New Genetic Associations Between Alzheimer's Disease and Its Key Risk Factors

ABSTRACT

Introduction

This study aimed to explore shared genetic architectures underlying Alzheimer's disease (AD) and its known risk factors.

Methods

Significant common variants between AD and its risk factors were identified using GWAS data. The 1000 Genomes Project genotyping data enabled the detection of linkage disequilibrium (LD) blocks and haplotype structures. Functional impact assessments, protein–protein interaction analyses, pathway mapping and enrichment studies were performed.

Results

Sixteen significant variants across nine genes were associated with AD and at least one risk factor (p ≤ 5 × 10⁻⁸). Genes APOE, ABCA1 and TOMM40 showed strong associations with AD (adjusted p = 9.75 × 10⁻⁹). High‐confidence interactions were identified among these genes, as well as APP and LRP1, within the AD pathway. Variant rs429358 (p ≤ 3 × 10⁻¹⁵) on the APOE gene was linked to AD, metabolic syndrome (MetS), diabetes, waist‐to‐hip ratio (WHR) and ageing. Variant rs2075650 (p ≤ 6 × 10⁻⁹) on TOMM40 correlated AD risk with MetS, WHR and body mass index (BMI). Variants rs483082 (p ≤ 2 × 10⁻³²) and rs71352238 (p ≤ 1 × 10⁻¹¹) on APOC1 and TOMM40 were associated with AD and MetS. Variants rs4420638 (p ≤ 2 × 10⁻³⁴) and rs1800978 (p ≤ 2 × 10⁻⁹) on APOC1 and ABCA genes were associated with AD and WHR. The rs13237518 (p ≤ 5 × 10⁻¹¹) was associated with AD risk in diabetic patients. Furthermore, the rs4277405 (p ≤ 9 × 10⁻²⁰) associated AD with cardiovascular disease (CVD). Haplotypic structures were also identified for all these variants (D′ and r ² ≥ 0.8).

Discussion

This study identifies genetic variants and LD blocks on APOE, ABCA1, TOMM40 and APOC1 genes shared between AD and its risk factors, revealing common genetic links and potential shared susceptibility pathways.

1. Background

Alzheimer's disease (AD) is one of the most important public health challenges and the leading cause of dementia, accounting for 60%–70% of cases [1, 2]. The prevalence of dementia has increased as the seventh leading cause of death in recent decades, and is projected to triple worldwide by 2050 [3, 4]. Despite extensive research efforts, effective treatments for AD remain elusive [5]. Several risk factors influence the development of AD. While modifiable factors contribute to nearly 30% of AD cases, such as obesity, diabetes, depression, heart disease, hypertension, alcohol consumption and smoking [6], approximately 70% of the risk of developing AD is attributed to genetics as a non‐modifiable risk factor [7]. To date, large‐scale GWAS meta‐analyses have identified more than 100 significant genomic loci associated with AD risk, and ongoing studies continue to expand the number of implicated loci [8, 9, 10, 11].

The interplay between obesity and AD is multifactorial and complex, and various obesity‐associated mechanisms influence the pathogenesis of AD. Because obesity‐induced chronic neuroinflammation may play an important role in the pathophysiology of AD, its imaging techniques in obese individuals may provide valuable insights into the future risk of developing AD [12]. Personalised therapeutic and risk stratification may play a key role in increasing the effectiveness of anti‐inflammatory interventions. Such strategies could help delay the onset of cognitive decline in individuals at risk of AD due to comorbid conditions like hypertension, obesity, or type 2 diabetes (T2D), which are known risk factors for neurodegenerative disease [13, 14]. Similar abnormalities are observed in the same brain regions in patients with AD and these comorbid conditions, such as T2D. For example, abnormal activation of GSK3β and DYRK1A, as two AD markers involved in the pathologies of T2D, leads to tau hyperphosphorylation and abnormal degradation of amyloid precursor protein (APP) [15]. Given the concerning rise of these conditions in ageing populations, it is crucial to unravel the pathophysiological mechanisms linking AD with its common comorbidities to identify common molecular targets for therapeutic interventions.

