Genetics of Vascular Dementia

Abstract

Genetic studies are transforming the way we diagnose, evaluate and treat patients. The era of genome-wide association studies promised to discover common risk variants in heterogeneous disorders where previous small-scale association studies had on the whole failed. However, as we enter the post-association era a degree of disappoint is felt regarding the lack of risk factors with large effect for a number of disorders including vascular disease. Vascular disorders are sporadic by nature, though a familial component has been observed. This review will focus on vascular dementia, the genetic risk factors for vascular disorders and highlight how new technologies may overcome the limitations of genome-wide association and nominate those genes that influence disease risk.

Introduction

The first draft of the human genome was released February 2001 and heralded a new era in both basic scientific endeavors and medical practice. The human genome is composed of 46 chromosomes that are estimated to encode for approximately thirty thousand genes; giving rise to millions of different protein isoforms^{1, 2}. Through DNA sequencing studies, vast amounts of variation have been observed at the nucleotide level; DNA sequence variants are estimated to occur one in every three to five hundred base pairs and result in a large diversity of phenotypes ranging from variation in traits (e.g. immune response) and susceptibility to disease. Polymorphism, literally translated as many forms, is the term used to describe common DNA variants (>1% frequency) that exist within a species. Polymorphism in a gene may result in increased/reduced protein production or affect the level of abnormal proteins generated by directly influencing gene transcription³. Increasing evidence shows that gene expression is regulated by complex interplay of a number of determinants from simple polymorphic variants in regulatory regions such as the promoter or `3 untranslated regions to the presence of microRNA species that can work through a negative or positive feedback mechanism to either down- or up-regulate expression^{4, 5}.

The human genome displays considerable inter-individual variability from simple single nucleotide polymorphisms (SNPs) and short repeats to large-scale deletions, multiplications and rearrangements. Large gene copy number variations (>100Kb) occur in the general population with no determinable disadvantage to carriers⁶. This phenomenon can also be pathogenic and result in disease phenotypes^7-10. The disease state is usually caused through either a ‘gain or loss of function’ that occurs from an altered balance in the level of an essential protein; however, the presence of variants that represent a much milder phenotype and produce a fractional increase/decrease in expression may result in the progressive phenotypes that typify numerous age-related diseases, including cancer, neurodegenerative disorders and vascular disease.

The work of the International HapMap project has helped to identify linkage disequilibrium patterns traversing the entire genome and facilitated the identification of haplotype ‘tagging’ SNPs to perform more cost-effective association studies^{11, 12}. Public databases, such as dbSNP, house data on about seventeen million candidate human SNPs, for which genotype data is validated for about ten million (http://www.ncbi.nlm.nih.gov/SNP/snp_summary.cgi)¹¹. This vast array of data, coupled with the incredible advances in genotyping and DNA sequencing technologies, create a situation whereby the geneticist can pinpoint trait loci^{13, 14}. Genome-wide association studies (GWAS) were proclaimed as the most powerful tool in the mapping of causative/modifying genetic loci in complex disorders ¹⁵. Unfortunately, to date the GWAS in cerebrovascular disease have nominated only a small number of loci, with only one consistent genomic locus being nominated^{16, 17}, even though epidemiologic studies support a sizeable genetic component in risk of cerebrovascular disease. A systematic review of heritability in stroke from previous studies estimated that first degree relatives of stroke patients have an approximately 50% higher risk of cerebrovascular disease¹⁸.

Vascular disease is a major factor in both mortality and morbidity. Fifteen million people worldwide suffer a stroke each year with devastating effects; one-third of these individuals die and another one-third remain permanently disabled¹⁹. The burden in the United States is increasing, with more than 780,000 new or recurrent strokes²⁰ and 240,000 transient ischemic attacks²¹ occurring each year. The scope of the public health problem is even greater when clinically “silent” events are also considered. Data from population-based studies with neuroimaging estimate that more than 11 million people experienced “silent” strokes in the US annually²². The effect of this “silent” disease on the public health remains unclear, but likely contributes to dementia burden.

Herein we review the genetic factors associated with cerebrovascular disease and specifically vascular dementia (VaD) from the perspective of what we know, and how the use of future genetic techniques can be applied to help clarify the genetic variants that effect disease risk.

Clinical Vascular Dementia

VaD is currently used to refer to dementia as a result of, or associated with, cerebrovascular disease. The National Institute of Neurological Disorders and Stroke (NINDS; 1991) published diagnostic criteria for VaD²³. The workshop received support from the Association Internationale pour la Recherché et l'Enseignement en Neuroscience (AIREN) and generated a diagnostic scheme for clinical assessment of VaD (NINDS-AIREN).

