Extracerebral microvascular burden links to dementia risk and brain structural changes: Evidence from a large‐scale cohort study

Abstract

INTRODUCTION

The contribution of extracerebral microvascular disease (MVD) burden to dementia risk remains unclear.

METHODS

Among 484,896 UK Biobank participants (median follow‐up 8.5 years), baseline extracerebral MVD (retinopathy, nephropathy, peripheral neuropathy) and incident dementia were identified via hospital records. Aging‐related brain changes were evaluated using magnetic resonance imaging (MRI) ‐derived volumes and white matter integrity.

RESULTS

MVD was 2.6 times more prevalent among participants who developed dementia than dementia‐free individuals (8.0% vs 3.1%) and ranked second among vascular comorbidities after coronary artery disease. Each additional MVD burden was associated with a higher dementia risk, comparable to coronary artery disease (hazard ratio = 1.69 vs. 1.72). Earlier onset conferred stronger risk, and coexisting vascular diseases further amplified associations (p‐trend < 0.001). MVD burden was also associated with global brain atrophy and reduced white matter integrity.

DISCUSSION

Extracerebral MVD burden is associated with dementia risks and brain imaging changes, supporting integrated vascular risk stratification and management for dementia.

Highlights

• Extracerebral microvascular disease (MVD) burden is associated with incident dementia.

• Coexisting vascular diseases are associated with higher dementia risk in MVD.

• Higher MVD burden is associated with global brain atrophy and reduced white matter integrity.

1. BACKGROUND

Dementia is a burdensome syndrome and currently the seventh leading cause of death, affecting more than 55 million people worldwide. ¹ Given dementia's incurable and progressive nature, public health strategies now prioritize early identification, screening, and intervention for risk factors within a presymptomatic window extending more than two decades, aiming to prevent and control dementia. ² Brain function relies substantially on oxygen‐ and nutrient‐rich blood pumped by the heart and circulated through the vascular system, establishing a close biological link between vascular health and cognitive function. ³ The 14 modifiable risk factors for dementia reported by the Lancet Commission include cardiovascular risk factors, such as smoking, obesity, and diabetes, ⁴ which induce inflammatory responses, arteriosclerosis, and endocrine system dysfunctions, ⁵ ultimately contributing to dementia. Therefore, vascular factors have emerged as the primary modifiable, treatable, and preventable contributors to dementia. ⁶ This realization has sparked enduring discussion and highlighted promising strategies to manage and mitigate the burden of dementia.

Accumulating evidence highlights that vascular diseases contribute to the onset and development of dementia. ⁷ Recent studies mainly focus on cerebrovascular and cardiovascular disease, such as stroke, ⁸ coronary heart disease, ⁹ and cerebral small vessel disease. ¹⁰ However, in the context of the vascular system, extra‐cerebral and extra‐cardiac vascular diseases may also be associated with cognitive impairment and dementia. Microvascular disease (MVD), represented by microvascular dysfunction in various vascular beds, often develops insidiously and has received less attention compared to macrovascular diseases. ¹¹ Previous studies have separately explored the association between regional microvascular dysfunction such as retinal microvasculature and poor cognitive performance. ¹² However, microvascular diseases often share risk factors and pathological features, rarely occur independently, and tend to interact and progress systematically. ¹¹ The relationship between comprehensive burdens of MVD and dementia remains poorly understood due to inconsistent clinical markers for MVD, small sample sizes, and cross‐sectional study designs. ¹² , ¹³ , ¹⁴ , ¹⁵

The UK Biobank (UKB) employed a prospective design ensuring temporal exposure‐outcome associations. Linked health records enable comprehensively identification of microvascular diseases and dementia outcomes, along with brain imaging scan facilitating the observation of neuroimaging phenotypes. ¹⁶ Leveraging this large‐scale research resource, our study aims to evaluate the association between regional and global burden of MVD and the risk of incident dementia. This research contributes a comprehensive perspective on extracerebral MVD burden for risk management of dementia.

