Imaging-validated correlates and implications of the pathophysiologic mechanisms of ageing-related cerebral large artery and small vessel diseases: a systematic review and meta-analysis

Abstract

Background

Cerebral large artery and small vessel diseases are considered substrates of neurological disorders. We explored how the mechanisms of neurovascular uncoupling, dysfunctional blood–brain-barrier (BBB), compromised glymphatic pathway, and impaired cerebrovascular reactivity (CVR) and autoregulation, identified through diverse neuroimaging techniques, impact cerebral large artery and small vessel diseases.

Methods

Studies (1990–2024) that reported on neuroradiological findings on ageing-related cerebral large artery and small vessel diseases were reviewed. Fifty-two studies involving 23,693 participants explored the disease mechanisms, 9 studies (sample size = 3,729) of which compared metrics of cerebrovascular functions (CF) between participants with cerebral large artery and small vessel diseases (target group) and controls with no vascular disease. Measures of CF included CVR, cerebral blood flow (CBF), blood pressure and arterial stiffness.

Results

The findings from 9 studies (sample size = 3,729, mean age = 60.2 ± 11.5 years), revealed negative effect sizes of CVR [SMD = − 1.86 (95% CI − 2.80, − 0.92)] and CBF [SMD = − 2.26 (95% CI − 4.16, − 0.35)], respectively indicating a reduction in cerebrovascular functions in the target group compared to their controls. Conversely, there were significant increases in the measures of blood pressure [SMD = 0.32 (95% CI 0.18, 0.46)] and arterial stiffness [SMD = 0.87 (95% CI 0.77, 0.98)], which signified poor cerebrovascular functions in the target group. In the combined model the overall average effect size was negative [SMD = − 0.81 (95% CI − 1.53 to − 0.08), p < 0.001]. Comparatively, this suggests that the negative impacts of CVR and CBF reductions significantly outweighed the effects of blood pressure and arterial stiffness, thereby predominantly shaping the overall model. Against their controls, trends of reduction in CF were observed exclusively among participants with cerebral large artery disease (SMD = − 2.09 [95% CI: − 3.57, − 0.62]), as well as those with small vessel diseases (SMD = − 0.85 [95% CI − 1.34, − 0.36]). We further delineated the underlying mechanisms and discussed their interconnectedness with cognitive impairments.

Conclusion

In a vicious cycle, dysfunctional mechanisms in the glymphatic system, neurovascular unit, BBB, autoregulation, and reactivity play distinct roles that contribute to reduced CF and cognitive risk among individuals with cerebral large artery and/or small vessel diseases. Reduction in CVR and CBF points to reductions in CF, which is associated with increased risk of cognitive impairment among ageing populations ≥ 60 years.

Introduction

The brain, a highly complex organ with nearly 100 billion neurones, is the most metabolically active organ in the body, consuming about 20% of the body's oxygen despite constituting only 2% of body weight [1, 2]. This high demand for essential oxygen and nutrients necessitates stable neurovascular coupling, ensuring an uninterrupted vascular system for adequate brain perfusion in addition to an effective glymphatic system for removing the metabolic waste [3, 4]. As the brain ages, cerebrovascular dysfunction becomes increasingly prominent with evidence of neurovascular uncoupling [5]. These ageing-related dysfunctional changes make the brain vasculature even more vulnerable to plaque accumulation, stenosis, stiffening, glymphatic disruption, and other endovascular issues, thereby compromising cerebral perfusion and waste removal [6,7,8,9]. Evidence suggests that exploring the mechanisms of neurovascular uncoupling, a dysfunctional blood–brain barrier (BBB), a compromised glymphatic pathway, and impaired cerebrovascular reactivity (CVR) and autoregulation in cerebrovascular dysfunctions may provide insights into ageing-related neurovascular alterations that contribute to cognitive decline [2, 10,11,12,13].

The clinical implications of these alterations are shown to range from silent to symptomatic manifestations, contingent on the severity, extent, size, morphology, and location of the underlying pathology [14]. The changes in the cerebral microvasculature present as cerebral small vessel diseases of presumed vascular origin, presenting typical phenotypes such as white matter hyperintensities (WMH), enlarged perivascular spaces (PVS), microbleeds, and lacune infarcts, as detected on neuroimaging [15,16,17]. Cerebral large artery diseases, on the other hand, present as intracranial atherosclerosis (ICAS) and intracranial arterial calcification (IAC) [18, 19]. Ageing-related cerebral large artery and small vessel diseases are considered strong predictors of neurological disorders, including stroke and cognitive impairment; however, their interconnectedness is complex [20,21,22]. Previous research has demonstrated a significant link between atherosclerotic diseases of the cerebral large arteries and the overall burden of cerebral small vessel diseases [23, 24]. However, contrasting evidence suggests that any coexisting association might be weak, arising primarily from shared risk factors common to both vascular conditions [25,26,27]. This dual perspective has spurred additional research, particularly in neuroimaging, to delve into the complex literature and offer a comprehensive understanding of these diverse interpretations. Notably, the advancements in neuroimaging hint that a deeper exploration of underlying mechanisms of cerebrovascular dysfunctions could uncover nuances beyond those explained by traditional risk factors [28,29,30,31,32].