Variables derived from the Framingham Heart Study Dementia Risk Score (FDRS) provide reliable predictions for AD and AD‐related dementias (ADRD) among populations with heart failure and atrial fibrillation. The incorporation of comorbidities and risk factors has only slightly enhanced predictive accuracy, suggesting that FDRS variables are appropriate for forecasting AD/ADRD in patients with heart failure and atrial fibrillation [16]. Hypertension exerts a multifaceted influence on the development of AD. Individuals with hypertension show increased AD pathology and a greater burden of white matter hyperintensity (WMH). The pathology of both conditions is independently associated with reduced thickness of the entorhinal cortex (EC) and impaired memory performance. Furthermore, regional reductions in cerebral perfusion, likely resulting from Aβ accumulation and hypertension, correlate with WMH burden in posterior brain regions and are interactively associated with elevated tau levels in the EC [17]. The H3K18 lactylation/NFκB pathway modulates age‐related inflammation (inflammaging) by regulating components of the senescence‐associated secretory phenotype, such as IL‐6 and IL‐8, thereby contributing to accelerated brain ageing and the emergence of pathological features in AD. This pathway may significantly influence the initiation and exacerbation of chronic inflammation in the brain, a critical factor in AD development [18].

Genetic correlation analyses of AD have been conducted in previous studies and have identified shared genetic factors between AD and a range of traits, including related dementias, type 2 diabetes, obesity and cardiovascular diseases [19, 20, 21, 22]. These findings highlight pleiotropic effects across traits and emphasise the importance of investigating the complex shared genetic architecture between AD and its established risk factors at multiple levels, including variant, gene, pathway and interaction‐network levels.

Given the importance of AD and the incomplete understanding of genetic associations between AD and its risk factors, this study aimed to explore shared genetic architectures underlying AD and its established risk factors, in order to identify common variants, genes and biological pathways contributing to AD susceptibility.

2. Methods

2.1. Study Pipeline

As illustrated in Figure 1, significant GWAS variants associated with AD, T2D, metabolic syndrome (MetS), ageing, smoking, depression, heart disease, Down's syndrome, hypertension and alcohol consumption were extracted from the GWAS catalogue. Common variants between AD and its risk factors were then identified. Linkage disequilibrium (LD) blocks were constructed using 1000Genome LD data. Following this, 1000Genome phase 3 genotyping data were employed to identify haplotypic structures associated with AD and its risk factors. Pathway and protein–protein interactions (PPI), and an OMIM investigation were conducted to prioritise and identify gene interactions between AD and its associated risk factor diseases or traits. Finally, the functional impact of the variants was assessed using ClinVar, GTExPortal and RegulomeDB.

FIGURE 1.

Study pipeline.

2.2. GWAS Variants Associated With AD and Its Risk Factors

To identify significant GWAS variants associated with AD and its risk factors, the GWAS catalogue EMBL‐EBI (gwas_catalogue_v1.0.2‐associations_e109_r2023‐05‐07) was downloaded (https://www.ebi.ac.uk/gwas/) [23]. The ‘disease/treat’ and ‘mapped variants’ sections were utilised to identify variants associated (p ≤ 5 × 10⁻⁸) [24, 25] with AD and its risk factors. The R programming language (version 4.2.1) was employed to identify common variants associated with both AD and risk factors, including MetS, body mass index (BMI), waist‐to‐hip ratio (WHR), obesity, T2D, ageing, smoking, cardiovascular disease (CVD) and alcohol consumption, which were designated as index variants. HaploReg v4.2 (https://pubs.broadinstitute.org/mammals/haploreg/haploreg.php) was used to find proxy variants that are in LD with index variants (with the highest degree: r ² ≥ 0.8 and D′ ≥ 0.8) [26]. GWAS LD variants were identified by combining index and proxy variants.

2.3. Haplotypic Structures Associated With AD and Its Risk Factors

To identify haplotypic blocks associated with AD and its risk factors, 1000 genomes phase 3 genotyping data for GWAS LD variants (derived from 2504 samples) were downloaded from the Ensembl Genome Browser 113 (https://asia.ensembl.org/index.html) [27]. Haplotypic blocks and LD plots were generated using HaploView V4.2. LD blocks were identified using the confidence intervals (default) algorithm and considering only genotypes with (100%) genotype frequency [28].

