A Meta-Analysis of the Relationship between Stroke and Alzheimer’s Disease: Therapeutic and Prognostic Implications

Abstract

Introduction

Several population-based studies have highlighted an association between stroke and dementia. Alzheimer’s disease (AD)-related dementia and vascular dementia are the most common causes of dementia, with clear pathophysiological mechanisms for the latter. Given the ongoing debate surrounding the link between stroke and AD, a systematic meta-analysis was performed to determine their relationship and the possible influence of some moderators (sex, age, and region).

Methods

We searched five databases (ISI Web of Science, Scopus, PubMed, Elsevier ScienceDirect, and Google Scholar) with no initial publication date restriction, and the last search was conducted in 2022. We included longitudinal population-based studies assessing the stroke-AD association, selecting those with reported effect sizes, standardized AD diagnosis, and an AMSTAR score ≥9. Case reports, reviews, animal studies, and non-English publications were also excluded. The meta-analysis, conducted using Comprehensive Meta-Analysis 3.1, presented pooled log odds ratios (LogOR) with 95% confidence intervals, heterogeneity analysis (Cochran’s Q, I²), and moderator analyses by age, sex, and region.

Results

The meta-analysis included 3 meta-analyses and 12 primary studies, comprising a total of 14,207 stroke cases and 1,952 AD cases. Our analysis revealed a significant association between ischemic stroke (IS), hemorrhagic stroke (HS), and microinfarcts (MI) and the risk of AD. Despite some heterogeneity across studies, no significant differences were observed in the stroke-AD association based on age, sex, or region.

Conclusion

Our study describes the risk of AD in patients with episodes of stroke (IS, HS, and MI) and suggests that the risk of AD may be higher in stroke patients than in matched controls without stroke incidence. Moreover, the moderator analysis supports the robustness of our results. The link between stroke and AD suggests that stroke may accelerate cognitive decline. This calls for tighter vascular control and indicates worse prognosis in dementia progression.

Introduction

Alzheimer’s disease (AD) is the most common cause of dementia in the elderly [1]. It is characterized by progressive accumulation of beta-amyloid (Aβ) plaques and neurofibrillary tangles, leading to neuronal dysfunction and cognitive decline [1]. By 2050, more than 13.8 million people will be affected by AD [2]. The estimated annual healthcare costs for a patient with AD over the age of 70 are over USD 81,000 and will reach USD 92,060 by 2030 [3]. However, there is no effective treatment for AD [4]. Recently, disease-modifying therapies that can change the underlying pathophysiology of AD, with monoclonal antibodies (e.g., aducanumab, bapineuzumab, gantenerumab, solanezumab, and lecanemab), have been successively developed and conducted in clinical trials [5]. These therapies do not improve cognitive function but rather slow their decline and have shown efficacy primarily in early stage AD patients with confirmed amyloid pathology [5].

Vascular and degenerative brain pathologies, such as stroke and AD, are two interrelated disorders that affect neurons in the brain and the central nervous system. Stroke is a neurological clinical sign of focal (or global) disturbance of cerebral function due to a vascular cause, including cerebral infarction, intracerebral hemorrhage, and subarachnoid hemorrhage, and is a major cause of disability and death worldwide. Stroke is the second leading cause of death and the third leading cause of disability worldwide, and its burden is rapidly increasing in low-income and middle-income countries [6].

Stroke classification distinguishes between ischemic and hemorrhagic stroke (HS), subarachnoid hemorrhage, and cerebral venous thrombosis [7]. Attending to the etiology of ischemic stroke (IS), we could distinguish atherothrombotic, small vessel disease, cardioembolic, and other causes). One in 5 stroke patients will develop dementia shortly after stroke [8]. When stroke occurs, a series of interconnected vascular and cerebral changes contribute to the progression of cognitive impairment [9]. Moreover, AD and stroke share some risk factors, including low education, sedentary lifestyle, and the presence of at least one heart disease [10]. In this study, we investigated three subtypes of stroke: IS, HS, and microinfarcts (MI). IS is defined as the sudden loss of blood flow to an area of the brain, resulting in loss of neurological function. It is caused by thrombosis or embolism, which occludes a cerebral vessel supplying a specific area of the brain [11]. Brain dysfunction caused by IS is often localized to the affected vascular territory [12]. HS is caused by bleeding in the brain due to blood vessel rupture [13]. Cerebral MI is defined as a microscopically demarcated region of cellular necrosis that is not visible upon macroscopic inspection of the brain [14]. They are commonly found in subcortical and cortical regions and are often associated with chronic hypoperfusion, small vessel disease, and impaired cerebral autoregulation. Although they are individually small, affected individuals may develop hundreds to thousands of these lesions, which cumulatively contribute to cognitive decline and have been linked to dementias, such as AD [9, 15].

The risk factors for stroke can be classified as modifiable or non-modifiable. The non-modifiable risk factors include age, sex, race/ethnicity, and heredity. Modifiable risk factors, which can be targeted for prevention, include hypertension, heart disease (particularly atrial fibrillation), diabetes mellitus, hypercholesterolemia, smoking, and alcohol abuse, among others [16]. In HS, poorly controlled hypertension is the most critical modifiable risk factor [17], and genetic predisposition has also been implicated in its development [18]. Hypertension remains the leading risk factor [19]. MI is often related to aging and vascular risk factors such as hypertension, which has been identified as the main underlying mechanism [20].

