Copy number variations and stroke

Stroke is the third leading cause of death worldwide after heart disease and all forms of cancers. Each year about 795,000 people in the USA suffer from a recurrent or new stroke [1]. This pathology is one of the leading contributors to death and long-term adult disability worldwide, and for this reason, the burden of stroke is felt physically, socially, economically, and emotionally by patients, by their relatives and health care services [2]. Stroke is defined as a syndrome characterized by a quick development of clinical signs and loss of cerebral functions, with symptoms lasting for over 24 h or leading to death, with an apparent cause of vascular origin [3].

Conventional and genetic risk factors contribute to damaging a cerebral blood vessel and, consequently, cause stroke [4]. Genetics plays a significant role in the development of this disease. In fact, several monogenic disorders cause stroke, as well as the interaction of multiple genes.

A new form of genetic variation, known as copy number variations (CNVs), has been recently identified as a new tool for understanding the genetic basis of many diseases, including stroke. CNVs are deletions and duplications (loss or gain) of segments of genome [5, 6].

CNV may alter the levels of gene expression, may also disrupt genes or regulation elements, may lead to frameshifts, and may generate new fusion products; all these genetic variations can result in a phenotypic variation, susceptibility of an individual to disease and/or a differentiated drug response [7, 8].

Today, modern high-resolution technologies, such as comparative genome hybridization (CGH) arrays, allow to detect simultaneously CNVs in multiple loci. These technologies may be clinically used to identify people who may be at risk for a stroke or might create benefit to identify specific therapies.

This review aims to provide a comprehensive overview of stroke types and their etiopathogenesis and summarize the current knowledge regarding the involvement of CNVs in stroke.

Occluded or ruptured cerebral blood vessel determines a reduction in normal cerebral blood flow in the affected vascular territory, resulting in reduced nutrient delivery to gray and white matter [9]. Without oxygen and nutrients from blood, neurons start to die within a few minutes in the core of the infarcted area. The region around the core, called “the ischemic penumbra,” contains functionally impaired cells but still viable for the presence of collateral vessels. This area may become infarcted at later time points due to secondary neuronal damage caused by the cascade of biochemical events that occurs after ischemia. This mechanism is common to all types of stroke: ischemic stroke (IS) [2], hemorrhagic stroke (HS), and transient ischemic attack (TIA).

IS represents up to 80 % of all stroke cases reported in epidemiological studies [2]. It is more often disabling rather than fatal, representing the most common life threatening neurological disorder. The remaining 20 % of stroke cases are caused by primary intracerebral hemorrhage (about 15 %) and subarachnoid hemorrhage (about 5 %) with a potential mortality rate from 30 to 50 % within 30 days [10, 11]. Last, TIA is similar to IS but differs in duration (less than 24 h).

This distinction in different stroke types is critical for therapeutic decision, although it is likely that these forms of stroke have both similar and different genetic susceptibility, risk factors, and etiologic overlaps. Furthermore, the patients’ global risk factor profile at the time of the stroke may influence the form of stroke that occurs.

Stroke is a challenging disease to study, because it can depend on a wide variety of risk factors. Conventional risk factors that increase a person’s likelihood of having a stroke can be controllable and uncontrollable (Fig. 1) [12].

Controllable risk factors, by medication or lifestyle changes, primarily include high blood pressure, diabetes mellitus, carotid artery stenosis, peripheral artery disease, atrial fibrillation, stress, alcohol, drug abuse, hypercholesterolemia, obesity, smoking, and physical inactivity.

Uncontrollable risk factors mainly comprise age, race (African–Americans have a much higher risk of death from a stroke than Caucasians—http://​www.​strokeassociatio​n.​org), ethnicity, family history, genetic factors, previous stroke or TIA, artery abnormalities, fibromuscular dysplasia, male gender, etc. Genetic factors contributing to onset of stroke have been identified in twin studies and familial aggregation studies [13]. Genetic predisposition to stroke can be categorized either as a single gene disorder or as a polygenic disorder, although the majority of the studies have mainly focused on monogenic forms of stroke. A recent study demonstrates that conventional cardiovascular risk factors, particularly smoking and hypertension, have been associated with an earlier stroke onset, highlighting the interaction between gene and environment, and the clinical importance of careful risk factor control even in patients with monogenic stroke disorders [14].