2.4. Enrichment, Pathway Analysis and Protein Interactions

Enrichment analysis on candidate genes was performed using the R programming language (enrichR package) connected to the Enrichr web server. The candidate gene list was examined in several selected databases, including KEGG 2021 Human, OMIM Disease and GWAS Catalogue 2023. For each database term, Enrichr calculated the significance of enrichment using Fisher's exact test to determine whether the overlap between the input genes and the annotated gene set was greater than expected by chance (genes were considered independent), then p values were adjusted for multiple testing using the Benjamini–Hochberg (false discovery rate) method provided by Enrichr. Enrichment results are presented based on adjusted p values and odds ratios [29, 30].

The candidate genes identified from GWAS data were compared with KEGG AD pathway (map05010) and the PPI were investigated using STRING Version 12.0 (https://string‐db.org/) with a high confidence score cutoff of 0.7. The significance of network connectivity was assessed using the PPI enrichment p value, which was calculated via Fisher's exact test [31].

2.5. Haplotypic GWAS SNP Functional Analysis

The gene associated with each variant was identified using dbSNP (https://www.ncbi.nlm.nih.gov/snp/). RegulomeDB (https://regulomedb.org/regulome‐search/) was utilised to determine the likelihood of variants being functional (a lower number indicates a higher score). Variants of expression quantitative trait loci (eQTL) were identified by checking the Genotype‐Tissue Expression (GTEx) portal (https://gtexportal.org/home/). To investigate the clinical significance of the identified variants, ClinVar (https://www.ncbi.nlm.nih.gov/clinvar/) was employed.

3. Results

3.1. Significant Variants Associated With AD and Its Risk Factors

Significant GWAS variants associated with the risk of AD and its risk factors were extracted from the GWAS catalogue. Sixteen significant genetic variants were associated with AD and various diseases/traits, including MetS, BMI, WHR, obesity, ageing, T2D, CVD and smoking. These variants were located on nine genes and were predominantly intronic variants (Table 1). No significant variants were associated with depression, high blood pressure, alcohol consumption or Down's syndrome.

TABLE 1.

GWAS significant variants associated with AD and its risk factors.

