Introduction

Alzheimer’s disease (AD) and cardiovascular diseases (CVDs) including atherosclerosis, strokes, hypertension, myocardial infarction, coronary heart disease, and congestive heart failure, are common in aging, especially among adults aged 65 years and older, which constitute increasing public health problems (Carter et al. 2017; de la Torre 2012; Gong et al. 2018; Justin et al. 2013; Tini et al. 2020). AD, the most common form of neurodegenerative diseases and dementia, has a substantial genetic component with heritability of 58 to 79% (Gatz et al. 2006). Ninety-five percent of all people who have AD are late-onset AD with occurrence in the eighth or ninth decade of life (American Psychiatric Association 2013; Bettens et al. 2010; Guerreiro et al. 2010). As for CVD, the heritability ranges from 0.30 to 0.50 for various quantitative CVD measures (O'Donnell and Nabel 2008), and the narrow-sense heritability of CAD is estimated to be about 40% (McPherson and Tybjaerg-Hansen 2016; Peden and Farrall 2011).

The co-occurrence of AD and CVD has been suggested due to shared risk factors between AD and CVDs, as well as due to a direct causal association. For example, cardiac diseases cause hypoperfusion and microemboli, which have been implicated in the etiology of AD (Attems and Jellinger 2014; de Bruijn and Ikram 2014; de la Torre 2012; Justin et al. 2013; Tini et al. 2020). It has been reported that CVD and cardiovascular risk factors are associated with an increased risk of AD and mild cognitive impairment (MCI) (de Bruijn and Ikram 2014; Gorelick et al. 2011; Tini et al. 2020). Furthermore, AD and CVD share similar genetic and biochemical profiles and common triggers (Mur et al. 2020; Tini et al. 2020). Several genes have been linked to AD and CVD including apolipoprotein E (APOE), MTHFR, HFE, and FTO (Eriksson et al. 2010; Kotze and van Rensburg 2012; Mahley 2016; Mansoori et al. 2012; Mur et al. 2020; Ray et al. 2008). Individuals who carry AD-associated APOE-ε4 allele usually have higher cholesterol levels (de Bruijn and Ikram 2014) and display a more marked association between the extent of CVD and the density of cardinal neuropathological lesions of AD than subjects who do not carry this APOE-ε4 allele (Tini et al. 2020). Recently, one study identified 90 single nucleotide polymorphisms (SNPs) on 19 different chromosomes that jointly conferred increased risk for AD and cardiovascular outcomes (Broce et al. 2019).

Among the suggested candidate genes for AD and CVD, the Rho GTPase activating protein 26 (ARHGAP26) gene (also known GRAF, GRAF1, OPHN1L, OPHN1L1) is located at 5q31.3 (Borkhardt et al. 2000; Ishikawa et al. 1998). GAP (GTPase-activating protein), one of the modulators of Rho GTPase family, is involved in oxidative stress contributing to vascular disease (Strassheim et al. 2019) and neuroplasticity changes in AD and other brain diseases (Guiler et al. 2021; Lefort 2015). For example, the ARHGAP26 gene was reported to be associated with neuropsychiatric diseases (Dahm et al. 2014), while another study found that ARHGAP26 was significantly upregulated in Parkinson disease (PD), a neurodegenerative disorder, when compared with controls (Jiang et al. 2019). AD is the most common form of age-related neurodegenerative disorder with dementia featured by progressive memory loss and cognitive decline (Lane et al. 2018; Lipnicki et al. 2017). It has been demonstrated that AD shares genetic, epigenetic, and environmental risk factors with PD (Cacabelos 2020; Desikan et al. 2015; Kelly et al. 2020; Ramanan and Saykin 2013). However, no study has investigated the associations of the ARHGAP26 gene with AD and/or CVD. This study aimed to investigate the association of ARHGAP26 gene with the risk of AD and CVD by using a case–control sample, followed by a replication using additional case–control study. A third sample of a family-based study design was also used for replication and confirmation of the current findings in this study. In addition, the associations of ARHGAP10 gene (also known GRAF2; PSGAP; PS-GAP), an important paralogon of ARHGAP26 (Katoh and Katoh 2004), with AD and CVD were examined in the present study.