2. MATERIALS AND METHODS

2.1. Study design and participants

This study was based on the UKB (application number of 92718) ¹⁶ , a population‐based prospective cohort of over 500,000 participants recruited in England, Scotland, and Wales between 2006 and 2010, with extensive baseline assessments of sociodemographics, lifestyle, medical history, physical measures, and cognition. Since 2014, a subset (∼70,000) underwent brain magnetic resonance imaging (MRI) and comprehensive cognitive tests. ¹⁷ Linked health records (hospital admissions, cancer registries, and mortality) were regularly updated. For the primary analyses examining the associations between vascular diseases and incident dementia, April 30, 2014, corresponding to the initiation of the imaging visit, was used as the index date to ensure an adequate sample size in the exposed group. Participants were followed from the index date until the first occurrence of incident dementia, death, loss to follow‐up, or October 31, 2022, whichever came first. We excluded participants who withdrew consent (N = 33), had prevalent dementia (N = 987), died before baseline (N = 9354), or had missing ethnicity (N = 7,119). Finally, the primary analysis set included 484,896 participants, among whom 8,408 were incident all‐cause dementia (ACD) cases. Analyses of the secondary outcomes were restricted to participants who attended the imaging visit and had relevant data available, including 65,361 participants for cognitive performance and 47,114 participants for MRI‐derived phenotypes. A flowchart detailing participant inclusion is shown in Figure 1A. Ethical approval for the UKB was obtained from the North West Multi‐centre Research Ethics Committee, and all participants provided electronically signed informed consent.

FIGURE 1.

Study population inclusion flowchart and baseline prevalence of vascular diseases. (A) Study population inclusion flowchart. (B) Baseline prevalence of vascular diseases among all populations, dementia‐free, and ACD. differences in prevalence between ACD and dementia‐free were calculated using the chi‐square test, all p values < 0.001. ACD, all‐cause dementia; CAD, coronary artery disease; MRI, magnetic resonance imaging; MVD, microvascular disease; PAD, peripheral artery disease

RESEARCH IN CONTEXT

1. Systematic review: We searched PubMed and found that previous studies have mostly discussed the impact of cardiovascular and cerebrovascular diseases on dementia. The relationship between the comprehensive burdens of microvascular diseases (MVDs) and dementia remains poorly understood due to inconsistent clinical markers for MVD, small sample sizes, and cross‐sectional study designs.

2. Interpretation: We found that MVD was associated with higher risks of incident dementia and its major subtypes, especially when MVD diagnosed at a younger age and amplified by other cardiovascular and cerebrovascular diseases. Global brain atrophy and reduced fractional anisotropy of white matter integrity were significant brain structural changes associated with MVD burden.

3. Future directions: MVD burden may support dementia risk identification and stratification. Future studies should use detailed clinical phenotyping to characterize organ‐specific MVD, validate these findings in diverse populations, investigate underlying mechanisms, and evaluate whether vascular disease management is associated with cognitive outcomes.

2.2. Dementia ascertainment, cognitive assessment, and brain MRI scan

The incident cases of dementia were extracted and ascertained from summary diagnoses of the hospital inpatient admissions according to the earliest record of International Classification of Diseases, Tenth Revision (ICD‐10) codes. The specific codes for ACD, Alzheimer's disease (AD), and vascular dementia (VaD) are listed in Table S1. The end of follow‐up was determined by the date of dementia recording (only for dementia cases), date of death, loss to follow‐up, or the end of the study (October 31, 2022), whichever occurred first.

Cognitive assessment was described in Supplementary Methods. The UKB imaging study is one of the largest imaging projects globally, which scans the brains and collects imaging‐derived phenotypes (IDPs). This study prioritized the assessment of global brain atrophy and total volume of white matter hyperintensities. To identify sensitive regions, volumes of cortical and subcortical regions from T1 data were additionally studied. Fractional anisotropy (FA) and mean diffusivity (MD) reflect the white matter integrity, which were extracted as tract‐averaged IDPs and calculated as global average levels from skeleton dMRI data. ¹⁷ Specifically, FA indicates the directionality of water diffusion, while MD describes the magnitude of water diffusion.