Gene	Variants	Type of variant	AD/risk factors	p	Population	Overall sample size (case/control)	References
APOE	rs429358	Coding	AD	5 × 10−286	European	14 895 (7316/7579)	[32]
			Ageing	1 × 10−126	European	837415*	[33]
			MetS	1 × 10−14	Korean	62 314 (10 684/51 630)	[34]
			WHR	4 × 10−37	European	697734**	[35]
			T2D	3 × 10−15	European	251740***	[36]
	rs7412	Coding	AD	2 × 10−31	European	12 772 (5705/7067)	[37]
			BMI	2 × 10−17	European	1122049**	[38]
			CVD	2 × 10−19	European/unknown	63 731 (18 467/45 264)	[39]
	rs769449	Intronic, 3′UTR	AD	2 × 10−83	European	17 480 (2741/14 739)	[40]
			BMI	2 × 10−49	European	1 122 049**	[38]
TOMM40	rs2075650	Intronic	AD	1 × 10−295	European/Hispanic	25 610 (3006/22 604)	[41]
			MetS	3 × 10−11	Taiwanese	107 230 (24 171/83 059)	[42]
			BMI	3 × 10−9	European/African American/Hispanic	238 944 (236 781; 887; 1276)**	[43]
			WHR	6 × 10−9	European	143 480**	[44]
	rs157582	Intronic	AD	4 × 10−63	European	17 480 (2741/14 739)	[40]
			MetS	1 × 10−8	European	10 564 (2637/7927)
	rs405697	Coding	AD	1 × 10−13	European	17 480 (2741/14 739)	[40]
			MetS	1 × 10−15	European	1 384 348**	[34]
	rs71352238	Upstream	AD	7 × 10−69	European	17 480 (2741/14 739)	[40]
			MetS	1 × 10−11	Taiwanese	107 230 (24 171/83 059)	[42]
APOC1	rs483082	Upstream	AD	1 × 10−15	European/African American/Japanese/Israeli‐Arab	33 263 (13 100/13 220; 1472/3511; 951/894; 51/64)	[45]
			MetS	2 × 10−32	European	291 107	[46]
	rs4420638	Downstream	AD	8 × 10−149	European	2229 (1291/938)	[47]
			WHR	5 × 10−8 (borderline)	European	663 598**	[48]
NECTIN2	rs6857	3′UTR	AD	7 × 10−18	European/African American/Japanese/Israeli‐Arab	33 263 (13 100/13 220; 1472/3511; 951/894; 51/64)	[45]
			MetS	4 × 10−28	European	1 384 348**	[34]
			BMI	6 × 10−26	European	806 834**	[35]
			T2D	7 × 10−9	European/African American/East Asian/Latino	71 604 (7111/49 688; 746/2825; 743/3992; 1147/5352)	[49]
	rs6859	Intronic, 3′UTR	AD	5 × 10−27	European	12 772 (5705/7067)	[37]
			BMI	4 × 10−8	European	1 122 049**	[38]
TMEM106B	rs13237518	Intronic	AD	5 × 10−11	European	487 511 (85 934/401 577)	[22]
			T2D	2 × 10−13	European/Hispanic/African American/Asian	1 407 282 (148 726/965 732; 24 646/,31 446; 8616/11 829; 46 511/169 776)	[50]
PPIAP55	rs4277405	Upstream	AD	9 × 10−20	European	487 511 (85 934/401 577)	[22]
			CVD	7 × 10−25	European	~459 000****	[51]
PTPRG	rs7609954	Intronic	AD	4 × 10−8	European	3457 (2478/979)	[52]
			Smoking	1 × 10−10	European/East Asian/Hispanic or Latin American/African	3 382 012 (1 432 937/1 949 075)	[53]
ABCA1	rs1800978	5′UTR	AD	2 × 10−9	European	487 511 (85 934/401 577)	[22]
			WHR	9 × 10−15	European	697 734**	[35]
SPATC1	rs79832570	Intronic	AD	4 × 10−13	European	451 469 (78 709/372 760)	[54]
			BMI	6 × 10−9	European	1 122 049**	[38]

^*The sample size corresponds to the effective European‐ancestry sample reported in the GWAS Catalogue for the association p value. This study represents a multivariate meta‐analysis of multiple GWASs (health span, parental lifespan, longevity), and the reported sample size is an aggregated estimate.

^**BMI, WHR and MetS were analysed as continuous traits; classical case–control groups are not applicable.

^***Original multi‐ancestry GWAS included 180 834 T2D cases and 1 159 055 controls; the GWAS Catalogue reports an effective sample size of 251 740 for the European‐ancestry subset relevant to the variant.

^****FINDOR method was applied to existing GWAS for CVD and other traits. Numbers reflect average sample sizes; no novel GWAS was performed, so case–control breakdowns are not available.

3.2. LD Blocks and Haplotypes

This study revealed several common haplotypes shared between AD and its risk factors. The GT is related to the GWAS significant eQTL variant rs429358, which has a good RegulomeDB ranking of 1f. This haplotype is located on the APOE gene and is shared between AD and four other traits, including MetS, WHR, T2D and ageing (D′ = 0.99 and r ² = 0.39). Additionally, the ACC haplotype, associated with the GWAS significant eQTL variant rs2075650 and a high RegulomeDB ranking of 1b, is located on the TOMM40 gene and is common among AD, MetS, BMI and WHR (D′ ≥ 0.94 and r ² ≥ 0.90). These results are illustrated in Figure 2.

FIGURE 2.

LD blocks and haplotypic structures shared between AD and (A) MetS, WHR, diabetes and ageing and (B) MetS, WHR and BMI.

Two haplotypes, GA (D′ = 0.99 and r ² = 0.99) and GCC (D′ ≥ 0.98 and r ² ≥ 0.83), associated with the significant eQTL variants rs4420638 and rs1800978, both with a high RegulomeDB ranking of 1f, are located on the APOC1 and ABCA1 genes and are shared between AD and WHR. These results are presented in Data S1.