Despite the progress, challenges remained in fully understanding how a dysfunction in one disease mechanism influences the others and whether such fluctuations impact the burdens of cerebral large artery diseases and small vessel diseases or their interconnectedness with cognitive impairment [28,29,30,31]. The varied applications of neuroimaging in various investigations with differences in interpretation and the complexity of integrating multimodal data have presented fragmented pieces of evidence [17, 33, 34]. Challenges stem from the limited generalisability of findings, as some studies focus exclusively on stroke patients, who may not represent the broader population affected by cerebral vascular diseases [35,36,37]. Furthermore, most studies often target specific aspects of cerebral vasculature or rely on particular imaging techniques, potentially overlooking the complex interplay of multiple disease mechanisms [24, 38,39,40,41,42]. Compounding these limitations has constrained our understanding of the comprehensive pathophysiology underlying both cerebral large artery and small vessel diseases [43, 44].

Therefore, this review aimed to explore how the mechanisms of neurovascular uncoupling, a dysfunctional BBB, a compromised glymphatic pathway, and impaired CVR and autoregulation, identified through diverse neuroimaging techniques, impact cerebral large artery and small vessel diseases. We further elucidated how these mechanisms facilitate the interconnectedness between the various cerebral large artery and small vessel diseases and cognitive impairment. The rationale for this endeavour was not merely to aggregate existing data but to apply critical synthesis techniques to explore patterns, relationships, and insights that are only visible through the lens of integrated analysis.

Methods

The review protocol was registered on PROSPERO with ID: CRD42024531238. This review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) [45].

Eligibility criteria

Eligibility was set in accordance with the PICO framework, whose elements include the population, intervention, comparison, and outcomes considered in each study.

P (population): studies that analysed stroke-free participants of any gender or ethnicity, above 18 years of age, with either ageing-related cerebral large or small vessel diseases.

I (intervention): studies employed single or combined neuroimaging examinations such as magnetic resonance imaging (MRI) and MR angiography (MRA) for the evaluation of atherosclerotic large artery diseases and small vessel diseases as well as disease mechanisms; computed tomography (CT) and CT angiography (CTA) to assess vascular integrity and detect calcifications and haemorrhages; transcranial Doppler (TCD) and carotid duplex ultrasound to measure haemodynamic mechanism and lumenography. These modalities were selected for their complementary strengths in assessing cerebral vascular conditions, and this was necessary to capture the multifaceted nature of these conditions, as relying on a single modality may limit our understanding.

C (comparator): studies reported brain patterns, cerebrovascular functions, or haemodynamic mechanisms observed in healthy controls or subjects with no cerebral large artery and small vessel diseases.

O (outcome): neuroimaging correlates of the ageing-related neuroparenchyma changes, neurovascular changes, cerebrovascular haemodynamics, as well as cognitive assessments. Cerebral small vessel diseases encompassed key phenotypes such as WMH, enlarged PVS, microbleeds, and lacune infarcts, whereas cerebral large artery diseases referred to ICAS and IAC.

Inclusion

This study considered only peer-reviewed observational studies published within the period 1990–2024 with various designs and demographics. This year range was considered in recognition of the advancement in neuroimaging and the evolving understanding of ageing mechanisms in the past few decades. Thus, including articles from 1990 onwards ensures that the latest literature is captured. All articles that reported on neuroimaging or neuroradiological findings on ageing-related cerebral large artery and small vessel diseases were deemed eligible for inclusion. Our study evaluated a range of cognitive impairments associated with cerebral large artery and small vessel diseases, focusing on specific domains such as executive function, visuospatial abilities, verbal skills, memory, abstract thinking, attention, and processing speed. We chose not to impose a specific diagnostic framework of vascular dementia to ensure a comprehensive overview of cognitive deficits reported in the literature. This approach allows us to capture the full spectrum of cognitive challenges observed in these conditions. While vascular dementia is indeed a known implication, our intention was to highlight the individual cognitive domains affected without overlooking important articles that might not explicitly categorise findings under vascular dementia. To avoid potential issues such as misinterpretations, omissions, or errors from translating articles [46], only those published in English or with English translations were included in the study.

Exclusion

Primary research articles that did not address the pathophysiological mechanisms of ageing-related cerebral large artery or small vessel diseases, as well as those focusing on neuroimaging findings involved with stroke or those unrelated to ageing, were excluded. The study also excluded research on other neurodegenerative diseases such as Parkinson’s syndrome, amyotrophic lateral sclerosis, and Alzheimer’s disease. Additionally, studies involving non-human subjects, those reporting on genetic, environmental, and molecular mechanisms, and those with insufficient data integrity were not considered. Secondary studies, preprints, grey literature, conference reports, editorials, duplicate publications, and articles without an English full-text version or lacking peer review were also excluded.

Ethical approval

Although ethical approval is not a requirement for this type of study, we followed ethical reporting standards, as recommended by Kahrass and colleagues [47], to enhance integrity, credibility, or adherence to best practices.

Data sources

Electronic databases, including Web of Science, PubMed, Embase, Scopus, and EBSCOHost (including CNHIL and PsycINFO), were chosen for their extensive access to a wide range of journals, websites, organisational links, and other databases in biomedicine, health, sciences, and engineering. According to Bramer and colleagues [48], this combination ensures an optimal search outcome. Additionally, a manual search on Google Scholar was conducted to identify eligible papers that met the selection criteria, following the recommendation of Haddaway and colleagues [49].