Stroke is not only a major risk factor for AD but is also closely linked to vascular dementia (VaD), a common form of dementia caused by cerebrovascular pathology [21]. VaD often arises after a stroke or due to chronic vascular insufficiency, leading to cognitive decline like AD, but with distinct pathophysiological mechanisms [22]. While AD is primarily driven by neurodegeneration, accumulation of β-amyloid plaques, and tau pathology, VaD results from ischemic damage, small vessel disease, and impaired cerebral perfusion. However, both conditions frequently coexist, leading to mixed dementia, wherein vascular and neurodegenerative processes contribute to cognitive impairment [23]. Given this overlap, distinguishing between AD, VaD, and mixed dementia remains a clinical challenge, underscoring the importance of understanding their shared and unique mechanisms [24].

Many theories have been proposed to explain the association between AD and stroke. First, the APOE ϵ4 allele is a genetic risk factor for AD and stroke [25, 26]. Recent genome-wide association studies revealed common risk genes between AD and IS, indicating common molecular pathways and pathophysiology [27]. Second, brain imaging results suggest that intracerebral vascular dysregulation may cause AD [28]. This hypothesis is based on the presence of vascular risk factors that reduce cerebral blood flow to a critical threshold, impairing the supply of essential nutrients, such as glucose and oxygen to neurons, which are necessary for maintaining normal brain function [29]. A reduction in cerebral blood flow may also favor the accumulation of Aβ, a key pathological hallmark of AD. Several studies have indicated that chronic hypoperfusion promotes Aβ deposition and exacerbates neurodegenerative processes and cognitive decline. Oxidative stress has also been established as a major pathological condition in both AD and stroke, further contributing to neuronal damage [30]. Third, Aβ, a major component of senile plaques in AD, tends to appear after stroke [31]. In a study on stroke-induced rats, the hippocampus showed a series of synergistic biochemical alterations, including microgliosis, increased Aβ protein, and cellular degeneration [32]. The presence of amyloid proteins is closely linked to neurodegenerative processes, including brain atrophy and progressive cognitive decline [33]. Finally, cognitive reserves play a fundamental role in the prevention of AD. Therefore, several studies have proposed effective interventions and protective strategies to promote cognitive reserve and thus prevent the development of AD [34].

Moreover, variables such as age, sex, and region may play a role in the development of AD and stroke. According to research, age is one of the strongest predictors of AD [35]. Several studies have suggested that patterns of brain gray matter atrophy may vary across the AD spectrum and depend on age and disease diagnosis [36, 37]. In this sense, AD is commonly categorized as either early onset or late-onset (LOAD) based on an age cutoff of typically 65 years [38]. AD affects 10–15% of individuals over the age of 65 and up to 47% of those over the age of 80 [39]. According to Van de Pol et al. [40], the hippocampal volume is independently affected by aging and AD. The presentation of LOAD is characterized by predominant impairment of anterograde episodic memory and dysfunction in additional cognitive domains such as visuospatial, language, and executive function, eventually resulting in global cognitive decline, complete dependency, and death [41]. Age is also a crucial factor in stroke. Nakayama et al. [42] concluded that age was not related to stroke; however, Kelly-Hayes et al. [43] demonstrated that the risk of stroke increased with age, with the incidence doubling every decade after the age of 45 years and more than 70% of all strokes occurring in individuals aged 65 years or older.

Differences in the risk of clinically diagnosed AD according to sex indicate that AD pathology is higher in women [44]. First, neuronal densities and estimates of the number of neurons are higher in men [45]. Second, men and women with AD exhibit different cognitive and psychiatric symptoms, and women show faster cognitive decline [46]. Third, the prevalence and effects of cerebrovascular risk factors for AD differ between men and women, with men having a higher prevalence [47]. Increasing evidence suggests that stroke is associated with sex-related differences. Barnes et al. [44] estimated, on the one hand, that the incidence of stroke is higher in men and, on the other hand, that women have a higher mortality rate than men after a stroke. In line with emerging trends, it is often not possible to assess the roles of sex (i.e., biological and pathophysiological factors) and gender (i.e., sociocultural factors) in isolation as they are closely intertwined [48]. Stroke epidemiology, diagnosis, access to care, treatment outcomes, and care received are factors that have been identified as potentially influencing the development of stroke [48].