The most common monogenic form is CADASIL (cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy) (OMIM 125310). It results from mutations in the gene NOTCH3, which encodes a transmembrane receptor. Mutations result in an odd number of cysteine residues within one of the 34 epidermal growth factor (EGF)-like repeats in the extracellular amino-terminal region of the Notch3 receptor, leading to its abnormal accumulation at the cytoplasmic membrane of vascular smooth muscle cells, in the vessels of patients [15].

Recently, genes involved in several other rare monogenic diseases have been recognized. CARASIL (cerebral autosomal recessive arteriopathy with subcortical infarcts and leukoencephalopathy) (OMIM 600142) causes lacunar stroke and early onset vascular dementia, and derives from recessive mutations in the HtrA serine protease (HTRA1) gene, which is involved in TGF-beta signaling [16].

COL4A1 and COL4A2 are two genes that encode the alpha 1 and alpha 2 chains of type IV collagen, which cause autosomal dominant porencephaly, infantile hemiparesis, and childhood hemorrhage [17, 18].

Another monogenic condition characterized by visual loss, stroke, and dementia is autosomal dominant retinal vasculopathy with cerebral leukodystrophies (RVCL) (OMIM 192315), a microvascular endotheliopathy. Mutations in the TREX1 gene are responsible for this disease [19].

Mutations in the genes underlying these monogenic forms of stroke are not recognized as the cause for multifactorial stroke, but may help in their comprehension.

Single mutations can induce stroke, but in most cases, this condition is caused by interaction among multiple genes. Several candidate pathways have been examined in stroke, including those involved in endothelial function, nitric oxide production, renin–angiotensin–aldosterone system, coagulation, haemostasis, and inflammation (Fig. 1).

Nowadays, because of the completion of Human Genome Project, modern high-throughput technologies, including the next generation sequencing (NGS), CGH, and single-nucleotide polymorphism (SNP) arrays, can be used to genotype simultaneously multiple genes involved in stroke. The subsections below will describe the CNVs and the methods to detect them to assess the potential association between CNVs and the development of stroke.

CNVs are defined as deletions or duplications of DNA that produce any change in copy number of a specific chromosomal region [5, 6, 20]. Their size varies from one kilobase (kb) to several megabases (Mb) [21, 22], and they often involve one or more genes [6].

In a diploid cell, the number of copies of a locus is two, a copy inherited from the mother and the other from the father, but some loci may contain CNVs.

It has been estimated that about 12 % of the genome is covered by CNVs and more than 41 % identifies CNVs overlap with known genes [6, 23, 24]. CNVs play an important role in the genome variability allowing humans to evolve and adapt [6, 20, 25].

CNVs have been recognized as source of both normal genetic variation and pathogenic mutation [26]. They can destroy regulation elements, generating new fusion products with various possible positive or negative consequences [20, 27]. Other studies indicate that larger CNVs are associated with pronounced clinical characteristics and deletions are associated with more severe phenotypes than duplications [6, 7].

CNVs may be divided into inherited or de novo types, and this depends on whether they are transmitted or not by at least one parent [28].

CNVs are classified into different categories (Table 1). Common CNVs usually represent normal genomic variation or benign. Rare CNVs can be likely benign variant specific to an individual or family, pathogenic variant, likely pathogenic variant and variant of unknown significance (VOUS)–CNVs with uncertain clinical and functional relevance. VOUS occur when a new CNV is identified. Family studies may help clinical interpretation, because the presence of a de novo CNV that segregates with the pathological phenotype strengthens the evidence that it is pathogenic. However, the importance of some CNVs may be still uncertain even after studies on families because of their variable expressivity; for this reason, it is extremely useful to perform comparative case–control data analysis in large populations to definitively associate specific CNVs to human diseases. Some CNVs do not lead to phenotypic effect in the carrier, but they can create genomic instability in future generations [20]. Normal genomic variants or benign CNVs may sometimes indirectly cause or contribute to pathogenicity, for instance, if:

For all these reasons, a better understanding of all mechanisms underlying CNVs is required.

Different methods for detecting CNVs are available, including real-time PCR (RT-PCR), NGS, and microarrays. The last one is now the primary method used for CNVs detection.

Microarrays include both SNP- and CGH-arrays. These technologies allow detection of CNVs at higher resolution than classical cytogenetic methods [5, 33]. The application of CGH- and SNP-arrays in control cohorts produces a genome-wide architecture of CNVs named ‘‘CNV landscape’’ [26, 34].

Array-based technologies have emphasized recurrent CNVs that seem to be associated with some diseases; in effect, they have been identified more frequently in patients compared with control populations.