Materials and Methods

Subjects

ADNI Sample

The Alzheimer’s Disease Neuroimaging Initiative (ADNI) (adni.loni.usc.edu) was launched in 2003 as a public–private partnership, led by Principal Investigator Michael W. Weiner, MD. The primary goal of ADNI has been to test whether serial magnetic resonance imaging (MRI), positron emission tomography (PET), other biological markers, and clinical and neuropsychological assessment can be combined to measure the progression of mild cognitive impairment (MCI) and early AD. The ADNI study began in 2004 as a multicenter that provides services in the United States and Canada. The ADNI is an ongoing, longitudinal, multicenter study designed to develop clinical, imaging, genetic, and biochemical biomarkers for the early detection and tracking of AD. In the present study, the merged data was used from several components of ADNI. There was an Institutional Review Board (IRB) exemption for the current study due to secondary data analysis.

Data of 800 individuals, 200 patients with early AD, 400 with MCI, and 200 with cognitive normal (CN) as controls were extracted from the ADNI cohort 1. Social-demographic factors included gender, age, race, and educational level. Gender was self-reported as either male or female. Age and years of education were considered as continuous variables. Race consisted of four subgroups: non-Hispanic White, non-Hispanic African American, Hispanic, and others. CVD was defined as Yes (with the cardiovascular disease history) or No (without cardiovascular disease history) with a total of 521 patients with CVD. Cerebrospinal fluid (CSF) biomarker included the total tau (t-tau). The Functional Activities Questionnaire (FAQ) was used to measure the activities of daily living, and the higher total scores represent increased disability (Pfeffer et al. 1982). The Clinical Dementia Rating Scale Sum of Boxes (CDR-SB) was utilized to examine the global cognition of participants. A higher CDR-SB score indicates a more severe degree of cognitive deficits (Morris 1993).

Genotyping was conducted using the Affymetrix technique. A total of 154 SNPs within the ARHGAP26 gene and 63 SNPs within the ARHGAP10 gene located at 4q31.23 were available in the ADNI sample. APOE-ε4 carriers were defined as individuals with at least one ε4 allele while non-carriers were defined as individuals with no ε4 allele (APOE-ε4-0) (Table 1).

Table 1 Descriptive statistics
Full size table

The Caribbean Hispanics Sample

To validate the association of ARHGAP26 gene, we used data from The Columbia University Study of Caribbean Hispanics with Familial and Sporadic Late Onset Alzheimer’s disease-dbGaP Study Accession: phs000496.v1.p1 (Lee et al. 2011). Gender was self-reported as either male or female, while age and years of education were considered as continuous variables. All individuals are Hispanic. CVD was defined as Yes (with the history of heart disease) or No (without history of heart disease).

Genotyping was conducted using the Illumina technique. A total of 249 SNPs within the ARHGAP26 gene and 89 SNPs within the ARHGAP10 gene were available. APOE-ε4 carriers were defined as individuals with at least one ε4 allele while non-carriers were defined as individuals with no ε4 allele (APOE-ε4-0) (Table 1).

NIA-LOAD Family Study

Three thousand seven individuals were available from the National Institute on Aging-Late Onset Alzheimer's Disease (NIA-LOAD) Family Study: Genome-Wide Association Study for Susceptibility Loci–Study Accession: phs000168.v2.p2. The details about these subjects were described elsewhere (Lee et al. 2008). Genotyping was conducted by the Center for Inherited Disease Research (CIDR) using the Illumina Infinium II assay protocol. A total of 1266 AD cases and 1279 non-AD individuals from 1386 pedigree (including 589 nuclear families) (Table 1) were included. A total of 351 individuals with CVD and 1553 without CVD was available. A total of 155 SNPs within the ARHGAP26 gene and 62 SNPs within the ARHGAP10 gene were available.

Statistical Methods

Genotype Quality Control

The Golden Helix Software (https://www.goldenhelix.com/products/SNP_Variation/index.html) was used to assess control genotype data for conformity with Hardy–Weinberg equilibrium (HWE). To deal with population stratification, the principal-component analysis approach (Price et al. 2006) in HelixTree was used to obtain first three principal components (PCs) for both data. Call rate and minor allele frequency (MAF) was determined for each SNP using The Golden Helix Software. SNPs with MAF < 1%, and/or call rate < 95%, and/or HWE < 10−4 were removed for further association analyses.

Multivariable Logistic/Linear Regression Models in PLINK Software

For the case–control designs, the multivariable logistic regression analyses of risk of AD and CVD as binary traits, adjusted for gender, age, education, racial group, APOE-ε4 genotype, and first three PCs were performed. The asymptotic p-values for the logistic regression models were observed while the odds ratio (OR) and its 95% confidence interval were estimated. For continuous outcomes such as FAQ, CDR-SB, and t-tau, the multivariable linear regression analyses, adjusted for gender, age, education, racial group, APOE-ε4 genotype, and first three PCs, were performed. The multivariable logistic/linear regression models were performed using PLINKv1.9 (www.cog-genomics.org/plink/1.9/ (Chang et al. 2015). To deal with multiple testing, Bonferroni correction was used for statistical significance.