The term “dementia” implies an acquired loss of cognitive abilities,²⁴ and as widely used, it has become almost synonymous with Alzheimer type dementia with characteristic and early memory impairment codified as an essential feature in Diagnostic and Statistical Manual of Mental Disorders (DSM-IV)²⁵. A definition of VaD has been proposed that would shift the emphasis from memory impairment to executive dysfunction severe enough to interfere with social or occupational functioning²³. DSM-IV stipulates that the clinical diagnostic criteria for dementia should have both memory impairment and cognitive disturbances in one or more of the following domains: language (aphasia), motor activities (apraxia), visuospatial skills (agnosia) or executive functioning²⁶. VaD is also argued not to have memory impairment in its pure form (i.e. not associated with other pathologies, most often Alzheimer type pathology) and instead might have prominent disturbances in frontal executive functioning resulting from the effects of cerebrovascular disease²³.

The cognitive symptoms of VaD are pleomorphic in that no single cognitive syndrome captures the observed range of symptoms and signs in the disorder. In a redefinition of the term VaD, Hachinski coined the phrase “vascular cognitive impairment”²⁷. The term vascular cognitive impairment (VCI) was later used to incorporate VaD under a broader clinical category that included any form of cognitive impairment resulting from cerebrovascular disease, including vascular mild cognitive impairment, VaD, and mixed vascular-Alzheimer type dementia²⁸. There is debate as to whether VCI should include VaD and so the term “vascular cognitive disorder” has been proposed to encompass both VCI and VaD^{23, 29}.

Table 1 summarizes the major features of the four main diagnostic schemes for VaD³⁰. There are often considerable periventricular and subcortical white matter changes on structural imaging (Figure 1). Impairment on neuropsychological tests is often characterized by prominent disorders in planning, sequencing, speed of mental processing and attention, with relatively better scores on memory tests³¹. Despite these characteristic features, there are no neuropsychological tests that unequivocally distinguish Alzheimer's disease (AD) from VaD, in part due to the heterogeneity of VaD. VaD may result from a single strategic stroke, multiple lacunar infarcts or cortical microinfarcts, or microvascular lesions that are not evident on neuroimaging³². Conversely, the detection of infarcts with neuroimaging does not necessarily equate to VaD, as some infarcts are clinically silent³³ .

Table 1.

Comparison of clinical diagnostic schemes for VaD

	ADDTC	ICD-10	NINDS-AIREN	DSM-IV
Stepwise/fluctuating deterioration	-	-	+	-
Focal brain damage as evident upon neurological testing	±	++	++	+
Evidence of >2 ischemic strokes	±	-	±	-
Evidence of significant CVD	++	++	±	+
Brain imaging requirements of CVD	+	-	++	-
Temporal relationship between stroke and onset of dementia	±	-	+	-
Etiologic relation to cognitive impairment	-	++	++	++
Unequal distribution of cognitive deficits	-	++	-	-

(Adapted from Pohjasvaara et al.)

required and/or requirement or requirement

++ indicates required element; + and/or required;; ± or required; - not required. One caveat of the ADDTC scheme is if the temporal relationship between stroke and dementia is established ≥ 1 infarct must be located outside of the cerebellum as evident upon imaging. A caveat of NINDS-AIREN is that it requires either a 3 month temporal relationship between stroke and onset of dementia and/or abrupt or stepwise deterioration in cognitive function, as well as evidence of multiple strokes or localization of a “strategic infarct”; DSM-IV does not require neuroimaging evidence of cerebrovascular disease (CVD), but focal neurological signs or laboratory evidence of CVD is required.

Figure 1.

(a) postmortem MR imaging of a patient with VaD reveals cribriform status of the putamen (solid arrow) as well as extensive increased signal in frontal white matter (dotted arrow). (b) At autopsy the cribriform change is grossly visible as dilated vessels (arrow) in the putamen that are associated with decreased signal change seen on MRI. (Panel A courtesy of Dr. Clifford R. Jack, Jr; Panel B courtesy of Dr. Joseph E. Parisi)

Neuroimaging

The advent of neuroimaging techniques such as X-ray computed tomography (CT), T2 magnetic resonance (MR) images, and fluid attenuated inversion recovery (FLAIR) sequence have greatly improved ante-mortem detection of white matter disease and infarcts in demented patients (Figures 1 and 2). Conversely, the recognition of similar lesions in normal, unaffected patients has made interpretation of the findings more difficult^{33, 34}. In a recent study that evaluated factors associated with dementia in the setting of a recent stroke, the presence of concurrent white matter disease, especially around the frontal horn, was associated with worse performance in multiple cognitive domains³⁵ . Whether cerebral white matter pathology (so-called leukoaraiosis³⁶) is due to vascular/ischemic mechanisms is still debated. With increasing age, white matter lesions are common in the periventricular region, while they are less predictable in the subcortical centrum semiovale. Multi-infarct dementia (MID) is a subtype of VaD where imaging is useful to locate the lesions. In patients with focal cortical infarcts, it is not always easy to differentiate the clinical neurobehavioral deficits associated with focal cortical syndromes (e.g., apraxia, aphasia) from the neurocognitive deficits that characterize VaD, especially in patients that do not meet other diagnostic criteria for dementia³¹. In addition, the presence of so-called silent strokes confounds interpretation of findings on imaging. They may not be the sole substrate of VaD, but may play a role in the presentation of VaD³⁷ .