Search strategy

Under the guidance of a certified librarian, researchers developed a robust search strategy incorporating critical features for an effective literature search in each database, a step deemed essential for authentic research by Grewal et al. [50]. This strategy was adapted for each database, considering their unique sensitivities and search requirements.

The search strategy incorporated keywords such as [(Ageing or aging); (Pathophysiology or mechanisms or “blood–brain barrier” or “cerebrovascular reactivity” or autoregulation or hypoperfusion or “oligodendrocyte precursor cells” or dysfunction or compromise or impairment or glymphatic or neurovascular coupling or “blood–brain barrier” or “BBB leakage” or "brain changes" or "vascular changes"); (Cerebrovascular or neurovascular or neurologic or cognitive or "brain vessels" or arteries or veins or microvasculature or vascular or haemodynamics or "haemodynamic mechanisms"); ("Cerebral artery diseases" or "large vessel diseases" or "small vessel diseases" or "large vessel diseases" or "white matter hyperintensities" or microbleeds or lacuna or "brain atrophy" or microangiopathy or "intracranial atherosclerosis" or ICAS or "intracranial arterial calcifications" or IAC); (Neuroimaging or magnetic resonance imaging or MRI or MRA or computed tomography or CT or CTA or ultrasound or transcranial ultrasound or TCD or TCCD or carotid doppler or Doppler)] based on the research purpose.

These keywords were further refined using Boolean operators (AND, OR) and truncation (*) to combine related terms and synonyms, maximising sensitivity to capture relevant articles [51]. Two reviewers (JAA and HD) independently conducted an electronic literature search from April 1st to 15th, 2024, and it was updated on February 2, 2025.

Study selection and data extraction

After the initial data search, results from each source were combined, duplicates were removed, and a comprehensive dual screening was conducted using EndNote, which Stoll et al. suggest improves the precision of study selection [52]. Two reviewers independently screened the titles and abstracts of all retrieved articles according to a pre-established review protocol (PROSPERO with ID: CRD42024531238) validated by expert researchers. A full-text screening of the selected studies was then independently performed by two reviewers (JAA and HZ), with any discrepancies resolved by a third reviewer (XC) in consultation with the team.

Data extraction was carried out on the included papers. Two independent reviewers filled out tabular templates with relevant information, summarised in Tables 1 and 2 for neuroimaging and cognitive outcomes, respectively. Extracted characteristics included references and year of publication, sample size, age, gender, predicting or exposure variable, neuroimaging correlates of outcome variable, cognitive domains, and key findings. The lead investigator convened a consensus meeting to resolve any discrepancies. These methods were adopted to ensure a high level of methodological consistency, as recommended by Charrois [53].

Table 1 Study Characteristics with neuroimaging outcomes

Table 2 Study characteristics with cognitive outcomes

Methodological quality assessment

Using the Newcastle–Ottawa Scale (NOS) for cohort studies, the included articles were rigorously evaluated against three domains (selection, comparability, and outcome) in order to assess their methodological quality.

Data synthesis and analysis

To integrate the results from the included studies, a narrative synthesis was employed to combine the findings from the qualitative and quantitative extracted data. This approach provides a comprehensive overview to bridge the research gap, as it allows for a robust synthesis and analysis of available data [54, 55]. A meta-analysis was performed using the Jamovi and R statistical software to integrate findings from nine studies. The 9/52 studies were selected because they were the only studies that compared the metrics of cerebrovascular functions measured between individuals with cerebral large artery and small vessel disease (target group) and those without any vascular disease (control group). CVR, cerebral blood flow, arterial stiffness, and blood pressure were the quantitative metrics of cerebrovascular functions measured between target and control groups. Since the metrics of cerebrovascular functions were quantified based on different scales and may change in opposing directions with disease progression, the standardised mean difference (SMD) was chosen to assess the outcome effect size in a random effects model. We defined the random-effects model based on the foundational method by DerSimonian and Laird, which accounts for the variability both within and between the studies included [56]. This model assumes that the true effects vary across studies due to differences in study populations, methodologies, and other factors. By using this approach, we could generalise our findings beyond the specific conditions of each study.

Using a two-step approach, we initially evaluated each metric separately then combined the four metrics—CVR, cerebral blood flow, arterial stiffness, and blood pressure—to obtain an overall effect size, as they co-occur in participants. This pooled SMD allows for a comparative evaluation of each metric's relative effect size against the direction and magnitude of the overall or combined effect size in the model. By examining the magnitude, direction, and statistical significance of each metric's SMD, we could identify the predominant metrics. From the statistical model, outcomes for random effects were reported, and the I² statistic was reliably reported for heterogeneity assessment, as it is less affected by the number of studies included in the meta-analysis. Subsequently, we performed subgroup analyses on participants who exclusively had cerebral small vessel diseases and those with cerebral large artery diseases. We focused on metrics with similar directional effects, specifically cerebrovascular reactivity (CVR) and cerebral blood flow in the subgroup analysis. Meta-regression with age, type of cerebral arterial disease, and risk of bias were performed to explore potential sources of heterogeneity. A two-sided P value < 0.05 was considered statistically significant.