Additionally, the research revealed differences in the prevalence of AD between regions. According to a previous study, the prevalence of AD is high in developing Asian and Latin American countries (≥5%) but consistently low (1–3%) in India and sub-Saharan Africa [21]. The lower prevalence in Africa and South Asia may be partially attributable to a lower survival rate of patients with dementia rather than a lower incidence [49]. In developed countries, the development of neurological diseases such as AD is related to population aging [50]. In fact, recent data from the study by Turana et al. [51] indicated that the prevalence of AD in Asian countries is related to age. Nevertheless, in Latin America, the higher prevalence of dementia and AD may be a result of poverty, cultural barriers, and socioeconomic vulnerability [52]. In the USA, it is estimated that 6.7 million individuals aged 65 years and older are living with AD in 2023, with projections reaching 13.8 million by 2060 if no medical advancements are made [53]. The prevalence increases significantly with age, affecting 10–15% of individuals over 65 years of age and up to 47% of those over 80 years of age. In Europe, the prevalence varies by country, but studies estimate it to be 5–7% in people over 65, with some of the highest rates observed in Southern and Eastern European countries, likely due to demographic trends and risk factor distribution [54]. In addition, East Asia (38.8%), Central Europe (31.7%), and Eastern Europe (31.6%) had the highest estimated risk of stroke among the different countries [55]. Epidemiological studies conducted in the USA, Europe, and Asia found that being overweight or obese was significantly associated with an increased incidence of IS; however, the association with HS incidence was inconsistent [56].

This study aimed to conduct a systematic review of the available literature and a meta-analysis of primary longitudinal studies that reported incidents of stroke in patients with AD. We estimated the risk of AD by comparing patients with a history of stroke (IS, HI, or MI) to those without a history of stroke. Therefore, this study estimates the effect size of the association between AD and stroke (IS, HS, and MI) and examines whether this effect size varies with different moderating variables (age, sex, and region).

The findings of this study could be important for planning healthcare resources for AD patients. Stroke is a modifiable risk factor and preventive measures can be established to avoid or delay AD. Promoting lifestyle changes, such as preventing an unhealthy diet, cardiovascular disease, hypertension, smoking, diabetes, obesity, metabolic syndrome, depression, and traumatic brain injury, could reduce the risk of stroke and AD. In addition, this study will allow clinicians to consider stroke occurrence when predicting the prognosis of patients with AD.

Method

Data Collection

To explore the association between different types of stroke (IS, HI, and MI) and AD, we conducted a systematic review and meta-analysis according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [57]. The PRISMA checklist is presented in the online supplementary material (for all online suppl. material, see https://doi.org/10.1159/000546395).

Based on other meta-analyses in this field, the following inclusion criteria were established: (1) reported effect size; (2) longitudinal meta-analyses or primary studies measuring the relationship between stroke and AD; (3) meta-analyses that provide sample data; (4) subjects with a diagnosis of AD; (5) AD diagnosed by diagnostic criteria (DSM and ICD); and (6) meta-analyses score greater than 9 on AMSTAR. Case reports, narrative reviews, letters, animal studies, articles in languages other than English, and articles reporting data on the interaction of certain types of drugs with stroke and AD were excluded.

The search was carried out using five databases: ISI Web of Science, Scopus, PubMed, Elsevier ScienceDirect, and Google Scholar. We used key search terms including “stroke,” “microvascular infarcts,” “ischemic stroke,” “hemorrhagic stroke, “meta-analysis, “dementia, and “Alzheimer’s disease.” No initial publication date was set for this study. The search was limited to English language publications and human studies. The search in the Google Scholar database was limited to titles. We also reviewed the reference lists of the relevant primary articles and reviews to identify studies that may have been missed in the search. The most recent meta-analysis was published in 2017. Finally, another search was conducted without restricting the initial publication date to identify primary longitudinal studies that were not included in the meta-analysis. The last search was conducted in 2022, and no studies have been identified. Each of the primary studies included in the meta-analyses was reviewed to select only those that met the inclusion criteria as some of these studies focused on nonspecific variables of the relationship between AD and stroke.

Data Extraction

Two authors (O.S. and A.P.) extracted key information from the studies using a preplanned form and recorded it in two separate databases, which were later compared and corrected for inconsistencies. When conflicts appeared in inclusion, exclusion, or data extraction, they were resolved by discussion or the involvement of a third reviewer (S.U.). The following variables were collected: study, year, type of stroke (IS, HS, and MI), population size (N), number of studies (K), regions, percentage of women (% F), mean age (M), main results of the study, effect sizes (odds ratio and 95% confidence limits), and AMSTAR scores. If the studies reported different types of stroke, the effect size for each type was defined.

The statistical measures used in the analysis were as follows: number of primary studies (k), logarithm of the odds ratio (LogOR), standard error (Se), 95% confidence interval (95% CI), odds ratio (OR), Q statistic for between-group heterogeneity (Qb), degrees of freedom (df), p value for statistical significance (p), and heterogeneity index (I²). For meta-analysis, we calculated the log OR (LogOR; logarithm of the OR) of AD for each type of stroke. The study results were pooled using measures of every type of stroke (IS, HS, and MI). We reported the associations between stroke and AD in each primary study (online suppl. Table 1S) included in the meta-analysis (Table 1). Statistical significance was set at p ≤ 0.05.

Table 1.