The GCATTAAAATCACAG haplotype (D′ ≥ 0.94 and r ² ≥ 0.88), associated with the significant eQTL variant rs13237518 on the TMEM106B gene and with a good RegulomeDB ranking of 1f, is common between AD and diabetics. The result is shown in Data S2. Two haplotypes, GC (D′ = 0.99 and r ² = 0.93) and CGGT (D′ ≥ 0.95 and r ² ≥ 0.88), related to rs483082 (APOC1) and rs71352238 (TOMM40) were shared between AD and MetS. The results are displayed in Data S3.

The ATAAAG haplotype (D′ ≥ 0.92 and r ² ≥ 0.79), associated with the significant eQTL variant rs4277405 from GWAS and with a good RegulomeDB ranking of 1f, is common between AD and CVD. The results are illustrated in Figure 3.

FIGURE 3.

LD blocks and haplotypic structures shared between AD and CVD.

3.3. Enrichment Analysis

In the enrichment analysis of GWAS_Catalogue data, the genes APOE, ABCA1, TOMM40, NECTIN2, SPATC1 and TMEM106B were significantly associated with AD (adjusted p value 9.75 × 10⁻⁹). Complete results are presented in Data S4. To evaluate the potential impact of gene‐size bias on enrichment results, genomic lengths of all candidate genes were retrieved from Ensembl and compared between AD‐associated and non‐associated genes. No significant difference in gene‐length distribution was observed between the two groups (Wilcoxon rank‐sum test, p = 0.7143), suggesting that the observed enrichments were unlikely to be driven by gene size. In the OMIM data, the APOE gene was significantly associated with AD (adjusted p value 1.2 × 10⁻²; Data S5). In addition, the ClinVar investigation indicated that the rs429358 variant on APOE is associated with AD. Detailed results are provided in Data S6. STRING pathway analysis revealed that among the nine genes in Table 1, only the APOE gene is located in the AD pathway.

3.4. PPI Interaction Analysis

Further PPI analysis for all genes identified in Table 1 demonstrated a strong association between APOE, ABCA1, TOMM40 and APOC1 (Figure 4A; PPI enrichment p value: 4.85e‐08; confidence ≥ 0.7). High‐confidence PPIs for APOE were identified, as shown in Figure 4B. When comparing these interactions with the AD, KEGG map05010 (high confidence 0.7), APOE, LRP1, and APP were found to be common within the AD pathway (Figure 4B; PPI enrichment p value: 1.2e‐13; confidence ≥ 0.7). Finally, the interactions of all nine genes in Table 1 and the two newly identified genes LRP1 and APP were investigated with a high confidence score of 0.7. The results indicated that APP is directly associated with APOE, TOMM40 and APOC1 and also interacts indirectly with ABCA1 genes through APOE (Figure 4C; PPI enrichment p value: 1.18e‐12; confidence ≥ 0.7).

FIGURE 4.

Interactions between genes from Table 1. (A) Strong PPI between APOE with ABCA1, TOMM40, APOC1; (B) high confidence APOE PPI; (C) high confidence score interaction between AD KEGG pathway genes and Table 1 genes.

4. Discussion

Given the significance of AD and the unclear genetic associations between AD and its risk factors, this study aimed to explore shared genetic architectures underlying AD and its established risk factors. The results identify common variants, genes and biological pathways contributing to AD susceptibility and its related risk factors, highlighting potential shared susceptibility mechanisms for future studies.