Haplotype Analysis in the ADNI Sample

Pairwise linkage disequilibrium (LD) statistics (D′) were assessed for controls, and haplotype blocks were constructed using HAPLOVIEW software (Barrett et al. 2005). Haplotype analysis of AD and CVD was performed using the multivariate logistic model using the PLINK v1.07 (Purcell et al. 2007). The asymptotic p-values for the logistic regression models were observed while the OR and its standard error were estimated using PLINK v1.07.

Family-Based Association Study

A family-based association analysis for AD and CVD was performed using the PLINK v1.07 (Purcell et al. 2007). The DFAM procedure in PLINK implements the sib-transmission disequilibrium test (TDT) and also allows for unrelated individuals to be included. This test can be used to combine discordant sibship data, parent–offspring trio data, and unrelated case/control data in a single analysis.

Functional Analysis In Silico

Functional annotation of SNPs was performed using three online functional prediction websites: SNPinfo (https://snpinfo.niehs.nih.gov/snpinfo/snpfunc.html), RegulomeDB (http://regulomedb.org/), and the Genotype-Tissue Expression (GTEx) (www.gtexportal.org). First, the associated SNPs were examined for microRNA binding sites, splicing sites, regulatory gene regions, and species-conserved regions using NIH-SNP Function Prediction (http://snpinfo.niehs.nih.gov/cgi-bin/snpinfo/snpfunc.cgi). Second, the potential functional consequences of associated SNPs were accessed using RegulomeDB (Boyle et al. 2012), which has a self-developed score system with a score ranging from 1 to 7, where a higher score indicates less functional significance. Third, the expression quantitative trait loci (eQTLs) of associated SNPs were assessed by the GTEx database.

Data Availability Statement

The datasets used during the current study are available in the (Alzheimer’s Disease Neuroimaging Initiative (ADNI)) repository, (http://adni.loni.usc.edu/data-samples/access-data/), (dbGaP:phs000496.v1.p1 Columbia University Study of Caribbean Hispanics and Late Onset Alzheimer’s disease) repository, (https://www.ncbi.nlm.nih.gov/gap/?term=phs000496.v1.p1), and (National Institute on Aging—Late Onset Alzheimer's Disease Family Study: Genome-Wide Association Study for Susceptibility Loci dbGaP Study Accession: phs000168.v2.p2) repository, (https://www.ncbi.nlm.nih.gov/projects/gap/cgi-bin/study.cgi?study_id=phs000168.v2.p2).

Results

Genotype Quality Control and Descriptive Statistics

The characteristics of participants of the three samples are presented in Table 1. After merging SNPs and phenotypes, 755 individuals in the ADNI sample kept in the study include 173 individuals with AD, 368 with MCI, and 214 with CN, while 2307 individuals in the Caribbean Hispanics sample were remained including 864 individuals with AD and 1412 with CN. The mean ages were 75.20 years, 76.13 years, and 75.53 years for ADNI sample, Caribbean Hispanics sample, and NIA family-based study, respectively. There were slightly more males (59.2%) than females in the ADNI sample, and more females than males in the Caribbean Hispanics sample (67.6%) and the NIA family-based study (64.6%).

Single Marker Association Analysis in the ADNI Sample

A total of 14 SNPs within the ARHGAP26 gene were associated with AD (p < 0.05) in the ADNI sample, and the best associated SNP was rs3776362 (p = 3.43 × 10−3). Furthermore, 37 SNPs revealed associations with CVD (top SNP rs415235 with p = 2.06 × 10−4). Interestingly, 13 SNPs were associated with both AD and CVD (top 3 SNPs for AD rs3776362, rs245852, and rs35291 with p = 3.43 × 10−3, 9.47 × 10−3, and 9.996 × 10−3, respectively, and p = 8.66 × 10−3, 2.91 × 10−3, and 4.16 × 10−3 for CVD, respectively). The SNPs with p < 0.05 with AD or CVD are presented in Table S1. The top 10 SNPs associated with AD or CVD are presented in Table 2. The top two SNPs (rs415235 and rs1370982) showing significant associations with CVD remained significant after a Bonferroni correction (p < 0.05/137 = 3.65 × 10−4). The best AD-associated SNP, rs3776362, was further associated with FAQ, and CDR-SB (p = 0.0229 and 0.0411, respectively). Within the ARHGAP10 gene, four SNPs were associated with AD (top SNP rs6836509 with p = 0.0278), but no SNP was found to be associated with CVD. Interestingly, rs6836509 in the ADNI sample was associated with t-tau protein (p = 4.63 × 10−4), and rs17475681 was associated with FAQ and CRD-SB and FAQ (p = 0.0311 and 0.0152, respectively).