Effect sizes related to AD and IS

Study name	Statics for each study
	sample	LogOR	Se	Ve	LLIC	ULIC	Z	p value	OR	LLIC	ULIC
Cao et al. [58]
Brayne et al. [59]	Baseline: IS, N = 100; follow: IS and AD, n = 51	0.64	0.40	0.16	−0.140	1.424	1.61	0.108	1.90	0.869	4.153
Strozyk et al. [60]	Baseline: IS, N = 258; follow: IS and AD, n = 47	0.10	0.51	0.26	−0.895	1.086	0.19	0.850	1.10	0.409	2.962
Strozyk et al. [60]	Baseline: IS, N = 143; follow: IS and AD, n = 84	0.41	0.45	0.20	−0.476	1.287	0.90	0.367	1.50	0.621	3.623
Troncoso et al. [61]	Baseline: IS, N = 179; follow: IS and AD, n = 79	1.39	0.35	0.12	0.706	2.067	3.99	0.000	4.00	2.025	7.899
Zhou et al. [62]
Qiu et al. [63]	Baseline: IS, N = 2,212; follow: IS and AD, n = 303	−0.20	0.24	0.06	−0.681	0.274	−0.83	0.404	0.82	0.506	1.316
Bermejo-Pareja et al. [64]	Baseline: IS, N = 3,864; follow: IS and AD, n = 184	1.50	0.24	0.05	1.024	1.972	6.20	0.000	4.47	2.784	7.184
Hayden et al. [65]	Baseline: IS, N = 3,215; follow: IS and AD, n = 121	1.86	0.28	0.08	1.303	2.422	6.52	0.000	6.44	3.679	11.266
Lindsay et al. [66]	Baseline: IS, N = 4,236; follow: IS and AD, n = 83	0.43	0.28	0.08	−0.123	0.987	1.53	0.127	1.54	0.884	2.682
Total Random	​	0.75	0.25	0.06	0.265	1.238	3.03	0.002	2.12	1.304	3.447

AD, n, Alzheimer disease cases; IS, N, ischemic stroke cases; LogOR, log odds ratio; Se, estimator bias; Ve, variance statistics; LLIC, lower limit confidence interval; ULIC, upper limit confidence interval; Z, standard punctuation; p, statistical significance; OR, odds ratio; LLIC, lower limit confidence interval; ULIC, upper limit confidence interval.

A p value ≤0.05 is statistically significant.

Data were entered into Comprehensive Meta-Analysis version 3.1 (Biostat Inc., NJ, USA) [67]. Heterogeneity between study samples was assessed using Cochran’s Q statistic [68]. The I² statistic was calculated to express the fraction of variation between studies owing to heterogeneity [69]. The I² statistic explains the percentage of variation in the observed effects due to the variation in the actual effects. An I² value of less than 25% was considered low heterogeneity, between 25% and 50% was considered moderate heterogeneity, and >50% was considered high heterogeneity [70].

Quality Assessment

To assess the quality of all included studies, we used the 11-item Assessment of Multiple Systematic Reviews (AMSTAR) tool [71], which has been shown to have good inter-rater agreement, reliability, and content validity [72]. The total scores for meta-analyses were calculated as the sum of the 11 items on a binary scale. Quality ratings were established as low (0–4), moderate (5–8), and high (9–11).

Results

A total of 448 meta-analyses were identified in the search: 68 from ISI Web of Science, 135 from Scopus, 49 from PubMed, 194 from Elsevier ScienceDirect, and 2 from Google Scholar. A total of 401 studies were excluded before screening: duplicate records (n = 67) and records removed for other reasons (n = 334): genetic studies (n = 87), unrelated to stroke (n = 92), relationship between dementia (not AD) and stroke (n = 109), and pharmacology studies (n = 46) (Fig. 1).

Fig. 1.

PRISMA flow diagram illustrating the study selection process for the systematic review.

Forty-seven meta-analyses were assessed for eligibility. Of these, some were excluded because they (1) did not report an effect size (n = 4), (2) the primary studies did not measure the relationship between stroke and AD (n = 3), (3) the primary studies did not provide sample data (n = 8), (4) subjects were not diagnosed with AD at baseline (n = 9), (5) AD was not diagnosed using clinical criteria (DSM, ICD, NINCDS) (n = 15), and (6) AMSTAR meta-analysis score was lower than 9 (n = 5).

Table 2 summarizes the key features of the three selected meta-analyses. To estimate the effect sizes of each meta-analysis according to the type of stroke, only studies included in each meta-analysis that met the inclusion criteria were considered (Tables 3–5).

Table 2.

Descriptive characteristics of meta-analyses of longitudinal studies examining the relationship between AD and stroke

Study	Type of stroke	Total N	K	Region (N)	% F	Age M	Effect size			AMSTAR scores
							Effect size: LogOR	95% CI: LL∼UL	p value
Pinho et al. [73]	HS	Follow: HS and AD, n= 284; baseline: HS, N = 2.201	2	USA (2)	59.06	78	0.57	0.125∼1.008	0.012	9
	MI	Follow: MI and AD, n= 81; baseline: MI, N = 139	1	USA (1)	41	79	1.58	1.114∼2.036	0.000	​
Cao et al. [58]	IS	Follow: IS and AD, n= 261; baseline: IS, N = 680	3	EU (1), USA (2)	50.42	54.8	0.70	0.134∼1.267	0.015	10
	HS	Follow: HS and AD, n = 51; baseline: HS, N = 201	1	EU (1)	21.2	90.7	0.41	−0.693∼1.504	0.469	​
	MI	Follow: MI and AD, n= 239; baseline: MI, N = 491	2	USA (2)	35	*	0.85	−0.384∼2.074	0.178	​
Zhou et al. [62]	IS	Follow: IS and AD, n = 691; baseline: IS, N = 13.527	4	EU (2), North America (2)	55.75	73.23	0.89	−0.052∼1.840	0.064	10

Table 3.