Results

Study selection

We retrieved a total of 1165 articles after eliminating 447 duplicates from an initial pool of 1612 records: EBSCOhost (n = 449), Embase (n = 222), Scopus (n = 123), Web of Science (n = 437), and PubMed (n = 381). Subsequently, the screening process, based on titles and abstracts, excluded 1089 articles. After the initial screening, 76 articles remained for full-text assessment to determine their eligibility. Following the full-text screening against the predefined eligibility criteria, 44 articles remained for inclusion. Additionally, a manual search of relevant reference lists identified 8 articles that met the inclusion criteria. Finally, this review included a total of 52 primary studies, with 9 incorporated for meta-analysis, and the entire approach is outlined in accordance with the PRISMA flow chart Fig. 1.

Fig. 1

PRISMA flow diagram for search and study selection

Study characteristics

The included articles comprised 52 observational studies [7, 8, 24, 28, 31, 57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103] published from 2000 to 2024, including cohort-based, cross-sectional, longitudinal, and case–control studies. From a comprehensive dataset of 52 studies involving 23,693 adult participants (average age = 67.4 years), diverse features related to cerebral large artery and small vessel disease, normal cerebrovascular characteristics, as well as cognitive effects were reported. From the 52 included studies, 45 reported complete demographic data on gender, with 47.9% (10377 out of 21,643) participants being males. The distribution of the entire 52 studies by countries is shown in Fig. 2. The results were deduced from neuroimaging investigations that utilised MRI, CT, and ultrasound. Cognitive assessment tools, including the Mini Mental State Examination (MMSE), the Repeatable Battery for the Assessment of Neuropsychological Status (RBANS), and the Montreal Cognitive Assessment (MoCA), were utilised by various investigators to evaluate overall cognitive impairments. Cognitive effects were assessed by capturing the cumulative evaluation of various cognitive domains, including executive function, visuospatial abilities, verbal or language skills, memory, abstract thinking, attention, and processing speed. The Trail Making Test was also commonly used to assess specific aspects of executive function. The results were presented according to qualitative and quantitative outcomes.

Fig. 2

Distribution of studies by countries. Overall, forest plot. A negative overall SMD indicates a reduction in effect size, while a positive SMD indicates an increase. This figure demonstrates an overall reduction in cerebrovascular function in participants with cerebral large artery and small vessel diseases compared to the control group without vascular disease

Qualitative outcome

We qualitatively examined how various key disease mechanisms, as underlying predictors—neurovascular uncoupling, dysfunctional cerebrovascular reactivity and autoregulation, blood–brain barrier (BBB) leakage, and glymphatic dysfunction—influenced the burden of cerebral large artery diseases and small vessel pathology as well as their consequent cognitive implications. While many of these underlying mechanisms are typically considered as predictive or exposure variables, it is important to note that some investigations have identified impairments in these mechanisms as consequence or outcome variables. For instance, some neuroimaging features such as atherosclerotic stenosis and calcifications have significantly predicted impairment in cerebrovascular functions (Table 1). This bidirectional relationship is further elaborated in the following sections.

Table 1 summarises the 48 studies [7, 8, 24, 28, 31, 57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99] that explored the intricate relationship between the ageing brain, underlying mechanisms, and their consequential neurovascular and neuro-parenchymal changes. Additionally, Table 2 summarises the 14 studies [28, 29, 59, 64, 66, 72, 75, 76, 86, 91, 95, 96, 98, 99] that meticulously documented the neurological and cognitive challenges accompanying the ageing-related cerebral large artery and small vessel diseases. The cognitive outcomes examined were exclusively associated with the ageing process, excluding other neurodegenerative conditions like Parkinson's disease, amyotrophic lateral sclerosis, or Alzheimer's disease.

The role of neurovascular coupling

The neurovascular unit, which is comprised of pericytes, astrocyte end-feet, and neurones surrounding the endothelial cells in the brain capillaries, maintains cerebrovascular health through neurovascular coupling [2, 104]. An effective neurovascular coupling regulates cerebral blood flow by coupling alterations of vascular dilation in response to the metabolic requirements for a given neuronal activity [105, 106]. Huang and colleagues combined resting-state functional magnetic resonance imaging and arterial spin labelling to investigate the dysfunctions in neurovascular coupling among 86 stroke-free individuals with cerebral small vessel diseases. The results showed that higher loads of cerebral small vessel diseases, specifically WMH, were attributed to neurovascular uncoupling with abnormal blood flow and poor functional connectivity strength noted in various regions of the prefrontal cortex, posterior cingulate cortex, thalamus, and parahippocampal gyrus [28]. The findings were corroborated by Porcu and colleagues, who observed similar trends and demonstrated that disruptions or reductions in neural activity within regions affected by white matter hyperintensities contributed to cognitive impairment in a study involving 75 healthy participants [71]. These outcomes could imply that reductions in cerebral neural activity and poor cerebral perfusion related with neurovascular uncoupling, may disrupt the cerebral microenvironment and facilitate neurotoxins due to impaired BBB [106]. The disturbances in cerebral microenvironment have been shown to affect the ability of the neurovascular unit to regulate blood flow responses to neuronal stimulations, which could disrupt the structural and functional integrity of the cerebral microvasculature with consequent cognitive impairment, depending on the region affected [107, 108]. While neurovascular uncoupling is recognised as a critical mechanism in cerebral small vessel disease at the microvascular level, further research is needed to explore its potential connections to cerebral large arterial disease, as current studies have yet to establish this link.