2.3. Diagnoses of baseline vascular diseases

Vascular diseases encompass a spectrum of conditions, classified into MVD, cerebrovascular disease, coronary artery disease (CAD), and peripheral artery disease (PAD) in this study. Because cerebrovascular and cardiovascular diseases have been extensively studied, we included them without further subtype classification solely to compare their effect estimates with those of MVD. As the primary exposure in this study, MVD comprised retinopathy, nephropathy, and peripheral neuropathy, based on the conventional classification of diabetic microvascular complications. ¹⁸ The ICD‐10 codes for each vascular disease are extracted from the hospital inpatient admissions and listed in Table S1. Cases of vascular disease diagnosed before the start of this study were identified as baseline exposures.

2.4. Data analysis

Detailed information on covariates is provided in Supplementary Methods and Table S2, including age, sex, ethnicity, Townsend Deprivation Index (TDI), education, apolipoprotein E (ApoE), family history of dementia and heart diseases or stroke, baseline diabetes, and individual cardiovascular health (CVH) score. Continuous variables are presented as median (P₂₅, P₇₅) and categorical variables as number (%), with group differences assessed using Mann–Whitney U‐test and the chi‐squared. Vascular diseases were defined as binary exposures. A MVD burden score (0–3) was constructed based on retinopathy, nephropathy, and peripheral neuropathy. Cox proportional hazards models were used to estimate associations between vascular diseases and incident dementia, with MVD further stratified by age at diagnosis (tertiles). Additive effects of MVD, cerebrovascular disease, CAD, and PAD were evaluated using participants without vascular diseases as the reference.

Sensitivity analyses were performed by excluding patients with mixed dementia, those diagnosed with dementia within 2 years after the index date, and those with cancer, nervous system diseases, or cerebrovascular diseases, or restricting the population to White participants. Furthermore, subgroup analyses were conducted using Cox proportional hazards models, stratifying by sex, age group (< median and ≥ median), TDI group (< median and ≥ median), family history of dementia, family history of heart diseases and stroke (no and yes), ApoE ε4 status (non‐carriers and carriers), baseline diabetes (no and yes), and the CVH status (< median and ≥ median). Two adjusted models were used: model 1 adjusted for age, sex, and ethnicity; model 2 additionally adjusted for TDI, family history of dementia, ApoE ε4, and CVH score status. Additional adjustments for education and individual CVH components were performed in sensitivity analyses.

Associations of MVD with cognitive performance were assessed using Poisson and logistic regression. IDPs were normalized, ¹⁹ outliers (> 4 SD) excluded and standardized before analysis. The cross‐sectional associations between baseline MVD burden and brain IDPs were analyzed by multiple linear regressions. In the sensitivity analysis of cognitive performance and brain IDPs, participants with mental disorders, nervous system diseases or cerebrovascular diseases (ICD‐10 Chapter F, G, and I60‐69) were excluded to avoid potential influences on brain structure or function. Moreover, the MRI site, head motion, head position and table/coil position were also adjusted in the sensitivity analysis for IDPs. False discovery rate (FDR) ‐adjusted p‐values < 0.05 were considered statistically significant to account for multiple testing, whereas FDR–p‐values from 0.05 to 0.20 were considered exploratory associations warranting further investigation. All statistical analyses were conducted using R (version 4.2.2).

3. RESULTS

3.1. Participant characteristics

During a median follow‐up period of 8.5 years involving 484,896 participants, a total of 8,408 incident cases of ACD were recorded. No statistically significant differences in participant characteristics were observed before and after imputation of missing data (all p > 0.05, Table S3). The median baseline age of the total study population was 63.5 years, with the majority being the White population (89.3%) (Table S4). Compared to participants without dementia at follow‐up, fewer incident ACD cases had received a college‐level education (32.6% vs. 20.3%, p < 0.001). Moreover, ACD cases had notably higher percentage of family history related to dementia, heart diseases, and stroke, ApoE ε4 carriers, and unfavorable combined CVH status (all p < 0.001).

As shown in Figure 1B, the baseline prevalence of MVD was 8.0% in ACD patients, ranking second only to CAD (17.9%), followed by cerebrovascular disease (6.6%), and PAD (2.0%). CAD and MVD were more than 2.5 times as prevalent among individuals with incident ACD as among those without dementia (all p < 0.001). Within the spectrum of MVD, retinopathy was the most prevalent in incident ACD cases at 5.0%, followed by nephropathy (2.9%), and peripheral neuropathy (1.1%).