Table 2 SNPs of ARHGAP26 and ARHGAP10 genes associated with AD and CVD in the ADNI sample
Full size table

Haplotype Analyses in the ADNI Sample

Using HAPLOVIEW software, we identified three haplotype blocks including top AD-associated and CVD-associated SNPs within the ARHGAP26 gene in the ADNI sample. Figure 1 shows the LD (D′) structure. The G-C haplotype from rs13161500 and rs1370982 (D′ = 0.85) revealed significant associations with AD and CVD (p = 0.0126 and 1.27 × 10−4, respectively) (Table 3). The G-A haplotype from rs3776362 and rs1347813 (D′ = 1.00) revealed significant associations with AD and CVD (p = 3.42 × 10−3 and 8.43 × 10−3, respectively).

Fig. 1
figure 1

Haplotype structure (D′) of SNPs within ARHGAP26 at 5q31.3 including the top SNPs with AD and CVD

Full size image
Table 3 Haplotypes of ARHGAP26 associated with AD and CVD in the ADNI sample
Full size table

Association Analysis in Caribbean Hispanics Sample

Single marker analysis showed that 17 SNPs within the ARHGAP26 gene were associated with AD (p < 0.05) and the best associated SNP was rs2107622 with p = 5.59 × 10−4. Furthermore, 12 SNPs were associated with CVD (top SNP rs6580265 with p = 2.91 × 10−3). Interestingly, two SNPs (rs2107622 and rs245828) were associated with both AD and CVD (p = 5.59 × 10−4 and 2.50 × 10−2 for AD and p = 2.35 × 10−2 and 2.92 × 10−2 for CVD, respectively). The SNPs with p < 0.05 with AD or CVD in Caribbean Hispanics sample are presented in Table S2. The top 10 SNPs associated with AD or CVD are presented in Table 4. Within the ARHGAP10 gene, two SNPs (rs6845865 and rs10003349) were associated with AD, and several SNPs were associated with CVD.

Table 4 SNPs of ARHGAP26 and ARHGAP10 genes associated with AD and CVD in the Caribbean Hispanics sample
Full size table

Association Analysis in NIA Sample Using a Family-Based Study Design

A family-based study showed that nine SNPs within the ARHGAP26 gene were associated with AD and three SNPs were associated with CVD in the NIA sample (Table 5). Four AD-associated SNPs (rs245852, rs37212, rs3776362, and rs37210) within the ARHGAP26 gene from the NIA sample were also associated with AD and CVD in the ADNI sample (Table 2). Within the ARHGAP10 gene, four SNPs were associated with AD, but no SNP was associated with CVD.

Table 5 SNPs of ARHGAP26 and ARHGAP10 genes associated with AD and CVD in the NIA sample
Full size table

In Silico Analysis

In the ADNI sample, rs3776362 associated with both AD and CVD was located at the gene regulatory region, and rs415235 associated with CVD was located at the species-conserved region. In the Caribbean Hispanics sample, rs7727006 associated with CVD was located at the gene regulatory region, and rs11167807 associated with AD was located at the species-conserved region.

RegulomeDB returned score of ≤ 3 is considered functionally important. Within the ARHGAP26 gene, three SNPs, rs3776362 (with 2b), rs463717 (3a), and rs1347813 (3a), associated with AD and CVD are functionally important in the ADNI sample (Table 2). Four SNPs, rs258764 (3a), rs11167807 (3a), and rs6894946 (2b) associated with AD, and rs17707291 (3a), associated with CVD are functionally important in the Caribbean Hispanics sample (Table 4). Two SNPs, rs3776362 (2b) and rs3776309 (2b), associated with AD are functionally important in the family-based study (Table 5).

Within the ARHGAP26 gene, a total of 14 SNPs in Table 2, 9 SNPs in Table 4, and 4 SNPs in Table 5 had significant eQTLs. Within the ARHGAP10 gene, a total of 2 SNPs in Tables 2, 3 SNPs in Table 4, and 4 SNPs in Table 5 had significant eQTLs. A more comprehensive list of SNPs with eQTLs is presented in supplementary Table S3.

