Molecular biomarkers for vascular cognitive impairment and dementia: the current status and directions for the future

Dementia is a syndrome with various underlying pathologies acting independently or in concert to cause cognitive dysfunction. The development of disease-specific treatments and targeted prevention strategies requires precise clinical sub-typing via etiology and pathophysiological processes. Furthermore, recent research advances in biomarkers, especially for Alzheimer’s disease (AD) diagnosis, have improved diagnostic precision for dementia. This has led to the proposal of an AT(N) framework for AD diagnosis, with A, T, and N signifying the demonstrated presence of amyloid (A) and tau (T) pathologies as well as neurodegeneration (N) respectively. Many authors have proposed to add other pathophysiological mechanisms to the traditional AT(N) framework, such as neuroimmune dysregulation, synaptic dysfunction, and/or blood-brain barrier (BBB) alterations, to make the ATX(N) more accommodating of a range of pathologies important for dementia. This has included the special consideration of vascular pathogenesis being represented by a “V” (Sachdev, 2020). This indicates that cerebrovascular disease (CVD) is a common cause of dementia alone or in combination with AD, and the importance of developing biomarkers of vascular “V” brain pathology has been argued. Research progress in vascular cognitive impairment and dementia (VCID), the second most common cause of dementia, has however been slow, partly because of limited research in VCID relative to AD. The diagnosis of VCID presently relies mainly on clinical information and neuroimaging including magnetic resonance imaging and computed tomography. Consequently, there exists an imperative to develop molecular biomarkers for VCID with high sensitivity and accuracy.

The primary characteristic of VCID is that cognitive deficits are attributable to CVD, encompassing small vessel disease (SVD), large vessel disease, and intracranial hemorrhage (Figure 1). Small vessels comprise small arteries, arterioles, venules, and capillaries. The neurovascular unit, a composite assembly of diverse cell types including endothelial cells, neurons, astrocyte end-feet, pericytes as the capillary mural cells, smooth muscle cells as the arteriole and venule mural cells, microglia, oligodendrocytes, and the extracellular matrix, forms the BBB within small vessels. The BBB protects the brain from toxic substances in the blood, while providing nutrients to brain tissue. Recent evidence has demonstrated that neurovascular unit is not exclusively associated with BBB, but there are also morphologically distinct neurovascular complexes throughout the cerebral vasculature that interact intricately with each other (Schaeffer and Iadecola, 2021). The material exchange between circulating blood and cerebrospinal fluid (CSF) occurs across the blood-CSF barrier, which is composed of epithelial cells of the choroid plexus, and the blood-arachnoid barrier, comprising arachnoid epithelial cells. Furthermore, endothelial cells and oligodendrocytes form the oligovascular niche that maintains oligodendrocyte homeostasis and white matter integrity through the exchange of soluble signals, including brain-derived neurotrophic factor and fibroblast growth factor. Pericytes assume a crucial role in maintaining the BBB, preserving vascular integrity, regulating inflammation, and facilitating angiogenesis. In rodent brain experiments, chronic hypoperfusion has been observed to induce pericyte degeneration and diminish pericyte coverage within cerebral blood vessels (Uemura et al., 2020). Smooth muscle cells are responsible for cerebrovascular autoregulation, adjusting vascular tone in response to alterations in arterial pressure through changes in myogenic tone, thereby maintaining cerebral blood flow within a constant range across the pressure spectrum.

Figure 1.

Various mechanisms of VCID.