3.2. Associations of vascular diseases with incident dementia

All baseline vascular diseases exhibited a significant association with a higher risk of incident ACD (Figure 2). Each increment in MVD burden, represented by retinopathy, nephropathy, and peripheral neuropathy, was associated with a 67% higher risk of ACD (hazard ratio [HR] and 95% confidence interval [CI]: 1.53, 1.82 in model 2). Among the subtypes of MVD, peripheral neuropathy demonstrated the highest risk with ACD after fully adjustment in model 2 (HR 95% CI: 3.31 [2.54, 4.32])), higher than the risk of cerebrovascular disease (HR 95% CI: 2.65 [2.36, 2.98]) and followed by nephropathy (HR 95% CI: 2.18 [1.85, 2.56]). For dementia subtypes, the association strengths between vascular diseases and VaD consistently greater than those between vascular diseases and AD. MVD burden was positively associated with risk of AD after fully adjustment in model 2, with the estimate similar to that of cerebrovascular disease, CAD and PAD (the range of HRs: 1.40‐1.61).

FIGURE 2.

Associations between baseline vascular diseases and incident dementia. All vascular diseases were treated as binary variables, except for the additionally derived MVD burden score, which quantified the cumulative burden of retinopathy, nephropathy, and peripheral neuropathy and ranged from 0 to 3. Model 1 adjusted for age, sex, and ethnicity; Model 2 adjusted for covariates in model 1 plus TDI, family history of dementia, family history of heart diseases or stroke, ApoE ε4 and the group of cardiovascular health score. All FDR‐p values < 0.01. ApoE, apolipoprotein E; FDR, false discovery rate; MVD, microvascular disease; TDI, Townsend Deprivation Index

Generally, the association between MVD and dementia remained significant in all sensitivity analyses (Tables S5, S6), except that baseline PAD showed no significant association with incident AD after excluding patients with mixed dementia or nervous system diseases (Table S6). Subgroup analyses showed that the associations of MVD with AD and VaD limited to the subgroup with unfavorable CVH score, and the associations with ACD consistently significant among all subgroups (Table S7). Importantly, baseline MVD was significantly associated with an increased risk of ACD, regardless of baseline diabetes status (HR was 2.06 and 1.36 among no diabetes and diabetes individuals, respectively)

3.3. Long‐term and additive associations between MVDs and incident dementia

The associations between baseline vascular diseases and incident dementia were also stratified based on exposure age tertiles (Figure 3A). Compared to individuals without MVD at baseline, the HRs for incident ACD in MVD cases gradually decreased with increasing age at MVD diagnosis (HR: 2.03, 1.59 and 1.55 for tertiles of MVD diagnosis age, respectively, p for trend < 0.05). Participants in the youngest tertile of age at MVD diagnosis had the highest risks of dementia, AD, and VaD compared with those without MVD, with significant trends across age tertiles (p for trend < 0.05). The median age at MVD diagnosis was 63 years old, specifically 62 years old for retinopathy and peripheral neuropathy, and 65 years old for nephropathy (Figure 3B).

FIGURE 3.

Long‐term and additive associations between MVD burden and incident dementia. (A) Stratification analyses on age of MVD diagnosis. HR and 95% CI of Cox proportional hazards model was showed after adjustment for age, sex, ethnicity, TDI, family history of dementia, family history of heart diseases or stroke, ApoE ε4 and the group of cardiovascular health score. (B) Distribution of age at MVD diagnosis. Dashed lines and numbers on the top of each figure represent the 33^rd, 50^th, 66^th percentiles of age at diagnosis, respectively. (C) Additive effects of baseline vascular diseases on the risk of incident dementia. HR of Cox proportional hazards model was showed after adjustment for age, sex, ethnicity, TDI, family history of dementia, family history of heart diseases or stroke, ApoE ε4 and the group of cardiovascular health score. All p for trend < 0.05. *** represents p < 0.001, ** represents p < 0.01, and * represents p < 0.05. ApoE, apolipoprotein E; CI, confidence interval; HR, hazard ratio; MVD, microvascular disease; TDI, Townsend Deprivation Index

The additive associations between baseline vascular diseases and the risk of incident dementia are presented in Figure 3C. Compared to healthy controls, the pure MVD group at baseline was associated with higher incident ACD, AD and VaD (HR: 1.46, 1.43, and 1.36 respectively, both p < 0.05). Additionally, the HRs for incident ACD, AD, and VaD increased with the number of coexisting classes of vascular diseases at baseline (all p for trend < 0.001). Notably, pure MVD and CAD without any of other coexisting vascular disease classes was associated with significant higher risk of incident AD (HR and 95%CI: 1.43 [1.18, 1.72] and 1.37 [1.20, 1.56]) when compared to healthy participants.