All these methods differ in their ability to detect deletions or duplications; for instance, more duplications are missed by SNP-array and NGS approaches than by CGH-array. Currently, CGH-array is the most sensitive tool for the research of small differences in CNVs [35].

CGH-array allows to detect chromosome imbalances too tiny to be seen with the microscope. DNA samples from a patient and from a control are labeled with two different fluorophores and, consequently, hybridized on array containing thousands of known DNA probes. The probes are arranged in a precise grid on a glass slide called “chip” [36, 37]. The most commonly used fluorochromes are red and green (cyanine 5 and cyanine 3). The chip is analyzed in a microarray scanner which measures the amount of red and green fluorescence on each probe. Last, an array analytical software calculates the ratio of fluorescence and in this way deletions or duplications in DNA can be identified (Fig. 2).

CNVs are associated with several complex disorders, and their potential association with risk for stroke has been object of lively discussion [38].

Until now, there are a relatively few association studies between CNVs and patients with IS.

In the first, genome-wide analysis was investigated whether CNVs could modulate risk for IS and was intended to provide a list of CNVs in IS patients, but no common genomic structural variation unequivocally linked to IS was detected [39]. CNVs were examined in 263 patients with IS and 275 neurologically normal controls using SNP chips (Illumina Inc., CA, USA). In 146 patients, the authors identified a total of 231 CNVs resulting in simple deletions or duplications. Most of the same CNVs were identified in healthy individuals too. Forty-five CNVs (19.5 %) were unique (Table 2). Within these new potential sites of structural variation, only one genomic region, on chromosome 1, contained recurrent CNVs in three individuals with IS. These individuals showed an apparently identical duplication spanning the genes SPRY domain-containing SOCS box protein 1 (SPSB1) and hexose-6-phosphate dehydrogenase (H6PD). Because of the potential clinical relevance of these alterations, they examined copy number at this locus in an additional 450 neurologically normal samples. These data showed the presence of CNVs at this locus in five of these samples (~1 %), suggesting that these variants were not a risk factor for IS. The remaining CNVs could have a role in the pathobiology of IS; however, due the low frequency of each individual alteration, screening of these variants in a greater cohort would be required to confirm the association unequivocally. In addition, it would be desirable to use another methodology to detect smaller CNVs or CNVs in genomic regions poorly covered by this technique that could confirm the risk for IS.

Nørskov and colleagues evaluated whether CNVs in glutathione S-transferases (GSTs) M1 and T1 genes were associated with an increased risk of ischemic vascular disease (IVD) including IS [40]. GSTM1 and GSTT1 detoxify the products of oxidative stress and may protect against atherosclerosis and IVD. Furthermore, epidemiological studies hypothesized that CNVs in GSTM1 and GSTT1 genes were associated with progressive decreases in their catalytic activity. In addition, they may modify risk of atherosclerosis and increase risk of IS (Table 2) [41]. The researchers included 6.557 IVD cases and 16.502 controls from 2 general population studies and 2 case–control studies. To genotype for the exact number of genes copies of GSTM1 and GSTT1, they used the RT-PCR. Principal findings in these studies individually or combined demonstrated CNVs in GSTM1 and GSTT1 were not associated with the risk of IS or any ischemic vascular event. Furthermore, the authors did not detect any associations between smoking exposure and GSTM1 and GSTT1 genotype (Table 2).

IS can also be caused by spontaneous cervical artery dissection (CeAD), in particular in healthy young adults [42, 43], but unfortunately the etiology of CeAD is still unknown. A genetic predisposition seems to be associated between CeAD and inherited microscopic and submicroscopic connective tissue alterations. Grond-Ginsbach and collaborators searched for causative CNVs in patients with and without connective tissue alterations that may predispose to CeAD [44]. They included 49 non-traumatic CeAD-patients with electron microscopic alterations (EM+ patients), 21 patients without alterations (EM− patients), and 403 control subjects. All patients were screened for CNVs through Affymetrix SNP6.0 microarrays. The authors concluded that rare genetic variants may contribute to the pathogenesis of CeAD in particular in EM+ patients (Table 2). However, the risk for CeAD might not be related to a single-gene or a single-genetic pathway, but it might be associated with different genetic variants (Table 2).