Discussion

The present study examined the associations of the ARHGAP26 gene polymorphisms with AD and CVD using three datasets. In the ADNI sample, there were 13 SNPs within the ARHGAP26 gene associated with both AD and CVD, while in the replication sample using a case–control sample, there were 2 SNPs associated with both AD and CVD. The results of an additional sample using a family-based study showed that nine SNPs were associated with AD, and three SNPs were associated with CVD. The top SNP rs3776362 associated with AD in the ADNI sample was associated with cognitive phenotypes (FAQ and CDR-SB). Haplotype analysis further supported the single marker of the ARHGAP26 gene in associations with AD and CVD in the ADNI sample. Several SNPs of this gene are functionally important using the RegulomeDB data. Within the ARHGAP10 gene, top SNP rs6836509 associated with AD and t-tau protein level in the ADNI sample. Additionally, over 20 SNPs of ARHGAP26 and ARHGAP10 genes had eQTLs using the GTEx database.

The ARHGAP26 (also known as GRAF) gene is a protein coding gene. Previous studies have shown that GRAF mRNA is expressed in a variety of avian tissues and is particularly abundant in brain and liver (Hildebrand et al. 1996; Lucken-Ardjomande Hasler et al. 2014). GRAF may be a regulator of the Rho family of small GTP-binding proteins that exhibit binding to a protein tyrosine kinase (Hildebrand et al. 1996; Shibata et al. 2001). Indeed, the GRAF gene is involved in leukemia (Bojesen et al. 2006; Borkhardt et al. 2000; Hildebrand et al. 1996), human brain cancer (Wang et al. 2013), gastric cancer (Wangxia et al. 2019), and ovarian cancer (Chen et al. 2019). ARHGAP26 is also critical for muscle development and may be involved in RhoA activity and progressive muscle degeneration (Doherty et al. 2011; Li et al. 2019). In addition, ARHGAP26 might be involved in the pathogenesis of mental retardation (Barresi et al. 2010) and neuropsychiatric diseases (Dahm et al. 2014). Interestingly, one recent study suggested that ARHGAP26 is involved in the development of dyslexia, a heritable neurodevelopmental disorder characterized by difficulties in reading and writing (Naskar et al. 2018). The patients with dyslexia are similar to patients with dementia who have language deficiencies with significantly lower results in the area of verbal expression, hearing comprehension, repetition, reading, and writing (Banovic et al. 2018). In the present study, we found ARHGAP26 gene polymorphisms were associated with AD and CVD. Especially, in the ADNI sample, there were 13 SNPs of the ARHGAP26 gene associated with both AD and CVD, while in the replication sample using a case–control sample, there were 2 SNPs associated with both AD and CVD. Additionally, four SNPs (rs245852, rs37212, rs3776362, and rs37210) were associated with AD both in ADNI and the family-based study. These findings suggest a shared genetic susceptibility among AD and CVD. To our knowledge, this is the first study that investigates the associations of ARHGAP26 gene polymorphisms with AD and CVD.

AD is the most common form of age-related neurodegenerative disorder with dementia (Lane et al. 2018; Lipnicki et al. 2017). Both AD and PD are neurodegenerative disorders and may share genetic, epigenetic, and environmental risk factors (Cacabelos 2020; Cifelli et al. 2019; Desikan et al. 2015; Jabir et al. 2015; Kelly et al. 2020; Ramanan and Saykin 2013; Ray et al. 2008). Although AD, vascular dementia (VD), and PD exert increasingly lethal or disabling effects, the associations among these diseases at the molecular level remain unclear (Chen et al. 2021). Interestingly, one recent study found that ARHGAP26 mRNA was significantly upregulated in PD, suggesting that the gene ARHGAP26 may be associated with PD (Jiang et al. 2019). Furthermore, it is suggested that common molecular mechanisms underlie the development of AD, PD, and cancer (Aliev et al. 2013; Driver et al. 2012; Forés-Martos et al. 2021; Jabir et al. 2015; Roe et al. 2010, Sharma et al. 2020). Previous findings and the current study suggest that ARHGAP26 may be one of the mechanisms to link AD, CVD, cancer, and PD; however, a confirmation study and a gene functional analysis are needed.

