significantly associated with cerebral hypoperfusion in older age. However, it is important to note that this was an observational study, and thus the findings did not actually prove that hypertension caused the signs of AD.

Conclusion
AD is a progressive, irreversible and multifactorial neurodegenerative disorder. Genetic risk factors have been well established to play an important role in AD, while inflammation, oxidative stress, mitochondrial dysfunction, metabolic alterations and insulin signalling impairment are some pieces in the aetiology of AD waiting to be assembled. Despite continued (but abating) debate on the epidemiological and clinicopathological relevance of systemic metabolic diseases to AD, there have been intense research efforts at defining convergent pathogenic processes using a repertoire of experimental models. It seems that cerebrovascular abnormalities commonly associated with metabolic disorders represent an important link that should be further investigated. In particular, the emerging evidence on the possible shared genetic background between AD and cardiovascular traits will provide new possible pathogenetic mechanisms. Moreover, despite the identification of shared risk factors including diet, lifestyle has emphasized the importance and feasibility of preventive strategies against the development of both cardiometabolic and cognitive conditions, uncovering of shared mechanisms, underlying these comorbid disease processes, point to an opportunity to develop/screen common treatments. As discussed in this review, some of the key disease mechanisms, including inflammation and insulin resistance, occur both in the periphery and in the brain suggesting common mediators (e.g. cytokines, pro-oxidants) and/ or deficiency in homeostatic regulators (e.g. insulin, IGF, HDL, antioxidants). A summary of these AD risk factors is illustrated in Fig. 1.
The search for the genetic risk factors contributing to AD has evolved enormously in recent decades. Amyloid precursor protein, presenilin 1 and presenilin 2 were first identified as a cause of autosomal dominant AD, followed by the identification of the ApoE ε4 allele as a major genetic risk factor for both early-onset and late-onset AD. GWAS have further identified over 20 risk gene loci, these susceptibility loci mainly act through three main pathways for disease pathogenesis: immune and complement systems/inflammatory response, endocytosis and cholesterol and lipid metabolism (Guerreiro and Hardy, 2011). Cardiometabolic risk factors including diabetes, obesity, high LDL cholesterol and low HDL cholesterol and high blood pressure have been linked with the development of dementia, although without evidence of a causal relationship. Despite the association between cardiometabolic traits and dementia that could be described as a causative relationship considering the potential burden of cerebrovascular chronic injury, some clinical experiences also documented a strong correlation between cardiovascular or metabolic factors and the pure form of AD. Interestingly, the new evidence from GWA studies suggested a possible genetic background which may be shared by both cardiometabolic traits and AD. This finding led to the formulation of the hypothesis that AD and cardiometabolic diseases could share a common genetic background which affects both physiopathological mechanisms (Bone et al., 2021). GWA studies documented the role of specific loci associated with inflammatory responses or metabolic pathways (Broce et al., 2019).
Systemic inflammation and insulin resistance may represent convergent mechanisms through which comorbid metabolic disorders promote the development of AD (Bhat and Thirumangalakudi, 2013). Special interest has also been given to the role of obesity, hypercholesterolemia and adipokines in dementia. Several population-based studies have suggested a role of obesity in neuroinflammation and AD development. Obesity is characterized by a chronic low-inflammatory state mediated by the production of several pro-inflammatory cytokines. The latter could cross the BBB and trigger neuroinflammation (Cai, 2013) leading to activation of microglial cells. Microglia, in turn, release potentially neurotoxic substances, leading to neurodegeneration (Eikelenboom et al., 2002). Moreover, hypercholesterolemia has also been suggested to be involved in the pathogenesis of AD, even though many therapeutic approaches failed to demonstrate real benefits in preventing or slowing the progression of the diseases.
There is no treatment that can halt, slow or reverse the progression of AD. Most of the recent clinical trials that attempted treating AD focused on inhibiting the main known causes of AD, such as Aβ. However, these attempts have failed because the therapeutic intervention was performed in patients with very advanced pathology or because the treatment was directed to the wrong target (Mehta et al., 2017). Therefore, an alternative strategy to fight AD could be the prevention of the known modifiable risk factors and related mechanisms for the disease. This includes management of associated comorbidities such as those listed in this review. In this regard, future studies are warranted to better explore the mechanisms underlining AD pathology, including genetic and metabolic factors. In detail, more accurate studies are needed to better define the genetic risk of AD, including not only single gene GWA studies but also haplotype studies, as the neurodegenerative processes of AD probably derive from polygenetic factors. Moreover, metabolic pathways, along with genetic susceptibility to AD and diabetes, need to be better understood, not only in preclinical models but also in humans. Dynamic studies with new PET CT tracers should be a promising strategy for shedding more light in the pathophysiology mechanisms but also for testing potential curative drugs, especially in the early phase of AD where symptoms are not already present.
In summary, although aging and ApoE ε4 status appear to be the most important, well-established, and validated non-modifiable risk factors for developing sporadic AD, cardiometabolic risk factors could be promising therapeutic targets for novel treatment approaches. (Fig. 1).

Disclosures
Dr Paul Edison (PE) was funded by the Medical Research Council and now by Higher Education Funding Council for England (HEFCE). He has also received grants from Alzheimer's Research, UK, Alzheimer's Drug Discovery Foundation, Alzheimer's Society, UK, Alzheimer's Association, US, Medical Research Council, UK and European Union. PE is a consultant to Pfizer, Roche and Novo Nordisk. He has received speaker fees from Novo Nordisk, Pfizer, Nordea, Piramal Life Science. He has received educational and research grants from GE Healthcare, Novo Nordisk, Piramal Life Science/Life Molecular Imaging, Avid Radiopharmaceuticals and Eli Lilly. He was an external consultant to Novo Nordisk and a member of their Scientific Advisory Board. He is in the scientific advisory board for CytoDyn.

Data availability
No data was used for the research described in the article.