The role of cerebral BBB

Given that the BBB comprises specialised endothelial cells that facilitate the functions of the neurovascular unit in maintaining cerebral homeostasis [104], an impaired BBB may be implicated in cerebral artery and small vessel diseases. With ageing, there is a reduction in capillary density and length, which compromises the efficiency of the blood–brain barrier (BBB) with a dysfunctional haemodynamics and impediment to the transport of essential molecules [63, 86, 88]. Beyond the fact that several studies [32, 86, 87], have associated BBB leakage, as quantified by dynamic contrast-enhanced (DCE-MRI), to higher burdens of cerebral small vessel disease, our critical synthesis revealed that BBB leakage is diffusely distributed beyond the white matter regions [31]. In their study, Kerkhofs and colleagues did not establish a cross-sectional association between cerebral small vessel disease and compromised BBB or leakage; however, a 2-year follow-up showed a decline in cognitive functions [101]. This finding suggested that the detrimental effects of BBB leakage accumulate over time, and its association with poor cognitive function may be time-dependent. However, other perspectives proposed that the neurological implications of BBB leakage might not be solely time-bound but rather influenced by the severity of the leakage rate as well as the specific region of leakage [88, 104, 109]. This view is supported by studies from Porcu and colleagues, which indicate that small vessel lesions in the periventricular region, as opposed to the deep subcortical and juxtacortical areas, may significantly disrupt regional cerebral activity and have the greatest impact on cognitive impairments [71]. Although, cerebral large artery diseases such as intracranial atherosclerosis and arterial calcification have been associated with endothelial dysfunction [5], the pathway by which BBB leakage contributes has not been clearly elucidated and warrants further investigation.

The role of cerebrovascular autoregulation and reactivity

Cerebrovascular autoregulation and reactivity are both implicated in both cerebral large artery and small vessel diseases and not just in small vessel pathology [1, 110, 111]. While cerebrovascular autoregulation maintains constant cerebral blood to ensure cerebral perfusion irrespective of arterial pressure fluctuations [30], CVR ensures the proper adjustment to adequate blood flow in response to metabolic demand [112]. Evidence suggests that the two mechanisms complement each other, and an impairment in one disrupts the cerebral haemodynamic functions of brain perfusion and neuro-parenchymal nourishment. [11, 72, 111, 113]. In a regional assessment of normal-appearing white matter and areas of small vessel lesions in the brain, as studied by Marstrand et al. [67] and Sam et al. [114] it was revealed that a reduction in CVR led to decreased blood flow with decreased fractional anisotropy and increased diffusivity—both indicating impaired brain parenchymal integrity [89, 114]—and in regions with white matter diseases. Similar findings were observed in a longitudinal study by Sam and colleagues, who further revealed that normal-appearing brain regions that reduced CVR at baseline eventually progressed to small vessel disease in a 1-year follow-up with poor structural integrity [61]. Such disease progression is partially explained by the insufficient haemodynamic function due to impaired autoregulation with poor cerebral perfusion disrupting the fibres of the brain matter with consequent cognitive impairment [64, 72, 73, 115, 116]. Reduced cerebrovascular autoregulation and CVR revealed in individuals with stenotic intracranial large artery diseases [65], is associated with arterial wall stiffness and poor blood flow [58, 117,118,119]. The study by Robert et al. found that arterial stiffening was linked to poorer cognitive functions, but these associations weakened after adjusting for cerebral small-vessel disease markers, highlighting the importance of addressing small-vessel disease in cognitive decline [77].

The role of the glymphatic system

The glymphatic pathway, which is responsible for cerebral fluid or waste clearance, is shown to be disrupted in ageing [4]. Through diverse pathways, different studies have associated impaired glymphatic system to cerebral large and small vessel diseases [33, 90, 98]. The evidence shows that different mechanisms may be implicated for glymphatic dysfunction depending on the underlying vascular disease. From the studies by Cai and colleagues [90], it can be inferred that small vessel lesions along the periventricular regions were predominantly attributed to venous impairment due to a dysfunctional glymphatic pathway. On the other hand, small vessel lesions along the deep subcortical regions are attributed to the cascades of ischaemia-hypoperfusion as well as glymphatic system impairment. A dysfunctional glymphatic pathway brings about the accumulation of waste cerebral metabolites and neurotoxins, which disrupts the cerebral microenvironment [4]. A dysfunctional cerebral waste drainage or clearance may promote neuroinflammation, endothelial dysfunction, and consequently atherosclerotic processes, affecting both the microvasculature as well as the basal cerebral arteries [3, 98]. Endothelial disruptions caused by glymphatic impairment, along with inflammatory changes, may initiate the atherosclerotic process in cerebral basal arteries, highlighting the role of glymphatic impairment in cerebral large artery diseases [120]. In a diffusion tensor imaging (DTI) analysis along the perivascular space, glymphatic dysfunction was shown to be associated with overall cognitive function, executive function, attention function, and memory, independent of the underlying phenotypes of small vessel diseases such as microbleeds, lacunes, WMH, enlarged PVS, and traditional vascular risk factors [91]. This has been partly explained by the neuro-axonal destruction in addition to the disruption of vascular integrity leading to stiffness and reduced vasomotor reactivity with consequent cognitive impairments [112, 121].