We also examined the association of the ARHGAP10 gene with AD and CVD. The ARHGAP10 gene, an important paralogon of ARHGAP26, is located at 4q31.23 (Katoh and Katoh 2004; Koeppel et al. 2004; Shibata et al. 2001). ARHGAP10 mRNA is expressed in the brain, heart, skeletal muscle, and testis (Ren et al. 2001; Shibata et al. 2001). Recently, ARHGAP10 gene mutations have been found to play a pathophysiologic and pathogenic role in schizophrenia (Hada et al. 2021; Sekiguchi et al. 2020). However, no study has focused on the effect of ARHGAP10 gene on AD and CVD. The present study showed that several SNPs in each of the three samples were associated with AD. Interestingly, rs6836509 in the ADNI sample was also associated with t-tau protein, and rs17475681 was associated with CRD-SB and FAQ. Additionally, several SNPs were found to be associated with CVD only in the Hispanic sample. These findings may suggest that ARHGAP10 gene may play a role in AD and related cognitive functions, but no SNPs on this gene have been found to be associated with both AD and CVD.

Previous studies have shown that the co-occurrence of AD and CVD might be due to shared risk factors between AD and CVDs, but there might also be a direct causal association of CVD and cardiovascular risk factors with AD (de la Torre 2012; Justin et al. 2013; Attems and Jellinger 2014; de Bruijn and Ikram 2014; Jabir et al. 2015; Monte and Vondriska 2014; Tini et al. 2020). AD and CVDs may also share genetic factors (Carter et al. 2017; Moreira et al. 2005; Mur et al. 2020; Nicot et al. 2007; Tan et al. 2013; Tini et al. 2020). Genome-wide association study also found that there may be a link between AD and CVDs using pathway analysis (Liu et al. 2014). A recent study identified 90 SNPs on 19 different chromosomes that jointly conferred increased risk for AD and cardiovascular outcomes (Broce et al. 2019). Our current findings and previous reports suggest three possible scenarios for genetic variants shared by AD and CVD: (1) genetic variants influence cardiovascular risk factors and AD independently, or (2) genetic variants influence AD through multiple cardiovascular risk factors (Broce et al. 2019), and/or (3) those disease-associated variants involved in common biology pathways result in the disease development. However, some genes could interact with cardiovascular risk factors in the development of AD (Bessi et al. 2020). Additionally, epigenetic influence at cellular and molecular levels in cancer, metabolic and cardiovascular complications are also involved in the development of neurological (psychosis) and neurodegenerative disorders such as AD and PD (Sharma et al. 2020).

There are several strengths of this study. First, we used a total of three samples: a population-based case control (ADNI) sample, a case control Caribbean Hispanic sample, and a family-based sample. A family study offers several advantages and has long been the favored approach for genetic study because of (1) reducing the type 1 error rate arising from population stratification and the NIA family–based study confirmed the results of the ADNI sample; (2) possessing several advantageous features for linkage and association tests compared with using samples of related subjects; (3) remaining a powerful and advantageous approach in complex trait genetics; and (4) reducing heterogeneity for the study due to related family members who tend to share similar lifestyles and genetic factors for a trait. Second, three samples revealed that several common genetic variants were associated with both AD and CVD, while top AD-associated SNP rs3776362 in the ADNI sample was also associated with cognitive phenotypes such as FAQ and CDR-SB. In addition, we performed functional analyses and found several SNPs with eQTLs.

There are a number of limitations in this study. The three samples may have ethnic heterogeneity even though there is a small proportion of Hispanic population representation in the ADNI sample. The majority in the ADNI sample and the NIA-LOAD family-based sample are non-Hispanic White; however, the Caribbean Hispanic sample is a genetic heterogeneous population. Furthermore, different genotyping platforms may lead to a limited overlap of the SNPs in three sets of samples. Finally, the results in the present study might be spurious or subject to type I error due to the large number of analyses performed in this study.

Conclusion

ARHGAP26 genetic variants are significantly associated with the risks of AD and CVD in three independent samples. A number of SNPs within the ARHGAP26 gene associated AD and CVD are functionally important or have had significant eQTLs based on in silico analysis. Furthermore, one AD-associated SNP rs3776362 in the ADNI sample was also associated with cognitive phenotypes. In addition, several ARHGAP10 genetic variants are significantly associated with the risks of AD in these independent samples. Within the ARHGAP10 gene, top SNP rs6836509 associated with AD in the ADNI sample was associated with t-tau protein. These results suggest a potential role of ARHGAP26 and ARHGAP10 in the pathogenesis of AD and/or CVD. Future functional study of these two genes may help to better characterize the genetic architecture of AD and CVD.