The GT haplotype, associated with the rs769449 and rs429358 variants of the APOE gene, is shared across AD, MetS, WHR, BMI and ageing, highlighting potential common susceptibility mechanisms. Previous GWAS studies have identified the roles of these variants in AD [55, 56], rs769449 in BMI [38], rs429358 in diabetes [50] and ageing traits [33]. The GA haplotype, associated with rs4420638 on the APOC1 gene, is related to the risk of AD and WHR condition. Its involvement in AD [57], diabetes [58] and obesity‐related conditions, including BMI [59], and WHR [48], has been established in prior GWAS studies. In addition, from all rs4420638 LD haplotypic variant, the role of rs56131196 variant on AD risk was identified in previous studies [60]. The GC haplotype of rs483082 (APOC1) variant was only associated with AD and MetS. recent studies identified the genetic pleiotropy between AD and age‐related macular degeneration and also APOC1 and APOE as pleiotropic genes [61]. In contrast, apoE2 appears to confer a protective effect against AD through distinct and overlapping mechanisms. ApoE is known to influence amyloid and tau pathology, as well as neurodegeneration and immune responses to neuronal injury. It functions as an upstream mediator in the intricate pathways that contribute to neurodegeneration and cognitive decline positions apoE as a promising therapeutic target for AD and related dementias. Therefore, future research using multi‐omic approaches, cell‐type specific analyses, and novel APOE variants may provide main insights into the development of AD therapy [62]. Moreover, the integration of innovative research techniques and therapeutic delivery methods will be essential for advancing treatment options for AD [63]. Research indicates that apoE pools in the brain and peripheral tissues function independently. Astrocytes serve as the primary source of APOE in healthy brains; however, reactive astrocytes in proximity to Aβ plaques in AD exhibit diminished APOE levels, whereas plaque‐associated microglia demonstrate elevated APOE expression. The interactions between apoE and Aβ in the extracellular environment may be limited under physiological conditions, as both compete for receptors such as LRP1, which are essential for Aβ clearance. Additionally, neuropathological investigations have detected APOE in pyramidal neurons located in neurodegenerative regions such as the hippocampus, with minimal interaction with tau tangles, suggesting that APOE lipoprotein particles may be internalised by neurons through the LRP1 receptor, also its high confidence interaction (APOE‐LRP1) identified in our study.

The ACC haplotype, associated with the rs2075650 variant of the TOMM40 gene was identified as a shared genetic component across AD, MetS, BMI and WHR, suggesting shared underlying susceptibility pathways. Its role in AD [64, 65], MetS [42], and obesity‐related conditions including BMI [35] and WHR [44] has been documented in previous GWAS studies. In addition, previous studies identified the role of rs11556505 variant (LD proxy variant of rs2075650) in the risk of AD [66, 67]. The CGGT haplotype of rs71352238 (TOMM40) variant represents a genetic factor common to AD and Mets. Previous GWAS studies have identified the roles of this variants in AD [40] and MetS [42]. In addition, the roles of rs34342646 LD haplotypic variants of rs71352238 were identified on AD [68] and MetS [42] in previous GWAS and whole genome sequencing studies. Research conducted by Kulminski and colleagues provides compelling evidence for the significant roles of APOE–TOMM40–APOC1 variants in the risk of AD [67]. Association between TOMM40 differential expression, AD, and mitochondrial dysfunction identified in previous studies [69], suggesting a potential role for mitochondrial dysfunction in AD. The recent studies have also suggested that APOE, APOC1 and TOMM40, as impact markers of mitochondrial health, should be investigated for their association with AD [63]. Investigating high and low‐risk combined genotypes alongside brain imaging using volumetric NACC data from key temporal‐limbic regions typically associated with AD risk and progression offers a unique advantage, allowing for the generalisation of findings related to the interplay between TOMM40, APOE, and brain health [63].

The GCC haplotype, associated with the rs1800978 variant of the ABCA1 gene, shows a shared association with AD and WHR. Its involvement in AD [22] and WHR [35] has been noted in earlier GWAS studies. Evidence indicates that enhanced ABCA1 activity is beneficial for AD and dementia‐related disorders. Additionally, several mechanisms link ABCA1 function to Aβ clearance and the elevated risk mechanisms associated with APOE4 [70].