Quantitative outcome

A meta-analysis compared differences in effect size of cerebrovascular function metrics—CVR, blood flow, arterial stiffness, and blood pressure—between stroke-free individuals with cerebral large artery and small vessel disease and those without these conditions. Nine studies (3,729 participants, mean age = 60.2 ± 11.5 years) were included, revealing significant differences in these parameters, with estimates indicating poor cerebrovascular function in the target group compared to controls. From the individual random-effects model (Fig. 3), the analysis revealed negative effect sizes of CVR [SMD = − 1.86 (95% CI − 2.80, − 0.92)] and cerebral blood flow [SMD = − 2.26 (95% CI − 4.16, − 0.35)], respectively indicating a reduction in cerebrovascular functions in individuals with cerebral large artery and small vessel diseases compared to their control group with no vascular disease. Conversely, there were significant increases in the measures of blood pressure [SMD = 0.32 (95% CI 0.18, 0.46)] and arterial stiffness [SMD = 0.87 (95% CI 0.77, 0.98)], which signified poor cerebrovascular functions in the target group.

Fig. 3

Forest plot assessing each metric of cerebrovascular function. From (a) and (b), the negative effect sizes of CVR [MSD = − 1.86 (95% CI − 2.80, − 0.92)] and cerebral blood flow [SMD = − 2.26 (95% CI − 4.16, − 0.35)], respectively, indicate a reduction in cerebrovascular functions in individuals with cerebral large artery and small vessel diseases compared to their control group with no vascular disease. Conversely, (c) and (d) respectively show positive effect sizes of blood pressure [MSD = 0.32 (95% CI 0.18, 0.46)] and arterial stiffness [MSD = 0.87 (95% CI 0.77, 0.98)] indicating an increase in these metrics, which signify poor cerebrovascular functions. Forest plot for subgroup analysis. CVR metrics significantly reduced in individuals with with cerebral large artery and small vessel diseases compared to the control group without vascular disease. Whiles, blood pressure and flow were elevated but this increase was insignificantly observed between the two groups. There were significant declines in cerebrovascular functions specifically noted in individuals with cerebral large artery disease (SMD= − 2.09 [95% CI − 3.57, − 0.62]

In the combined model (Fig. 4), the overall average effect size was negative [SMD − 0.81 (95% CI − 1.53 to − 0.08), p < 0.001]. Comparatively, this suggests that the negative impacts of CVR and cerebral blood flow reductions significantly outweighed the effects of blood pressure and arterial stiffness, thereby predominantly shaping the overall model.

Fig. 4

Forest plot for combined metrics of cerebrovascular functions. The forest plot shows measures of CVR and cerebral blood flow were reduced in the target group compared to their control. Conversely, blood pressure and arterial stiffness were increased in the target group compared to their control group. There is indication that the negative effects of CVR and cerebral blood flow predominantly influenced the overall negative effect size of the random effects model in the combined measures of CVR, blood pressure, cerebral blood flow, and arterial stiffness

Heterogeneity assessment

The substantial heterogeneity among study outcomes may be largely associated with the impact of the different metrics of cerebrovascular function—such as CVR (I² = 87.7%), cerebral blood flow (I² = 94.4%), arterial stiffness (I² = 13.5%), and blood pressure (I² = 83.7%)—which capture distinct aspects of cerebrovascular functions, with varying impacts on disease burden. Meta-regression analysis revealed that age, sample size, and risk of bias scores do not account for the variability in findings. Meanwhile, the type of vascular disease (cerebral large artery disease or small vessel disease) significantly contributed to the variability (model coefficient, F = 7.16, p = 0.017). Figure 5 presents a subgroup analysis demonstrating that cerebrovascular functions, specifically CVR and cerebral blood flow, are significantly diminished in individuals who exclusively had cerebral large artery disease (SMD = − 2.09, 95% CI − 3.57 to − 0.62) and those with small vessel disease (SMD = − 0.85, 95% CI − 1.34 to − 0.36) compared to their respective controls. Due to the limited number of studies (n = 9), we did not further stratify the analysis by sample size.

Fig. 5

Subgroup analysis according disease subtype. Using the measures of cerebral blood flow and CVR as the measures of cerebrovascular function in subgroup analyses: (a) Shows an overall negative effect size [SMD = − 0.85 (95% CI − 1.34, − 0.36)], which signifies a reduction in cerebrovascular functions in participants presenting exclusively with small vessel diseases. (b) Shows an overall negative effect size [SMD = − 2.09 (95% CI − 3.57, − 0.62)], which signifies a reduction in cerebrovascular functions in participants presenting exclusively with large arterial diseases. Schematic model for cerebral large artery and small vessel diseases. A simple diagram that could serve as a guide to explore the interconnectedness of underlying mechanisms implicated in cerebral large artery and small vessel diseases

Overall quality assessment outcome and risk of bias

In strict compliance with the Newcastle–Ottawa Scale (NOS) guidelines for evaluating cohort studies, an excellent 100% risk-of-bias assessment outcome was attained among the analysed studies (supplementary table), demonstrating exceptional rigour and the high quality of the evidence presented in this study. These studies meticulously met all the criteria across the three critical domains of the NOS: selection, comparability, and outcome. Moreover, given the small number of studies included (n < 10), a publication bias assessment was not performed. This decision aligns with the Cochrane meta-analysis guidelines, which advise that publication bias outcomes from such a limited dataset are likely to be skewed and non-representative [122].