Figure 2.

(A) Dissection of an atherosclerotic vessel from a pathologically confirmed case of VaD reveals complicated atheromatous plaques, with a markedly expanded intima filled with grumose material. (b) In another patient rarefied white matter and obliterative small vessel disease are associated with subcortical white matter hyperintensities on neuroimaging. (c) An arteriosclerotic vessel in the basal ganglia shows an intima expanded by cholesterol containing, lipid material from lipohyalinosis. (d) Marked gliosis and a paucity of CA1 neurons are evident in hippocampal sclerosis. (Panel A courtesy of Dr. Joseph E. Parisi)

Treatment and Pharmacogenomics

Similar to AD, there are neurotransmitter deficits in VaD, although they may not be as specific since the locus of vascular pathology varies from individual to individual and is not as stereotypic as in AD³⁸. Moreover, there is evidence to suggest that in pure VaD, in contrast to AD, cholinergic deficits may not be present³⁹. The authors examined cholinergic deficits in the temporal cortex of subjects with pure VaD (n=9), concurrent VaD and AD (n=12), AD (n=10) and normal controls (n=12). They observed a cholinergic deficit in both AD and VaD with AD; however, the pure VaD patients showed similar values as the controls. These findings suggest that cholinergic therapies may benefit VaD patients who also have AD pathology. Given the heterogenous nature of VaD, it can not be ruled out that different patterns of cholinergic deficits are associated with small-vessel disease⁴⁰. Nevertheless, cholinesterase inhibitors have been used empirically to treat VaD, with modest effects on cognition and clinical global impression scales⁴¹. Ischemia of the basal forebrain and of cholinergic pathways is postulated to account for improvements with cholinesterase inhibitors,⁴² but the much more likely explanation for the appearance of any benefit of cholinesterase inhibitors in VaD is the coexistence of Alzheimer pathology in a large number of clinically diagnosed VaD cases⁴³. Prevention focused on treating vascular risk factors would seem to have more public health relevance than symptomatic therapies⁴⁴. Along this line, it is worth noting that antihypertensive drugs have been observed to decrease the risk of stroke and VaD in those at risk for both⁴⁵.

Neuropathologic diagnosis

Presently, standardized neuropathologic criteria for cerebrovascular disease consistent with VaD have not been generally accepted. The heterogeneity of cerebrovascular disease lesion types, including microinfarcts, lacunar infarcts (< 1 cm) and large cortical infarcts complicates the operationalization of a widely accepted scoring method. The location of the cerebrovascular disease pathology further complicates the classification of subtypes and neuropathologic diagnosis⁴⁶. Additionally, a pathologic diagnosis consistent with VaD can be made more difficult by the presence of concomitant Alzheimer-type pathology. The prevalence of cerebrovascular lesions has been reported to range from 40% to 80% in autopsy proven AD⁴⁷. The lack of an accepted standard definition is reflected in a lack of consistency in methods for assessing VaD. A survey of multiple centers that perform neuropathological examinations found that the heterogeneity of applied criteria may impair the verification of vascular lesions⁴⁸. While they did not attempt to define a neuropathological set of criteria, they point out the lack of uniformity across centers and the need for a consensus on procedures.

Kalaria and coworkers proposed a list of key variables to be noted in assessing cerebrovascular disease⁴⁶ that could be scored in increasing numerical order, with 0 for absent and 1 to 3 used to represent mild, moderate and severe pathology. The variables to be included are the type of infarct (i.e. anemic or hemorrhagic), location, laterality and size. For the purposes of distinguishing “pure” VaD from mixed dementia, it is crucial to exclude other common pathologic processes that occur in the aged brain, such as AD, Lewy body disease and argyrophilic grains⁴⁹. The presence of neuronal loss and gliosis in the vulnerable sectors of the hippocampus, so-called hippocampal sclerosis, also needs to be assessed⁵⁰. A prospective longitudinal study of subcortical ischemic vascular dementia evaluated the contribution of neurofibrillary tangles, hippocampal sclerosis and vascular pathology to cognitive status⁵¹. Their proposed cerebrovascular disease assessment evaluated subcortical and cortical lesions in the gray and white matter. The scoring system was based on infarct type: cystic infarcts were scored by volume; lacunar infarcts, by count; and microinfarcts, by density (i.e. 1-10, 11-24, 25+). They found that neurofibrillary tangles, hippocampal sclerosis and vascular disease all independently contributed to worsening cognitive status. A summary of pathologic features of VaD is provided in Table 2.

Table 2.