Effect sizes related to AD and HS

Study name	Statics for each study
	sample	LogOR	Se	Ve	LLIC	ULIC	Z	p value	OR	LLIC	ULIC
Pinho et al. [73]
Epstein et al. [74]	Baseline: HS, N = 435; follow: HS and AD, n = 186	0.33	0.61	0.37	−0.873	1.525	0.53	0.594	1.39	0.418	4.597
Honig et al. [75]	Baseline: HS, N = 1,766; follow: HS and AD, n = 98	0.60	0.24	0.06	0.129	1.080	2.49	0.013	1.83	1.138	2.944
Cao et al. [58]
Brayne et al. [59]	Baseline: HS, N = 201; follow: HS and AD, n = 51	0.41	0.56	0.31	−0.693	1.504	0.72	0.469	1.50	0.500	4.500
Total random	​	0.54	0.21	0.04	0.134	0.954	2.60	0.009	1.72	1.144	2.596

AD, n, Alzheimer disease cases; HS, N, hemorrhagic stroke cases; LogOR, log odds ratio; Se, estimator bias; Ve, variance statistics; LLIC, lower limit confidence interval; ULIC, upper limit confidence interval; Z, standard punctuation; p, statistical significance; OR, odds ratio.

A p value ≤0.05 is statistically significant.

Table 5.

Effects of sex, age, and region in different types of stroke (IS, HS, and MI)

Moderator	Statics for each study
	variable	k	LogOR	Se	Ve	LLIC	ULIC	Z	p value	Qb
Stroke all types
Sex	Men	7	0.86	0.22	0.05	0.419	1.295	3.83	0.000	0.02, p = 0.900
	Women	8	0.81	0.28	0.08	0.262	1.362	2.90	0.004
Age	≤65	4	0.85	0.24	0.06	0.377	1.328	3.52	0.000	0.03, p = 0.853
	>65	11	0.79	0.25	0.06	0.307	1.270	3.21	0.001
Region	Europe	5	0.65	0.35	0.13	−0.041	1.342	1.84	0.065	0.43, p = 0.514
	North America	10	0.92	0.20	0.32	0.518	1.315	4.51	0.000
IS
Sex	Men	1	0.64	0.40	0.13	−0.140	1.424	1.61	0.108	0.10, p = 0.753
	Women	7	0.80	0.33	0.11	0.160	1.449	2.45	0.014
Age	≤65	2	0.98	0.53	0.28	−0.070	2.020	2.02	0.067	0.16, p = 0.692
	>65	6	0.71	0.39	0.15	−0.041	1.470	1.47	0.064
Region	Europe	3	0.65	0.58	0.33	−0.483	1.775	1.12	0.262	0.12, p = 0.725
	North America	5	0.88	0.35	0.12	0.194	1.574	2.51	0.012
HS
Sex	Men	3	0.54	0.21	0.04	0.134	0.954	2.60	0.009	0.00, p = 1.000
	Women	0	–	–	–	–	–	–	–
Age	≤65	1	0.60	0.24	0.06	0.129	1.080	2.49	0.013	0.24, p = 0.624
	>65	2	0.37	0.41	0.17	−0.441	1.179	0.89	0.371
Region	Europe	1	0.41	0.56	0.31	−0.693	1.504	0.72	0.469	0.07, p = 0.790
	North America	2	0.57	0.23	0.05	0.125	1.008	2.51	0.012
MI
Sex	Men	3	1.21	0.30	0.09	0.617	1.793	4.02	0.000	0.80, p = 0.371
	Women	1	0.84	0.28	0.08	0.288	1.387	2.99	0.003
Age	≤65	1	0.84	0.28	0.08	0.288	1.387	2.98	0.003	0.80, p = 0.371
	>65	3	1.21	0.30	0.09	0.617	1.793	4.01	0.000
Region	Europe	1	0.84	0.28	0.08	0.288	1.387	2.98	0.003	0.80, p = 0.371
	North America	3	1.21	0.30	0.09	0.617	1.793	4.01	0.000

Table 4.