Discussions

From our qualitative synthesis, the impact of neurovascular uncoupling, impaired cerebrovascular reactivity and autoregulation, glymphatic impairment, and blood–brain barrier leakage is diverse and may not be specific to particular cerebral artery diseases; however, they collectively underscore the complex interplay of cerebrovascular dysfunction contributing to cognitive decline in ageing populations [9, 28, 71, 81, 102, 103, 116]. Cerebrovascular dysfunctions predominantly influence the severity or higher burdens of the cerebral large artery and small vessel diseases, although a bidirectional relationship may exist. In a meta-analysis, individuals with cerebral large artery and/or small vessel diseases were shown to exhibit significantly reduced or impaired cerebrovascular function compared to healthy controls. The meta-regression results suggest that for stroke-free individuals in their 60 s, the status of their cerebrovascular functions is significantly influenced by whether they have cerebral artery or small vessel diseases or they do not. The significant effects of CVR, cerebral blood flow, blood pressure, and arterial stiffness hint at the reliability that these parameters could serve as indicators of cerebrovascular health in a stroke-free population. Our model further suggested that the negative impact of cerebrovascular reactivity and cerebral blood flow reductions may outweigh other metrics such as arterial stiffness and blood pressure. This suggests that reductions in CVR and cerebral blood flow are more linked to decreased cerebrovascular functions. Given that reduction in cerebrovascular functions is previously associated with reduced cognitive functions [64, 112], our findings suggest that reductions in CVR and cerebral blood flow in individuals in their 60 s may indicate risks of cognitive impairment.

Neurovascular and neuroparenchyma changes

It is speculated that with advanced ageing, neurovascular uncoupling and BBB leakage compromise the cerebral microvascular circulation, but their association with the burdens of cerebral large artery diseases is not well-interrogated [28, 106]. Previous studies have speculated that ageing-related changes in arterial wall calibre, haemodynamic functions, and endothelial integrity may initiate the formation of new vasa vasorum in both the anterior and posterior intracranial arterial walls, increasing their vulnerability to plaque development and hypoxic-ischaemic events [6, 8, 123]. This association is still debatable, and elucidating the exact mechanism linking the presence of vasa vasorum to intracranial large arterial plaque formations may provide insights into the pathogenesis of cerebral large artery diseases and probably how the microvasculature impacts this process.

In actual cases of intracranial atherosclerotic stenosis or occlusion, the circle of Willis, through its communicating arteries, often serves as a compensatory network; however, its efficacy may be diminished by age-related increased vascular stiffness and loss of arterial compliance [39, 124]. Such ageing-related alterations—implicated in dysfunctional cerebrovascular reactivity and autoregulation—could aggravate cerebral hypoperfusion and subsequent ischaemic events [8, 9, 58, 67, 72]. The prominent regions showing parenchyma alterations include the prefrontal cortex and posterior cingulate cortex for lacune infarcts and microbleeds; periventricular and deep subcortical regions for white matter hyperintensities and lacune infarcts; basal ganglia and centrum semiovale for enlarged perivascular spaces; thalamus for lacune infarcts, as well as diffuse and focal alterations in the parahippocampal gyrus, brainstem, and brain matter tracts such as the corpus callosum [28, 31, 81, 86, 87, 89, 125].

From our synthesis, we could infer that the neuroparenchyma deterioration may be intricately linked to the vascular changes of both large and microvasculature, as they both impact blood flow and nutrient delivery, which, when disrupted, may exacerbate neuronal vulnerability and precipitate the pathogenesis of neurological disorders [126]. Over the years, a consensus, based on MRI investigations, has emerged among various researchers [61, 65, 80, 86, 96, 114], who have consistently reported similar findings of an age-related increase in neuroparenchyma diffusivity and a decline in anisotropy, indicating a reduction or loss of brain parenchymal integrity. What is also uncertain is whether ageing-related haemodynamic dysfunctions in non-occluded extracranial arteries (carotid or vertebral) could significantly influence those of the intracranial major vasculature as well as the microvasculature.

The interrelatedness of mechanisms

The interrelatedness of mechanisms describes how neurovascular uncoupling, BBB leakage, CVR, cerebral autoregulation, and glymphatic impairment are interconnected in contributing to the pathogenesis of both cerebral large artery diseases and cerebral small vessel diseases. Neurovascular uncoupling disrupts the coordinated response between neuronal activity and blood flow, leading to inadequate cerebral perfusion [2, 28, 127]. This, combined with impaired CVR and cerebral autoregulation, leaves this condition unresolved, which may exacerbate the ischaemic damage presenting as lacune infarcts and WMH [30, 59, 72, 73, 89, 103]. BBB leakage allows neurotoxic substances to enter the brain, and glymphatic impairment further hinders the clearance of these substances, contributing to neuroinflammation and vascular damage presenting as enlarged PVS and microbleeds [3, 31, 32, 86, 88]. Persistent inflammatory changes are potent factors for the endothelial alterations that may initiate atherosclerosis and calcification in various vascular beds, including the intracranial arteries [120, 128]. Together, these processes create a vicious cycle that accelerates the progression of both large and small vessel pathologies [9, 62]. Given this cycle, it can be inferred that a disruption in one of these key mechanisms directly or indirectly impacts the other, leading to cerebrovascular dysfunction. We also posit that the association between cerebral large artery disease and small vessel diseases may be bi-directional, such that changes in the microvasculature could disrupt the integrity of large arteries and vice versa. Notably, elevated blood pressure is a recognised moderator that contributes to the pathogenesis of cerebral large artery disease by promoting endothelial dysfunction, arterial stiffness, and atherosclerosis, while in small vessel disease, elevated blood pressure may facilitate arteriolosclerosis, microvascular damage, and impaired autoregulation, exacerbating ischaemic and haemorrhagic lesions [59, 93, 129,130,131]. Badji and colleagues contend that even with medication, uncontrolled high blood pressure significantly compromises cerebral perfusion and waste clearance [81], which in turn could accelerate cognitive decline as part of the advanced ageing process [66].