3.4. Associations of MVD burden with neuroimaging phenotypes

We further investigated the associations between MVD burden and dementia phenotypes represented by macro cognitive performance (Table S8) and micro brain IDPs (Table S9 and Figure 4). Baseline MVD burden was significantly associated with a longer completion time for trail making test A, indicating poorer executive function (exp[estimate] and 95% CI: 1.04 [1.02, 1.07] in model 2 of main analysis, p < 0.05). However, no significant associations were found between MVD burden and cognitive domains in the fully adjusted sensitivity analysis.

FIGURE 4.

Associations between MVD burden and brain imaging‐derived phenotypes. (A) Global imaging‐derived phenotypes. (B) Regional and track‐averaged imaging‐derived phenotypes. (C) Significant sites of reduced volume associated with MVD. Primary analysis adjusted for age, sex, ethnicity, TDI, family history of dementia, family history of heart diseases or stroke, ApoE ε4 and the group of cardiovascular health score. Sensitivity analysis excluded patients with mental, behavioral, nervous system disorders or cerebrovascular diseases at baseline, and adjusted for covariates in primary analysis plus the MRI site, head motion, head position and table/coil position. ApoE, apolipoprotein E; MRI, magnetic resonance imaging; MVD, microvascular disease; TDI, Townsend Deprivation Index

Regarding brain IDPs derived from MRI data, higher MVD burden was associated with global brain atrophy, particularly lower grey matter volume, in both the primary and sensitivity analyses (global brain atrophy: estimates [95% CI], ‐0.08 [‐0.15, ‐0.02] and ‐0.11 [‐0.19, ‐0.03], respectively; grey matter volume: ‐0.09 [‐0.16, ‐0.03] and ‐0.10 [‐0.17, ‐0.03], respectively; all FDR‐p < 0.05; Figure 4A). Among regional volumetric IDPs, lower left lateral occipital cortex volume showed the most consistent association with MVD burden (estimates = ‐0.18 and ‐0.20 in the primary and sensitivity analyses, respectively; both FDR‐p < 0.01; Figures 4B, C). Lower bilateral pallidal volume was also observed in the primary analysis (estimate = ‐0.15 for both hemispheres, both FDR‐p = 0.006) and was directionally consistent in the sensitivity analyses at the exploratory threshold (FDR‐p < 0.20). Exploratory associations of MVD with lower volumes of the left putamen, hippocampus, accumbens area, and right amygdala did not persist in sensitivity analyses. For white matter microstructural IDPs (Figure 4B), higher MVD burden was associated with lower global FA and higher global MD in the primary analysis (estimates = ‐0.13 and 0.11, respectively; both FDR‐p < 0.05), but these associations were not statistically significant in the sensitivity analyses. Lower FA in the posterior thalamic radiation and fornix cres/stria terminalis was also observed in the primary analysis and was directionally consistent at the exploratory threshold in the sensitivity analyses (FDR‐p < 0.20).

4. DISCUSSION

In this prospective study involving 484,896 participants from the UKB, tracked over a median 8.5‐year period, we found that the MVD burden was significantly higher in individuals with dementia than those without. Baseline MVD exhibited associations with increased risks of ACD, AD, and VaD, with an association akin to that of CAD. Notably, midlife MVD had an enduring and cumulative association with the onset of dementia in later life. Neuroimaging assessments uncovered potential changes in brain structure, such as grey matter atrophy and lower white matter integrity in posterior thamamic radiation, linking MVD with dementia progression.