Effect sizes related to AD and MI

Study name	Statics for each study
	Sample	LogOR	Se	Ve	LLIC	ULIC	Z	p	OR	LLIC	ULIC
Pinho et al. [73]
Suter et al. [76]	Baseline: MI, N = 139; follow: MI and AD, n = 81	1.58	0.24	0.06	1.114	2.036	6.69	0.000	4.83	3.045	7.662
Cao et al. [58]
Arvanitakis et al. [77]	Baseline: MI, N = 233; follow: MI and AD, n = 192	1.39	0.20	0.04	0.989	1.788	6.81	0.000	4.01	2.689	5.979
Sonnen et al. [78]	Baseline: MI, N = 258; follow: MI and AD, n = 47	0.12	0.52	0.27	−0.892	1.136	0.24	0.813	1.13	0.410	3.115
Total random	​	1.49	0.22	0.05	1.053	1.917	6.74	0.000	4.41	2.866	6.798

AD, n, Alzheimer disease cases; MI, N, microinfarcts cases; LogOR, log odds ratio; Se, estimator bias; Ve, variance statistics; LLIC, lower limit confidence interval; ULIC, upper limit confidence interval; Z, standard punctuation; p, statistical significance; OR, odds ratio.

A p value ≤0.05 is statistically significant.

A total of 14 effect sizes were extracted from three meta-analyses comprising 12 primary studies (see online suppl. Table 1S). According to the results of the meta-analysis, k = 8 effect sizes provided information about IS and the risk of AD (57.1%), k = 3 for HS (21.4%), and k = 3 for MI (21.4%).

For the pooled OR analysis, we analyzed the effect sizes of the primary studies (k = 12). The total effect size was LogOR = 0.82; se = 0.18; 95% CI = 0.470, 1.169; OR = 2.27; and 95% CI = 1.599, 3.218, and heterogeneity was high (Qb = 65.98; df = 14; p = 0.000; I² = 78.78). The results (Qb = 3.27; df = 2; p = 0.195) did not find any differences in effect sizes according to the type of stroke.

Ischemic Stroke and AD

Seven primary studies examined the association between IS and AD risk (k = 8 effect sizes; N = 14,207 participants with IS; n = 952 with AD and IS). The meta-analysis carried out by Cao et al. [58] (k = 4 effect sizes; N = 680 with IS; n = 261 with AD and IS) showed a significant association, whereas Zhou et al. [62] (k = 4 effect sizes; N = 13,527 with IS; n = 691 with AD and IS) found no significant association (Fig. 2).

Fig. 2.

Forest plot of the meta-analysis of risk rates AD in patients with IS.

However, the total random-effects analysis confirmed a statistically significant association between IS and AD (LogOR = 0.79; se = 0.29; 95% CI = 0.212, 1.362; Z = 2.68; p = 0.007; I² = 84.93). The effect sizes for IS are listed in Table 1.

Hemorrhagic Stroke and AD

Two meta-analyses reported a relationship between HS and AD. Pinho et al. [73] (k = 2 effect sizes; N = 2,201 participants with HS; n = 284 with HS and AD) found significant associations between HS and AD, whereas Cao et al. [58] (k = 1 effect size; N = 201 participants with HS; n = 51 with HS and AD) did not find significant associations between HS and AD (Fig. 3). Similar to Pinho et al. [73], this meta-analysis (k = 3 effect sizes; N = 2,402 with HS; n = 335 with HS and AD) found a significant overall association between HS and the risk of AD (LogOR = 0.54; se = 0.21; 95% CI = 0.134–0.954; Z = 2.60; p = 0.009; I² = 0.000).

Fig. 3.

Forest plot of the meta-analysis of risk rates of AD in patients with HS.

Microinfarcts and AD

Two meta-analyses examined the association between MI and risk of developing AD. In this vein, meta-analyses of studies conducted by Pinho et al. [73] (k = 1 effect size; N with N = 139 with MI; n = 81 with AD and MI) showed significant associations between MI and AD, and Cao et al. [58] (k = 3 effect sizes; N with MI = 491; n = 239 with AD and MI) did not find significant associations between MI and AD (Fig. 4). Similar to the study conducted by Pinho et al., this meta-analysis found a significant overall association between MI and AD (k = 3 effect sizes; N = 630 with MI, n = 320 with MI and AD) (LogOR = 1.49; se = 0.22; 95% CI = 1.053, 1.917; Z = 6.74; p = 0.000; I² = 69.59).

Fig. 4.

Forest plot of the meta-analysis of risk rates of AD in patients with all MI.

Moderating Variables Analysis

Moderator analyses were performed to explore possible parameters that may explain the differences between the effect sizes. These analyses were performed on categorical variables comparing studies by sex (0, men; 1, women), age (1: ≤65 years; 2: >65 years), and region (1: Europe; 2: North America).

The results showed no differences in the association between stroke and AD based on sex, age, or region. In addition, there were no differences in the association between each of the different types of stroke (IS, HS, and MI) and AD according to these variables (see Qb). In women, however, there was a significant association between IS and AD, but this association was not significant in men. Furthermore, there is an association between HS and AD in individuals younger than 65 years but not in those older than 65 years. In terms of region, we note that in North America, there is a significant association between all types of stroke and AD, IS and AD, and HS and AD, whereas in Europe, this association is not significant.

Discussion

We performed a systematic review of meta-analyses on the association between stroke and AD, as well as a meta-analysis of the primary longitudinal studies included in the three selected meta-analyses that met the inclusion criteria. We pooled the risk of developing AD in stroke patients. The results showed a significant association between IS, HS, MI, and the risk of AD. This review found no differences in the association between any type of stroke and AD based on age, sex, or region despite the heterogeneous effects observed between studies.