Figure 6 summarises the results, depicting the underlying mechanisms of cerebral small vessel disease or/and cerebral large artery diseases and their interrelatedness with cognitive impairment.

Fig. 6

Schematic diagram of study results

The imaging assessment of mechanisms

The neuroimaging assessments of the ageing-related pathophysiological mechanisms underlying cerebral large artery and small vessel diseases present varied and unique advantages in the neurological examinations (Table 3).

Table 3 Imaging assessments of mechanisms

Clinical implications

Utilising multimodal imaging to non-invasively explore assessment metrics, as summarised in Table 3, can enhance the precision of diagnostics, preventative strategies, and disease-specific interventions. Blood oxygen level-dependent (BOLD) MRI signals are used to reflect functional connectivity in blood flow and oxygenation, where discrepancies between neuronal activity and BOLD signals can indicate neurovascular uncoupling. Arterial spin labelling (ASL) MRI can quantify changes in cerebral blood flow in response to metabolic demands, such as variations in arterial partial pressure of carbon dioxide, which is critical for assessing CVR. Transcranial ultrasound complements ASL-MRI by providing real-time data on cerebral blood flow velocity, aiding in the evaluation of autoregulation through continuous monitoring of blood pressure changes. Additionally, dynamic contrast-enhanced MRI is commonly used to evaluate BBB permeability by measuring leakage volume or rate. Diffusion tensor imaging (DTI) assesses diffusivity and the along perivascular space (ALPS) index, reflecting glymphatic function.

These metrics are essential for assessing the underlying mechanisms of cerebral large artery and small vessel diseases, enhancing our understanding of disease progression and the impact on cerebrovascular function, ultimately aiding in the maintenance of cognitive health in the elderly. We further recommend the clinical evaluation of CVR, which is an easily assessed cerebrovascular functional parameter, to be utilised to assess the vascular risk of cognitive impairment among the elderly. For this purpose, transcranial ultrasound, which is widely available, relatively cheaper, and non-ionising, has proven very effective and reliable in accessing the cerebral blood flow and global CVR of individuals [115, 132,133,134].

Limitations

This study acknowledges some limitations related to the study methodology and outcomes. The meta-analysis included only 9 studies with four different metrics measuring major mechanisms of cerebrovascular functions. This diversity in metrics introduced substantial heterogeneity, which may affect the generalisability of the findings. However, this approach enabled us to capture a comprehensive understanding of how different metrics influenced cerebrovascular function, revealing that the observed reductions in CVR and cerebral blood flow and increases in blood pressure and arterial stiffness are consistent across various assessment methods. Additionally, the limited number of studies constrained our ability to conduct more detailed subgroup analyses or assess publication bias reliably and hence more studies are needed in this domain. Nevertheless, the meta-analysis comprised exclusively high-quality studies, all free from significant risk of bias, with a cumulative sample size of 3,729 participants. This substantial sample size provides significant statistical power and enhances the reliability of the findings. Including studies of high methodological quality reflects the reliability of our findings. Again, the study excluded other neurodegenerative conditions like Parkinson's disease, amyotrophic lateral sclerosis, or Alzheimer's disease, reducing the generalisability of the results to a broader context of neurodegenerative diseases.

Conclusion

In a vicious cycle, the mechanisms of neurovascular uncoupling, BBB leakage, dysfunctional CVR and autoregulation, as well as glymphatic impairment, accelerate the progression of both large and small vessel pathologies. The association between cerebral large artery disease and small vessel diseases may be bi-directional, such that changes in the microvasculature could disrupt the large arteries and vice versa. Individuals with cerebral large artery and/or small vessel diseases exhibit significantly reduced cerebrovascular function compared to healthy controls. The negative impact of cerebrovascular reactivity and cerebral blood flow reductions may outweigh other metrics such as arterial stiffness and blood pressure. This suggests that reductions in CVR and cerebral blood flow are more linked to decreased cerebrovascular functions. Given that reduction in cerebrovascular functions is previously associated with reduced cognitive functions, our findings suggest that reductions in CVR and cerebral blood flow in individuals in their 60 s may indicate higher risks of cognitive impairment. This study confirms that the application of multimodal neuroimaging offers comprehensive insights that facilitate precise evaluation of cerebrovascular pathologies, enhancing our understanding of disease patterns and elucidating how various pathophysiological mechanisms influence cognitive impairment in stroke-free populations.