Our prospective findings not only validate the well‐known impact of cerebrovascular and cardiovascular diseases ⁸ , ⁹ on dementia, but also emphasize MVD burden as an underestimated microvascular markers associated with risks of ACD and its common subtypes, AD and VaD. Vascular injury contributes to cognitive impairment and dementia, but its intertwined relationship with dementia subtypes shows heterogeneity and remains uncertain. ²⁰ As the second most prevalent form of dementia after AD, VaD is widely acknowledged for its pathological basis in cerebral vascular injury, ²¹ which is in line with our results that all vascular diseases exhibit a stronger association with VaD compared to AD. However, AD is a multifactorial disorder involving several different pathogenic mechanisms, ² the precise role of vascular diseases in the AD continuum, whether as a causative factor, a precursor, or a comorbidity, remains a subject of debate. ⁷ A case–control study and in vitro experiment indicated that vascular injury‐induced blood–brain barrier disruption is an early procedure for cognitive dysfunction. ²²

In this population‐based study, a higher baseline burden of MVD was prospectively associated with an increased risk of incident AD. Most previous studies of MVD have been limited to individual organ‐specific microvascular diseases. For instance, a decade‐long follow‐up of 584 participants revealed that increased retinal arteriolar tortuosity and venular calibers was associated with dementia risk. ²³ Besides, many prospective studies have highlighted significant associations between renal diseases and dementia. ²⁴ , ²⁵ Among the three MVD components examined, peripheral neuropathy showed the strongest association with incident ACD, possibly reflecting both microvascular injury and neural damage. ²⁶ Moreover, early peripheral neuropathy is often asymptomatic or underrecognized, clinically recorded cases may represent more symptomatic or advanced disease and may also indicate severe diabetes. ²⁷ Therefore, residual confounding by cumulative diabetes burden cannot be excluded despite adjustment for multiple diabetes‐related factors. Differences in diagnostic sensitivity and clinical severity across retinopathy, nephropathy, and peripheral neuropathy may partly explain the variation in their associations. Although these conditions involve distinct organ‐specific pathophysiology, some overlapping vascular and inflammatory processes have been proposed. ¹¹ Because these diseases are commonly recognized as clinical manifestations of diabetic microvascular complications, we constructed a diagnosis‐based composite score to capture the cumulative clinical burden of these conditions and report their combined association with cognitive outcomes. Comprehensive research efforts are warranted to enhance our systematic comprehension of vascular diseases, particularly the underestimated MVD.

Our study showed that younger age at MVD onset was associated with a higher risk of dementia. These findings align with a prior cross‐sectional population‐based study (N = 2,452, mean age = 79.5 years), which linked microvascular and macrovascular abnormalities to impaired cognitive function. ²⁸ Importantly, this association remained significant, regardless of traditional cardiovascular risk factors and prior CVD history. ¹⁵ Our study, based on a larger and younger cohort (N = 484,896, mean age = 63.5 years), contributes essential longitudinal evidence and complements existing research gaps. The critical long‐term associations of MVD emphasizes the necessity of early detection and management. Furthermore, we also found the coexistence of other vascular diseases was associated with a higher risk of incident dementia, indicating the additive effect pattern. A case–control study found that heart failure predicts coronary and cerebral microvascular dysfunction, which are both associated with cognitive dysfunction. ¹⁴ Although the systematic research on both macrovascular and microvascular diseases was insufficient, these observations highlight the intricate and interconnected nature of vascular conditions, underscore the need for additional evidence from broader natural populations and more in‐depth investigations into causal mechanisms to enhance our current understanding.

The imaging analyses in this study underscore the association between MVD and alterations in brain structure, extending previous research focused on individual MVD. Higher MVD burden was associated with global brain atrophy, particularly lower grey matter volume and left lateral occipital cortex volume, suggesting that cumulative microvascular injury may contribute to brain vulnerability. These findings align with previous studies showing that vascular burden is associated with adverse brain imaging phenotypes. Epidemiological studies have linked greater vascular burden to lower brain volumes and impaired white matter microstructure, ²⁹ while kidney microvascular dysfunction ³⁰ and retinal microvascular abnormalities ³¹ have been associated with brain atrophy and altered white matter integrity, respectively. Associations involving additional subcortical regions and white matter microstructural measures were not consistently retained in our sensitivity analyses and should therefore be interpreted as exploratory. Several mechanisms may underlie these associations. Microvascular endothelial dysfunction can impair neurovascular unit function, disrupt cerebral blood flow regulation, and compromise blood–brain barrier integrity. ³ Persistent cerebral hypoperfusion and impaired oxygen metabolism may promote neuronal and white matter injury, thereby contributing to progressive structural brain damage. ³² , ³³ Collectively, these findings highlight MVD burden as a potential contributor to aging‐related brain changes, and warrant further longitudinal investigation to clarify causal mechanisms.