Pathologic features consistent with VaD

Cerebrovascular disease	• Atherosclerosis – large vessels • Arteriolosclerosis – small vessels (<100 μm diameter), often showing lipohyalinosis • Cerebral amyloid angiopathy
Infarcts	• Ischemic (anemic) or hemorrhagic • Location: cortical (multimodal, unimodal, paralimbic), subcortical (basal ganglia, thalamus, internal capsule), white matter, brainstem, cerebellum
Size	• Microinfarcts (not visible to naked eye) • Lacunar infarcts (<1.5-cm) • Encephalomalacia (larger infarcts, usually in major vascular territory)
Involvement	• Gray matter, white matter or both • Arterial circulation ○ Anterior, middle or posterior cerebral arteries ○ Watershed zones between major circulations ○ Laterality: right or left
White matter disease	• Location: periventricular, centrum semiovale or both: • Myelin pallor or ischemic injury, associated with gliosis and vascular collagenosis
Absence of other significant pathology	• Amyloid plaques (diffuse or neuritic) and neurofibrillary tangles • Lewy bodies (a-synuclein positive brainstem or cortical-type inclusions) • Argyrophilic grains • Hippocampal sclerosis • Vascular vs. degenerative • Frontotemporal lobar degeneration • Tauopathy

(Adapted from Kalaria et al 2004 and Chui 2006)

Risk factors

Risk factors for VaD can be divided into four main categories: demographic, atherosclerotic, brain imaging findings and genetic risk factors^{52, 53}. Demographic risk factors included age, race, sex and education. As noted above, the frequency of VaD increases with age. VaD appears to be more prevalent in Africans and Asians, though there are inconsistencies in reports on Asians^{53, 54}. Studies on gender differences in VaD have had mixed results, but the predominant conclusion is that men are at higher risk than women⁵⁵. Education is reported to be protective, but whether or not this is a result of the association of higher education with higher income or occupation type remains to be determined⁵³. Atherosclerosis risk factors include hypertension, cigarette smoking, myocardial infarction, diabetes mellitus and hypercholesterolemia⁵². One study reported a three-fold increase in risk of VaD in patients who had hypertension and heart disease⁵⁶. Brain imaging risk factors include cerebral atrophy, white matter changes (i.e. leukoaraiosis) and cerebral infarcts. Risk factors for post-stroke VaD include location of the infarct, with greater likelihood for left-sided infarcts, infarcts in “strategic sites,” and bilateral infarcts; volume of tissue loss produced by the infarct(s); and recurrent strokes ⁵⁷.

Genetics of Vascular Dementia

As previously stated the complex nature of vascular disease reflects a multigenic condition with synergistic interplay of genes and environment. Genes influencing risk of pure vascular disorders have been difficult to identify; however, the combination of alternate approaches exploiting both population and family genetics have successfully nominated a number of key genetic determinants in related disorders such as AD and stroke. In the following sections, we will highlight some of the genes that have been identified within AD and stroke and how these relate to vascular disease and result in distinct forms of VaD.

Familial Genetics

One of the most successful approaches to the mapping of disease-related genes has been the identification of rare Mendelian forms. The classical linkage method employing large familial aggregates that display patterns of disease inheritance (dominant/recessive) has identified genes involved in monogenic forms of AD and cerebrovascular disorders ⁵⁸.

Familial AD genes

AD and VaD share several features and both are generally considered complex and heterogeneous^{59, 60}. The majority of patients with AD are sporadic with between 5-10% AD patients that are considered familial, with autosomal dominant inheritance seen in about 1%^{61, 62}. Three gene have been implicated in autosomal dominant AD families, with rare mutations in the amyloid precursor protein gene (APP) and the presenilin 1 (PSEN1) and 2 (PSEN2) genes^{61, 63, 64}. The presenilins are components of the proteolytic γ-secretase complex that together with β-secretase (BACE 1) generates β-amyloid peptide fragments from APP⁶⁵. Two major β-amyloid species of either 40 (Aβ40) or 42 (Aβ42) amino acids in length are produced, with Aβ42 being the more fibrillogenic and neurotoxic species. The identification of mutations in genes encoding both the substrate and the key enzyme for generation of β-amyloid has led to major advances in the characterization of the pathophysiology of AD and provides the main support for the amyloid cascade hypothesis^{66, 67}. This hypothesis suggests the central event in AD is the imbalance between the production, maturation and clearance of β-amyloid in brain leading to neuronal degeneration and dementia. Over time this hypothesis has evolved principally in the nature of the pathogenic β-amyloid species that is proposed to initiate deleterious events and cause AD^{67, 68}.

The APP gene was first implicated in dementia, as Down syndrome patients have an extra copy of chromosome 21q21, which leads to overproduction of APP and deposition of Aβ peptide in amyloid plaques and the development of early-onset AD associated with cerebral amyloid angiopathy (CAA)⁶⁹. Recently, individual duplications at the APP locus in 5 families with autosomal dominant early-onset AD associated with CAA have been identified⁷⁰. These results were supported by a second study, which demonstrated that APP duplication is sufficient to cause early-onset AD with CAA⁷¹. These results suggest that increased APP gene dosage may initiate a cascade of events leading to β-amyloid plaques and neurofibrillary tangles, although it remains to be evaluated to what extent APP duplication leads to protein overexpression. In addition it provides a hypothesis that a mild increase in APP expression due to variation in 5’ or 3’ regulatory regions could be a risk factor for late-onset sporadic AD. In fact, APP promoter mutations have been associated with an in vitro increase in transcriptional activity^{72, 73}.