On the one hand, the results demonstrated a significant association between IS and the risk of AD. Both stroke and AD share several common pathophysiological mechanisms [79, 80], including excitotoxicity (which triggers damaging events in neuronal cells, such as calcium dysregulation and oxidative stress) [81], the presence of ApoE4 (a key genetic risk factor for LOAD, which is also associated with increased stroke risk) [82], chronic neuroinflammation (one of the most important pathophysiological conditions in AD and stroke) [83], and gut dysbiosis (which contributes to systemic and cerebral ischemia) [84]. In addition to these shared mechanisms, IS may contribute to AD development through specific pathological processes. These include blood-brain barrier disruption, cerebral hypoperfusion, accumulation of white matter lesions, ischemia-induced amyloid-beta aggregation, and tau phosphorylation [85]. Such mechanisms, specific to cerebrovascular events such as IS, may act as direct triggers or accelerators of neurodegeneration. Similarly, Chi et al. [86] and Vijayan et al. [87] determined that IS is one of the most significant vascular risk factors for AD.

Furthermore, we found an association between HS and risk of developing AD. This finding is consistent with those of other studies that have demonstrated significant associations between HS and AD [88, 89]. Several common pathophysiological mechanisms may explain this association. Cerebral microbleeds and white matter lesions caused by vascular damage in HS have been linked to AD-related neurodegeneration [90]. Additionally, chronic hypoperfusion, blood-brain barrier disruption, and neuroinflammation, all of which occur in HS, are implicated in AD pathogenesis. Moreover, cerebrovascular injury may exacerbate cognitive dysfunction when a concomitant degenerative process such as AD is already present, accelerating disease progression [62].

Finally, we found a significant association between MI and AD. It is true that MI degrades the integrity of microvascular and microstructural tissues, resulting in Aβ deposition and tau phosphorylation, which are associated with AD. In the same vein, other studies have demonstrated an association between MI and AD [91, 92]. MI may alter significant cognitive networks, which may explain a portion of the neurological dysfunction observed in AD [14].

Our findings support the association between stroke and increased risk of AD, reinforcing the role of cerebrovascular pathology in neurodegeneration. However, it is important to acknowledge that VaD plays a significant role in post-stroke cognitive decline [22]. Unlike AD, which is characterized by amyloid accumulation and tau pathology, VaD is primarily driven by ischemic events, MI, and chronic hypoperfusion, leading to white matter damage and executive dysfunction [23]. Despite these differences, growing evidence suggests a substantial overlap between AD and VaD, particularly in cases of mixed dementia, where both neurodegenerative and vascular mechanisms contribute to cognitive impairment [24].

On the other hand, the results showed no differences in the association between stroke and AD based on the moderator variables of sex, age, and region. Despite finding no moderating effects, we discovered a significant association between IS and AD in women but not in men. In contrast, the relationship between MI and AD was significant in both sexes. Regarding sex, other studies have found contradictory results. For instance, Kawas et al. [93] concluded that women with a history of stroke were more likely to develop AD than men. In fact, some evidence suggests that conditions related to pregnancy and menopause are female-specific risk factors for AD [94]. Similarly, research indicates that premenopausal women may experience a lower incidence of stroke than men of the same age, whereas postmenopausal women may experience an increase in stroke rate [94, 95]. Therefore, further analysis is necessary to understand the relationships between sex, stroke, and AD.

Understanding the interaction between stroke and AD may shed light on the relationship between the two pathologies, as well as between some of the moderators studied. There was an association between all types of stroke and AD, as well as MI and AD, among participants older than 65 years and younger than 65 years. However, some studies have yielded contradictory results. For example, Cook et al. [96] concluded that the incidence of stroke among patients with and without AD dementia was higher for individuals aged 50–69 years but decreased with increasing age. In addition, increasing age in people with stroke is associated with a decreased relative risk and an increased absolute risk of AD in individuals with a history of stroke. Some authors have explained that elderly patients may have a poor initial neurological status, which can lead to the development of stroke and AD [42]. Age has been shown to be associated with comorbid conditions such as hypertension, atrial fibrillation, cancer, and AD as well as a decreased likelihood of independence prior to stroke [91]. Consistent with our findings, Michalski et al. [97] found in an experimental study with mice that, following a stroke, myelin basic protein immunoreactivity was strongly affected throughout the ischemic nucleus, striatum, ischemic border zone, and lateral neocortex of the ischemic hemisphere, regardless of age and genetic background. Although different studies have found that the incidence of stroke increases with age [76, 98], Kokmen et al. [99]found that the risk of dementia doubled in the stroke cohort during the entire follow-up period, even after 25 years, regardless of age. Similar results were found in the Framingham study 10 years after stroke, after adjusting for age, sex, education, and individual stroke risk factors [100]. However, in this study, we also found a significant association between HS and AD in participants younger than 65 years but not in those older than 65 years. Seizures are a risk factor for AD and stroke in younger individuals, which may explain this finding [96].