To our knowledge, this study represents the first attempt to assess the associations between MVD burden and incident dementia within a large‐scale community‐based cohort. From a holistic perspective of vascular diseases, we establish a prospective relationship between MVD and incident dementia. It also facilitates comparisons of the varying associations of different vascular diseases on dementia among different subgroups and under independent/combined exposure patterns, adding in identifying gaps and priorities in management strategies of vascular risks. By adopting a comprehensive index to represent the MVD burden, this research contributes to a more nuanced understanding of the vascular mechanisms implicated in dementia. Furthermore, the inclusion of precise brain imaging data provides crucial insights into the neuroimaging changes involved within the continuum of dementia.

This study also has the following limitations. The MVD definition was derived from classic diabetic microvascular complications. As the three component conditions are identified through clinical diagnoses and differ in diagnostic sensitivity, clinical severity and the relationship with diabetes, subclinical or early‐stage organ‐specific MVD may have been undercaptured, potentially resulting in non‐differential exposure misclassification. Moreover, the relatively modest cognitive associations and the attenuation of some findings after full adjustment and sensitivity analyses warrant cautious interpretation. These findings may partly reflect the limited sensitivity of brief, unsupervised touchscreen‐based cognitive tests in the UKB. ³⁴ Detailed clinical examination data are needed to better characterize preclinical changes and cognitive associations of organ‐specific MVD. Additionally, the prospective observational design precludes causal inference, and residual confounding cannot be fully excluded, including the cumulative glycemic injury, inflammatory status, diabetes duration or severity, multimorbidity, healthcare utilization, medication use, and lifestyle factors not fully captured at baseline. Furthermore, the relatively healthier and less socioeconomically deprived UK Biobank population, together with potential selection and survival bias in the imaging subcohort, may limit the generalizability of our findings. Further validation in diverse populations, including high‐risk groups, is needed to determine the broader applicability of these findings.

Based on this large‐scale cohort and from a comprehensive perspective of vascular diseases, this study reveals that MVD burden shows a substantial but often underestimated association with dementia risk, along with the long‐term and additive associations of other vascular diseases. Notably, brain imaging analyses highlight brain atrophy and reduced white matter integrity as critical vascular‐related neuroimaging pathologies, meriting further validation through prospective evidence and specific mechanism investigation. This prospective research enhances our understanding of the associations between vascular disease and dementia risk, supporting risk stratification and management.

AUTHOR CONTRIBUTIONS

Jialin Li: Conceptualization, Formal analysis, Writing—Original Draft, Visualization; Yingzhe Wang: Conceptualization, Methodology, Writing—Original Draft; Chunyu Yin: Conceptualization, Methodology, Writing—Original Draft; Mei Cui: Writing—Review & Editing, Funding acquisition; Jincheng Li: Methodology; Chen Suo: Resources, Writing—Review & Editing; Kelin Xu: Data Curation, Funding acquisition; Zhenqiu Liu: Data Curation, Writing—Review & Editing; Li Jin: Project administration; Xingdong Chen: Funding acquisition; Yanfeng Jiang: Conceptualization, Resources, Writing—Review & Editing, Supervision, Funding acquisition.

CONFLICT OF INTEREST STATEMENT

The authors declare no competing interests. Author disclosures are available in the Supporting Information.

ROLE OF THE FUNDER/SPONSOR

The funders had no role in the study design, data collection, analysis, interpretation, or manuscript preparation.

CONSENT STATEMENT

Ethical approval for the UKB was obtained from the North West Multi‐centre Research Ethics Committee, and all participants provided electronically signed informed consent.

Supporting information

Supporting Information

Supporting Information

Supporting Information

DATA AVAILABILITY STATEMENT

The datasets generated and analyzed during the current study are available on the UKB website upon request (www.ukbiobank.ac.uk/).