Familial stroke genes

The term autosomal dominant retinal vasculopathy with cerebral leukodystrophy (RVCL) encompasses three disorders cerebroretinal vasculopathy (CRV), hereditary vascular retinopathy (HVR) and hereditary endotheliopathy, retinopathy, nephropathy and stroke (HERNS). RVCL patients are clinically characterized with visual loss, stroke and dementia beginning in middle age with death occuring in most families 5 to 10 years later⁷⁴. Recently it was reported that RVCL was caused by truncation mutations in the TREX1 gene resulting in the protein losing part of its C-terminus^{74, 75}. TREX1 (DNase III) is a DNA-specific 3’ to 5’ exonuclease. The wild-type TREX1 protein is localized to the perinuclear region, while the TREX1 proteins found in individuals with RVCL, which lack part of the C-terminal end of the protein, are diffusely distributed in the cytoplasm and the nucleus⁷⁴. TREX1 is a part of the SET complex that normally resides in the cytoplasm, but translocates to the nucleus in response to oxidative DNA damage⁷⁶. The haploinsufficiency caused by TREX1 mutations in RVCL may reduce the interaction with SET proteins and diminish SET complex formation. This SET complex is believed to target DNA repair factors to damaged DNA under conditions of oxidative stress. An alternative hypothesis may be that the truncated TREX1 isoforms in the nucleus and cytoplasm may have toxic effects.

Cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy (CADASIL) is a rare form of small vessel occlusive disease. The genetic locus involved in CADASIL was originally mapped to chromosome 19 in two families and thirteen additional linked kindreds were subsequently identified^{77, 78}. In 1996 Joutel and colleagues reported the identification of the first genetic locus for stroke, with pathogenic mutations in the NOTCH3 gene observed to result in CADASIL⁷⁹. The Notch pathway is a fundamental signaling mechanism determining cell fate⁸⁰. The Notch family of receptors plays a pivotal role in the regulation of this cascade (Figure 3). Notch1 through notch4 are cell surface receptors, which interact with membrane-bound ligands transducing signals between neighboring cells. The Notch3 receptor promotes vascular smooth muscle cell survival. To date, the confirmed pathogenic CADASIL mutations are located in the epidermal growth factor (EGF)-like repeat domains at the extracellular N-domain of the receptor, with missense, splice-site mutations and in-frame deletions observed in patients. Mutations appear to affect highly conserved cysteine residues, although the role of mutations affecting other residues and protein domains remains unclear⁸¹. Within each wild-type EGF-like repeat there are six cysteine residues, while in CADASIL mutation patients there is a loss or gain of a cysteine residue. This uneven number of cysteine residues has been hypothesized to effect differential protein interactions and the possible multimerization of mutant Notch3. Recently, a novel mutation was reported p.L1515P that is hypothesized to hyper-activate the Notch3 receptor, suggesting alternative mechanisms of pathogenicity⁸². The pathomechanism behind NOTCH3 mutations and the ischemic stroke, which is characteristic of CADASIL, remains unclear.

Figure 3. NOTCH Signaling.

Mammals express four members of the Notch family of receptors (notch 1–4), and two families of ligands (delta-like 1,3,4 and jagged 1,2). Notch receptors and ligands are both single-pass transmembrane proteins that enable signaling between neighboring cells. Binding of the ligand triggers proteolytic cleavage of the Notch receptor, releasing the Notch intracellular domain (NICD) into the cytoplasm. The final step of proteolytic cleavage is mediated by the γ-secretase complex, of which presenilin (PSEN) acts as the catalytic subunit. The cleaved NICD translocates to the nucleus where it forms an active transcriptional complex with the DNA binding protein RBPJ (recombination signal binding protein for immunoglobulin J-kappa region), the co-activator mastermind-like (MAML) and other co-activators (CoA) (Adapted from High & Epstein JA, 2008; 33).

There does not appear to be a consistent genotype-phenotype correlation among the more than 50 mutations that have been described⁸³. The clinical phenotype usually presents with ischemic episodes, cognitive deficits, migraine with aura or psychiatric disturbance. At autopsy CADASIL is characterized by the presence of granular osmiophilic material (GOM) deposits (Figure 4). Ischemic stroke is by far the most common manifestation accounting for up to 85% of patients⁸⁴. The disease course of CADASIL is variable, even in the same family⁸⁵. The age-at-onset can range from 30-94 years, with early-onset forms not necessarily predicting a more severe symptomatic progression, and disease duration range of between 3-43 years⁸⁶. There is no specific treatment for CADASIL, with most therapies focused on symptomatic control; the mean age-at-death is estimated to be approximately 60 years⁸⁴. Interestingly, a NOTCH3 mutation was identified in an octogenarian patient with a minor stroke and no family history of stroke, demonstrating the potential role of NOTCH3 mutations with less severe presentation^{87, 88}. Notch3 signaling has also been recently implicated in ischemic stroke and suggests that Notch3 protein is a determinant of stroke burden via vascular smooth muscle cells⁸⁹.