This study identified the rs13237518 variant and its associated haplotype on the TMEM106B gene as a shared genetic component between AD and diabetes. Its role in AD [22] and diabetes [50] has been established in prior GWAS studies. Genetic variations in TMEM106B are known to affect the risk and incidence of several neurodegenerative diseases, particularly AD. TMEM106B exhibits variable expression across different human brain tissues, with notably high levels in the cerebellum. The rs1990622 haplotypic LD variant may modulate TMEM106B expression in human brain tissues; this variant has been associated with both AD risk and TMEM106B expression in the cerebellum [71]. Recent studies employing cryogenic electron microscopy (cryo‐EM) have successfully isolated and characterised new protein filaments formed from a C‐terminal fragment (CTF) of the lysosomal membrane protein TMEM106B. These filaments have been extracted from post‐mortem human brain tissue associated with various neurodegenerative conditions and natural ageing. The role of TMEM106B in the context of natural ageing and neurodegenerative diseases is particularly significant [71]. Age‐dependent amyloid‐like filaments of TMEM106B form in the human brain, and the presence of abundant intraneuronal amyloid filaments in human tissues has been linked to various diseases.

The ATAAAG haplotype, associated with the rs4277405 variant located within the CYB561 and PPIAP55 genes, was identified in this study as a genetic component shared between AD and CVD. Its involvement in AD [22] and CVD [51] has been documented in previous GWAS studies. Additionally, the role of rs4291 variants in this haplotype with AD risk identified in previous studies [72], this variant was also associated with CVD risk [73]. Further research is needed to elucidate the mechanisms and causal relationships involving the association of this haplotype and its variants with AD and CVD.

The rs7609954 variant on the PTPRG gene represents a common genetic association with AD and smoking, highlighting a potential common susceptibility mechanism. The rs7609954 role in AD [52] and smoking [53] has been previously reported in GWAS studies. Single‐cell and spatial transcriptomic analyses have revealed that a specific subpopulation of microglia expressing PTPRG can induce the expression of VIRMA in neurons through intercellular signalling. The PTPRG protein, prominently expressed in neurons, enhances RNA stability by binding to VIRMA, which in turn increases the level of m6A PRKN modification. This modification reduces the stability of newly synthesised RNA, ultimately inhibiting mitophagy, inducing neuronal death and actively contributing to the progression of AD [74].

The polygenic and multifactorial nature of AD and its related risk factors represents a limitation of this study, as individual genetic variants typically exert modest effect sizes and may be influenced by environmental factors. The convergence of overlapping variants, LD regions, genes, and enriched biological pathways across traits provides insight into shared biological mechanisms underlying disease susceptibility. However, observed correlations may reflect shared genetic architecture rather than fully independent effects, limiting the ability to disentangle distinct causal contributions. Despite these limitations, gene and pathway level analyses provide a complementary framework by capturing aggregated biological signals that are not detectable at the single variant level. In addition, this study used GWAS summary statistics, which do not allow the assessment of genetic interactions at the individual level or measure cumulative risk. Therefore, the findings primarily reflect shared genetic susceptibility rather than direct interaction effects. Future studies using individual‐level genotype data and complementary genome‐wide approaches, such as genetic correlation analyses and Mendelian randomisation, will be required to further resolve shared and potentially causal biological mechanisms underlying these traits.

In summary, this study identified several structures and genes (especially APOE, ABCA1, TOMM40 and APOC1) shared between AD and established risk factors, including diabetes, CVD, obesity, as well as ageing and smoking. These findings highlight a common genetic link between AD and its associated conditions, providing insights into potential shared susceptibility mechanisms. They reveal biologically relevant connections that could serve as a foundation for future functional studies and the development of multi‐trait genetic panels.

Funding

This study was supported by the Endocrinology and Metabolism Research Institute, Tehran University of Medical Sciences (Grant 1403‐2‐221‐73223).

Ethics Statement

The study was approved by the Research Ethics Committee of Endocrinology and Metabolism Research Institute, Tehran University of Medical Sciences (IR.TUMS.EMRI.REC.1403.059).

Consent

The authors have nothing to report.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Data S1: LD blocks and haplotypic structures associated with AD and waist–hip ratio.

Data S2: LD blocks and haplotypic structures associated with AD and diabetes.

Data S3: Haplotype structures associated with AD and MetS.

Data S4: Disease/trait and gene enrichment analysis.

Data S5: OMIM enrichment analysis.

Data S6: ClinVar enrichment analysis.

Data Availability Statement

The data that support the findings of this study are available in the Supporting Information of this article.