Nik-Zainal and colleagues examined a case of a 35-year-old male with a ring chromosome 12 originally diagnosed 20 years before IS appeared [45]. CGH-array analysis revealed a submicroscopic microdeletion and microduplication within 12p13.3 and a microdeletion in 12q24.33. FISH analysis further revealed that in this patient, the duplication from exons 35–52 of Von Willebrand factor (VWF) gene was in an inverted orientation within the ring chromosome. VWF plays a critical role in maintaining the normal balance of the clotting cascade via multiple complex interactions with factor VIII, platelets, collagen, and subsequent degradation by a metalloprotease called ADAMTS13 (A Disintegrin And Metalloprotease with ThromboSpondin Type-1 Motif, 13). Partial duplication of this gene suggests that a potential mechanism for generating a prothrombotic state may have contributed to a premature stroke (Table 2).

At present, only two studies from the same authors reported the relationship between CNVs and HS, in particular with subarachnoid hemorrhage (SAH) (Table 2).

In 2008, Bae and colleagues genotyped SNPs on CNV regions for the CNVs identification. They found out 597 SNP markers with a multiallelic CNV genotype, known as the common deletion polymorphism, within the CNV region. Among 597 CNV markers, CNV region around rs1242541 (nearest gene: SEL1L) showed the most significant association with the risk of SAH [46].

In 2010, they executed the first genome-wide association study to investigate the relationship between common CNVs and SAH. They hypothesized that CNVs can predict the risk of SAH [47]. The authors identified a total of 4.574 CNVs from a Japanese population sample (n = 473) and discovered 1.644 unique CNV regions containing 1.232 genes. The researchers carried out a genome-wide CNV association analysis using a logistic regression model, controlling for age and sex, to determine the association between the identified CNVs and the risk of SAH in 187 CNVs with frequency >1 %. Interestingly, two CNV regions, deletion 4q31.3 and duplication 10p15.1 have been significantly associated with the risk of SAH. In the case of chr4:153210505–153212191, the frequency of deletion in the patients group was higher than that in the control group. This result suggests that the deletion allele may be a risk factor for SAH. In the case of chr10:6265006–6267388, the frequency of duplication in the patients group was higher than that in the control group. This latter finding indicates that the increase in copy number in the region may influence the onset of SAH. Unlike their previous work, in this study, no significant association has been detected between CNV region around rs1242541 and the risk of SAH. Probably, these discrepancies in the results may be due to the fact that this last study was conducted at a larger scale.

Finally, investigations on the association between CNVs and intracerebral hemorrhage (ICH) have never been reported in the literature.

Several factors may increase the risk of stroke, including genetic ones, and in particular, CNVs. Even if the role of CNVs in the genetic etiology of stroke is not yet well established, there is an increasing interest in CNVs because of their usefulness as a powerful tool in understanding the genetic basis of numerous diseases.

Until now, there are a relatively few association studies between CNVs and stroke. Some studies concluded that rare CNVs may contribute to the pathogenesis of stroke, while other studies detected no significant association between specific CNVs and the risk of stroke. These seemingly contradictory data can arise, because in most studies, the method used for detecting CNVs was SNP-microarray. To confirm an unequivocally association between CNVs and stroke and extend the current findings, it would be desirable to use another methodology to detect smaller CNVs or CNVs in genomic regions poorly covered by this technique, for instance, CGH-array. In addition, more duplications are missed by SNP-based array and sequencing than by CGH-array and SNP-array, which have limited ability to detect single-exon CNVs due to the distribution of SNPs across the genome. Currently, CGH-array is the most sensitive tool for the research of CNVs. Another strategy to improve the detecting of CNVs would be that to combine SNP-microarray and CGH-array into one platform providing a genetic screening in a more efficient manner.

Furthermore, screening of CNVs in a suitable number of patients would be required to confirm unambiguously the association between CNVs and risk for stroke.

It is clear that the discovery of disease-associated CNVs will lead to improvements in clinical genetic diagnosis and genetic counseling. This will not only help to make more appropriate diagnosis but may help to design treatments which could be allocated according to genetic etiology rather than meeting strict diagnostic criteria set for each separate disorder. Therefore, the identification of CNVs could lead to personalized medical treatments which would be targeted for each patient and his genome, and could, therefore, improve treatment success.

Finally, as patients with shared genetic etiologies of stroke will be identified, studies of genotype–phenotype correlation, natural history, and therapeutic response to specific drugs can be performed, which will lead to improved long-term care and outcomes for patients.

In light of these observations, further studies will be required to clarify how CNVs may affect an individual’s susceptibility to stroke, to confirm the associations in larger populations, and to know if there are some association between CNVs and the different subtypes of stroke.