Figure 4. Granular osmiophilic material.

Granular osmiophilic material (GOM) in cerebral autosomal dominant arteriopathy with subcortical infarctions and leukoencephalopathy (CADASIL). A low-power image of a small artery in CADASIL with reduplicated basal lamina, degeneration of smooth muscle cells, and extensive GOM in the outer media (original magnification × 9000; image courtesy of Dr. Wen-Lang Lin).

Recently, mutations in the HTRA1 gene encoding a serine protease were identified to cause cerebral autosomal recessive arteriopathy with subcortical infarcts and leukoencephalopathy (CARASIL)⁹⁰. CARASIL is by ischemic small-vessel disease that is associated with alopecia and spondylosis^{91, 92}. Linkage analysis in five CARASIL families of Japanese ancestral origin identified a region on chromosome 10q26, and with fine-mapping and sequence analysis, two nonsense and two missense mutations were observed in the HTRA1 gene⁹⁰. These mutations are reported to result in nonsense-mediated decay of HTRA1 mRNA or a reduction in protease activity. The HTRA1 protein is a serine protease that regulates TGF-β signaling⁹³. TGF- β signaling is associated with vascular angiogenesis and has a role in vascular endothelial and smooth muscle cells. It is now crucial to determine both the role of HTRA1 genetic variation and TGF- β signaling in the more general forms of sporadic ischemic stroke.

Population Genetics

The complexity of common human disorders such as vascular disease has frustrated the research community for many years. Association studies in population-based samples have generally failed to identify reliable disease markers. This is due in part to limited marker density, and restricted numbers and heterogeneity in samples. Given the age-related sporadic nature of vascular disease and particularly VaD, it is hypothesized that these types of disorders are the results of multiple genetic factors of small effect that interact with a second stimulus, which could be an environmental agent. In this section, we will focus on common variants that have nominated genes for AD and cerebrovascular disease through candidate and unbiased GWAS.

The best example of a common major genetic risk factor for both AD and VaD is apolipoprotein E (APOE). There are three main genetic variants of APOE - ε2, ε3 and ε4 - with several studies indicating an increased risk of VaD associated with APOE ε4, especially VaD associated with cerebral amyloid angiopathy⁹⁴. The ε4 allele of the APOE gene has been the only consistently replicated functional genetic risk factor for late-onset AD, VaD and vascular disease^{95, 96}. Meta-analyses showed a 3-fold increased risk to develop AD in individuals carrying one copy of the APOE ε4 allele and a 15-fold increased risk for APOE ε4 homozygote individuals⁹⁷. The APOE ε4 allele further modifies the disease by shifting its onset to an earlier age^{98, 99}. Several mechanisms have been proposed to explain APOE ε4 detrimental effects in dementia (for review, see¹⁰⁰). Through its fundamental role in lipid transport, APOE is implicated in numerous cellular pathways linked to dementia, including cholesterol redistribution, oxidative stress, neurite outgrowth, tau phosphorylation, and β-amyloid clearance and aggregation.

Sporadic AD genes

For a continuously updated list of the genes most strongly associated with AD based on meta-analyses of all published studies, we refer to the AlzGene database (http://www.alzforum.org/res/com/gen/alzgene/default.asp)¹⁰¹. The emergence of GWAS rather than smaller candidate gene studies means no a priori hypothesis is necessary. A number of novel susceptibility genes have been nominated through GWAS (Table 3)^102-112. Other than APOE, no other gene is consistently observed across the studies. This may be due to limited sample sizes, differences in study design and the stringent corrections for multiple testing inherent to these studies. In an attempt to overcome this last hurdle, Feulner et al. performed an initial analysis of their GWAS data only including the current top 10 candidate genes according to the AlzGene database^{101, 113}. Using this approach, all four genes that were previously identified through GWAS (GAB2, PGBD1, PCK1 and LMNA) showed nominally significant association, compared to only two genes (MAPT and SORL1) identified through candidate gene studies.

Table 3.

GWAS performed in late-onset AD.