In terms of region, we note that in North America, there is a significant association between all types of stroke and AD, IS and AD, and HS and AD, whereas in Europe, this association is not significant. In Europe, the association between MI and AD has been significant. One study concluded that the incidence of dementia in Europe has decreased by 13% per decade over the past 25 years as a result of lifestyle, education, and health interventions (i.e., blood pressure control and antithrombotic medication) that have been implemented to prevent vascular diseases [101]. In fact, according to a recent meta-analysis [102], the incidence of stroke in Europe is 191.9 per 100,000 person-years (95% CI: 156.4–227.3), with 195.7/100,000 person-years in men (95% CI: 142.4–249.0) and 188.1 per 100,000 person-years in women (95% CI: 138.6–237.7). This rate is lower than that reported in the UA, where the incidence was 373 per 100,000 person-years (95% CI: 351–396) from 1987 to 2011, with an incidence of 219 per 100,000 person-years in individuals under 65 and 529 per 100,000 person-years in those over 64 [102]. The main discrepancies between the US and European populations are attributed to lifestyle differences, such as the adoption of the Mediterranean diet, which may be associated with a lower risk of stroke [103]. In a study by Román et al. [104], a strict Mediterranean diet was associated with a very low prevalence of cardiovascular diseases such as stroke. Therefore, the Mediterranean diet is a promising tool for AD prevention. By not adopting this lifestyle, the US population may be more susceptible to cardiovascular disease, hypertension, diabetes, smoking, and obesity, all of which increase the risk of stroke [102].

The findings of our study must be interpreted in light of its strengths. First, we used a large and well-established primary-care database; 12 primaries and longitudinal studies were analyzed. Second, AMSTAR was used to fulfill the criteria for quantitative data synthesis to avoid publication bias. Third, we included an independent selection of studies by multiple authors. Fourth, the inclusion of only longitudinal studies in this meta-analysis provides reliable empirical evidence that stroke severity is a true risk factor for AD. However, our study had a few limitations. Significant study heterogeneity is likely attributable to the variability in stroke identification, high variability in follow-up periods, and other significant sociodemographic variables. For example, the estimation of a causal effect for a comparison between a disease group and a non-disease group may be biased because of the issue of self-selection associated with a patient’s specific prognostic factors. In large observational studies, cases and controls frequently exhibit numerous distinguishing characteristics.

Several questions remain unanswered in terms of future research, including the impact of white matter changes and the associated pathology of AD as well as the impact of pre-existing cognitive states in stroke and AD. Additionally, the possibility of AD without stroke should be investigated. The clinical significance of silent infarcts in all forms of dementia could be an additional area of study. In summary, future research should focus on the incidence and prevalence of AD after stroke, the predisposing etiologies of stroke, pre-stroke impairment, and imaging factors that determine the state of the brain following stroke.

Conclusion

Our study describes the risk of AD in patients with episodes of stroke (IS, HS, and MI) and suggests that the risk of AD may be higher in patients with stroke than in matched controls without a history of stroke. Moreover, moderator analysis supported the robustness of our results, showing that the association between stroke incidence and AD remains consistent across sexes, age-groups, and regions. Because stroke occurs more frequently than AD and is a modifiable risk factor, its prevention may also reduce the risk of AD. This highlights the clinical relevance of our findings: through the correction and management of stroke risk factors, not only AD but also VaD and mixed dementia, both of which represent the second most common causes of dementia, could be prevented or delayed. Implementing preventive measures, such as promoting a healthy diet, controlling cardiovascular diseases, managing hypertension, reducing smoking, treating diabetes, addressing obesity, metabolic syndrome, and depression, and preventing traumatic brain injury, could lower the risk of stroke, AD, and other dementia subtypes. Moreover, these results have important implications for health care planning. Understanding the link between stroke and dementia can help in resource allocation for patients with dementia and in developing targeted strategies to reduce dementia incidence through stroke prevention. In addition, this study allows clinicians to consider stroke occurrence as a prognostic factor in patients with AD.

Statement of Ethics

This study is a systematic review and meta-analysis based exclusively on previously published data and does not involve human participants, personal data, or animal subjects. Therefore, ethical approval was not required. All included studies were assessed to ensure they had obtained appropriate ethical approval and informed consent from participants as reported in their respective publications. This research was conducted in accordance with the principles outlined in the Declaration of Helsinki.

Conflict of Interest Statement

All authors certify that they have no affiliations with or involvement in any organization or entity with any financial or non-financial interest in the subject matter or materials discussed in this manuscript.

Funding Sources

This study received grants aimed at supporting recognized research groups from public universities of Castilla y León starting in 2024 (BU043G24).

Author Contributions

O.S. contributed to the conceptualization, methodology, formal analysis, investigation, data curation, and visualization and was responsible for writing the original draft. S.U., as corresponding author and project supervisor, was responsible for project administration, supervision, funding acquisition, conceptualization, validation, and review and editing of the manuscript. A.P. contributed to software management, formal analysis, data curation, methodology development, and resource provision and also participated in the review and editing of the manuscript.

Funding Statement

This study received grants aimed at supporting recognized research groups from public universities of Castilla y León starting in 2024 (BU043G24).

Data Availability Statement

All data generated or analyzed during this study are included in this published article and its online supplementary information files.