GWAS	Study Design	Sample origin	Number of samples in GWAS (patients/controls)	Genes or SNPs most associated with late-onset AD
Carrasquillo et al.	Clinical and neuropathological case-control	USA	844/1255	APOE, PCDH11X
Beecham et al.	Clinical case-control	USA	492/496	APOE, FAM113B
Feulner et al.	Clinical case-control	Germany	491/479	APOEa
Abraham et al.	Clinical case-control	UK	1082/1239	APOE, LRAT
Betram et al.	Family based	USA	941/404	APOE, ATXN1, CD33, rs11159647 at 14q31.2
Li et al.	Clinical case-control	Canada and UK	753/736	APOE, GOLM1, rs9886784 at 9p24.3, rs10519262 at 15q21.2
Grupe et al.	Clinical case-control	USA and UK	380/396	APOE, GALP, TNK1, PCK1
Coon et al. and Reiman et al.	Clinical and neuropathological case-control	USA and Holland	664(446)/422(290)	APOE, (GAB2 in APOEε4+ carriers)
Lambert et al.	Clinical case-control	France	2032/5328	APOE, CLU, CR1
Harold et al.	Clinical case-control	USA, UK and Germany	3941/7848	APOE, CLU, PICALM

^aNominally significant association was also found for GAB2, PGBD1, PCK1, LMNA, MAPT and SORL1 but p-values were not corrected for multiple testing. AD = Alzheimer's disease; APOE = Apolipoprotein E; PCDH11X = protocadherin 11, x-linked; FAM113B= family with sequence similarity 113, member B; LRAT=lecithin retinol acyltransferase; ATXN1= Ataxin 1, CD33= CD33 molecule; GOLM1= golgi membrane protein 1; GALP = galanin-like peptide precursor; TNK1 = non-receptor tyrosine kinase 1; PCK1= phosphoenolpyruvate carboxykinase 1; GAB2 = GRB2-associated binding protein; CLU= clusterin/apolipoprotein j; CR1= complement (3b/4b) receptor 1; PICALM= phosphatidylinositol binding clathrin assembly protein.

Sporadic Cerebrovascular Disease genes

Several biologically plausible candidate genes have been identified for cerebrovascular disease and, subsequently, VaD (Table 4). For example, a substitution in the angiotensinogen protein (methionine to threonine at position 235) is linked to a promoter variant that is reported to affect transcription of the gene and increase risk of coronary artery disease¹¹⁴. Angiotensinogen is a precursor for angiotensin, a polypeptide in the blood that causes vasoconstriction, increased blood pressure, and aldosterone release from the adrenal cortex. Angiotensin converting enzyme (ACE) catalyzes the conversion of angiotensin to angiotensin II, a potent vasoconstrictor, and inactivation of bradykinin, a potent vasodilator. These two actions of ACE make it an ideal target in the treatment of conditions such as high blood pressure, heart failure and stroke. A insertion/deletion variant has been associated with myocardial infarction and AD, while a deletion allele has been associated with stroke^115-117. This variation has been suggested to affect transcription of the gene thereby altering plasma levels of ACE¹¹⁸, leading to the suggestion that ACE inhibitors may be a possible therapeutic target to lower vasoconstriction and reduce the risk of stroke.

Table 4.

Candidate Genes That Are Reported to Associate With Stroke

Gene Name	Chromosome	Polymorphism
Factor V Leiden	1q23	R506Q
5,10-Methylenetetrahydrofolate reductase	1p36.3	C677T
AGT	1q42-q43	Ins/Del
Interleukin-6	7p21	−174G>C
Paraoxonase 1	7q21.3	L55M, Q192R
Matrix metalloproteinase 3	11q22.3	5/6 repeat A
Platelet glycoprotein Ilb/IIIa	17q21.32	A1/A2 Variant
ACE	17q23.3	−6 G>A, M235T
Apo-E	19q13.2	ε2, ε3,ε4

Isolated populations, which are due to expansion from a limited number of founders, provide an ideal background to study the genetics of common disease, keeping in mind the caveat of ethnic-specific variations. DeCODE Genetics has employed this approach in Iceland, with genotypic and medical data from over 100,000 individuals - more than 50% of the adult population of Iceland - in tandem with their genealogical data linking together the entire present-day population and stretching back over 1,100 years. By mining these datasets, DeCODE can effectively trace the inherited components of a given disease, pinpointing the key disease genes as well as the specific markers or haplotypes within these genes that correlate with the disease. In 2003, DeCODE reported that carriers of one copy of an “at-risk” phosphodiesterase 4D (PDE4D) haplotype had a nearly 2-fold increased risk of carotid or cardioembolic ischemic stroke and those with two copies are at an almost 4-fold increased risk¹¹⁹. The investigators also reported changes in PDE4D isoform expression profiles between affected and unaffected individuals, including affected individuals with or without the high-risk haplotype ¹¹⁹. The authors speculated that the associated variant may result in transcriptional regulation of PDE4D7 isoform. Further studies from DeCODE reported evidence for an association between the gene encoding 5-lipoxygenase activating protein (ALOX5AP) and stroke in the Icelandic population ¹²⁰. A 4-SNP haplotype (HapA) was significantly associated with myocardial infarction (OR= 1.8) and stroke (OR= 1.67), while another (HapB) was found to be a risk factor of myocardial infarction in a British population (OR= 1.95)¹²⁰. Attempts to replicate associations of specific PDE4D and ALOX5AP polymorphisms with stroke have been inconsistent^121-132. A recent meta-analysis failed to confirm the association of PDE4D gene with stroke, suggesting that the association may be restricted to Iceland or neighboring Scandinavian countries¹³³.

The first GWAS in ischemic stroke was reported in 2007¹⁷. The study was a preliminary investigation of the complex genetics underlying ischemic stroke and involved a relatively small series of about 250 ischemic stroke patients and controls from the US Caucasian population. The study generated more than 200 million individual genotypes in total and nominated a number of potentially interesting SNPs. Positive associations (with P-values < 0.00001) were identified across twelve chromosomes. The authors highlight the fact that of the eleven characterized genes nominated, two are involved in potassium transport (KCNIP4 on Chr4p15 and KCNK17 on Chr6p21) and two are located at chromosome 9p21 (Chr9p21), which was previously associated with myocardial infarction. The study highlighted the fact that no single genetic factor has substantial magnitude of effect to cause ischemic stroke. Potential confounders such as lifestyle or co-morbidities may influence association, and a more detailed analysis of these findings is now warranted. A larger GWAS of stroke (involving almost 20,000 subjects) recently nominated a novel locus on Chromosome 12p13¹⁶. The region contains two genes NINJ2 and WNK1, which are both biologically plausible candidates. These findings recently failed replication in a large case-control meta-analysis¹³⁴.

Chromosome 9p21

Genome-wide association and subsequent replication studies in myocardial infarction/coronary artery disease have confirmed an association of SNPs at Chr9p21 and disease risk (odd ratio ~1.2-1.7; reviewed in¹³⁵). Further studies have suggested that the same genomic locus shows association with other disorders such as type-2 diabetes, stroke and VaD^136-138. Within the genomic region on Chr9p21 showing association there are three genes encoding proteins and one non-coding RNA transcript. The region encompasses the INK4/ARF tumor suppressor locus containing two cyclin-dependent kinase inhibitor genes (CDKN2A and CDKN2B), and the 5'-methylthioadenosine phosphorylase (MTAP) gene. In addition, a recently identified non-coding RNA called ANRIL has been identified to overlap at it 5’-end with the CDKN2B gene¹³⁹. A meta-analysis in over 8,000 subjects attempted to fine-map the association of variants with coronary artery disease¹⁴⁰. These results nominated variants within the ANRIL non-coding RNA as the prime candidate for disease risk in coronary artery disease.

Further analysis showed that the ANRIL transcript is expressed in cell-types that may be associated with atherosclerosis, including vascular endothelial and smooth muscle cells¹⁴⁰. Recent expression studies have shown that at least three alternate isoforms of ANRIL exist, including one long and two short transcripts¹⁴¹. Jarinova and colleagues later showed that expression of the shorter ANRIL transcripts are increased in carriers of the Chr9p21 coronary artery disease -risk haplotype¹⁴¹. In addition, they showed that a variant (rs1333045) in a conserved region (CNS3) is associated with increased gene expression in primary aortic smooth muscle cells. No sequencing studies have been performed to characterize the unknown genetic variation within the ANRIL locus, including the exons encoding the different transcripts. Studies are now warranted to investigate the genetic variation at the ANRIL locus and assess the role of alternatively spliced isoforms in stroke and VaD.

Future advances in the Genetics of VaD

With aging being one of the greatest risk factors for VaD and an increasing aging population, increased awareness of the various types of dementing disorders is important. Epidemiological studies estimate the number of dementia patients will double every 20 years, increasing the global demented population from 24.3 million to 84.1 million people by 2040. VaD is the second most common form of dementia, with the frequencies ranging from 15 to 20%¹⁴². A meta-analysis of VaD involving 23 studies from Europe, United States, Taiwan, China and Japan found an increase in incidence, and an exponential increase over the age of 65 years.¹⁴³ Given these statistics, it is crucial that further work be performed to find preclinical biomarkers and that new targeted therapeutics be developed. Genetics will play a role in the diagnosis and treatment of VaD.

The future holds great promise for the elucidation of genes and genetic variants that predispose individuals to vascular disorders. Over the next decade genetic discoveries will be increasingly integrated into the clinical practice. Individualized medicine will determine the therapeutic intervention strategies employed for each patient based on genomic risk profile. The latest DNA sequencing technologies will become a standard for variant identification and association with disease risk. Whole-genome sequencing remains an expensive goal. Its cost will be dramatically reduced over the coming years, and the aim of the $1000 genome no longer appears to be beyond our grasp. Genomic capture approaches that allow the researcher to focus on specific genomic regions have revolutionized the way genetic sequencing studies are performed. Indeed, the application of whole-exome sequencing and transcriptome sequencing have already proved to be cost-efficient methods for detecting new variants involved in disease^144-147. Whole-exome sequencing involves the capture of every known coding exon throughout the genome for sequencing, whereas transcriptome sequencing screens all mRNA species, identifying different mRNA isoforms and giving insight into gene expression patterns.