The pathophysiological processes of VCID underly various mechanisms, including SVD, LVD, and hemorrhage, caused by endothelial dysfunction, BBB breakdown, inflammation, oxidative stress, clotting pathway, neuronal and glial cell degeneration. Created with Microsoft PowerPoint. ADMA: Asymmetric dimethylarginine; BBB: blood-brain barrier; BDNF: brain derived neurotrophic factor; CRP: C-reactive protein; GFAP: glial fibrillary acidic protein; HIF-1α: hypoxia-inducible factor-1a; ICAM: intercellular adhesion molecule; IGF: insulin-like growth factor; IL: interleukin; LCN2: lipocalin 2; Lp-PLA2: lipoprotein-associated phospholipase A2; LVD: large vessel disease; MDA: malondialdehyde; MMP: matrix metalloproteinase; MR-proADM: mid-regional pro-adrenomedullin; NfL: neurofilament light chain; NGF: nerve growth factor; OxLDL: oxidized low-density lipoprotein; PAI: plasminogen activator inhibitor; PlGF: placental growth factor; RAGE: receptor for advanced glycation end products; sPDGFRβ: soluble platelet-derived growth factor receptor-β; SVD: small vessel disease; TGF: transforming-growth-factor; TIMP: tissue inhibitors of metalloproteinase; TM: thrombomodulin; TNF: tumor necrosis factor; tPA: tissue plasminogen activator; VCAM: vascular cell adhesion molecule; VCID: vascular cognitive impairment and dementia; VEGF: vascular endothelial growth factor; vWF: von Willebrand factor.

SVD serves as an inclusive designation encapsulating pathological alterations in small vessels. These heterogeneous pathological changes include acute ischemia and/or chronic hypoperfusion, which lead to endothelial dysfunction, inflammation, oxidative stress, clotting pathway, BBB breakdown, and neuron and glial cell degeneration. The degeneration of pericyte caused by prolonged hypoperfusion leads to BBB breakdown. Large vessel disease embodies a comprehensive term embodying pathological modifications in sizable to medium-sized arteries, thereby contributing to the development of atherosclerosis and atherothrombosis culminating in the rupture of atheromatous plaques. Large vessel disease and SVD share many pathophysiological processes and are closely related, as is evident in the concept of ‘large and small artery crosstalk’. It has been shown in mice that with approximately 50% stenosis induced in bilateral common carotid arteries, the tight junctions in the capillaries expand and the BBB is disrupted at least 2 hours later, signifying that hemodynamic changes caused by alterations in large vessels can also affect small vessels (Hattori et al., 2014). Furthermore, the loss of the elastic plate of the proximal artery owing to atherosclerosis can result in loss of the cushioning function in systole and increased pulsatile stress in downstream small arteries, thereby causing small vessel damage, which is referred to as atherosclerosis windkessel dysfunction. A recent study has shown that atherosclerosis, particularly in intracranial vessels, results in reduced synaptic signaling, increased RNA splicing, and oligodendrocyte development and myelination, and a subset of the related proteins and modules has also been found to be associated with AD (Wingo et al., 2020).

Intracranial hemorrhage in cortical or subcortical regions has been suggested to have differing etiologies. While hemorrhage in the subcortical regions is susceptible to the effects of aging and hypertension, hemorrhage in the cortical regions is related to cerebral amyloid angiopathy, which is comorbid in more than 90% of patients with AD (Michaud et al., 2013).

Previous studies have identified a large number of candidate molecules that are relevant biomarkers for the respective pathophysiologies of VCID (Figure 1). Published studies have approached this topic from different perspectives, examining the roles of biomarkers to differentiate patients with VCID from healthy controls or those with neurogenerative diseases including AD, dementia with Lewy bodies and frontotemporal dementia. There is some evidence that biomarkers related to neuronal and glial degeneration (e.g., neurofilament light chain and glial fibrillary acidic protein), endothelial dysfunction (e.g., vascular cell adhesion molecule 1, asymmetric dimethylarginine or von Willebrand factor) and inflammation (e.g., interleukin and tumor necrosis factor) can differentiate neurodegenerative diseases from healthy controls, but they are agnostic to syndromic diagnosis. A limitation of these biomarkers is that they are non-specifically elevated in systemic diseases as well as CVD. Some researchers suggest choosing biomarkers with high brain specificity, such as placental growth factor and mid-regional pro-adrenomedullin.

In differentiating VCID from neurodegenerative diseases, particularly AD, it is very difficult to distinguish between them, as VCID shares much pathophysiology with AD. Some researchers have therefore proposed the possibility of subcategorizing VCID to address more specific pathophysiology. In particular, subcortical SVD shows higher levels of BBB permeability compared to AD. Patients with subcortical ischemic vascular dementia had an elevated CSF/serum albumin quotient in comparison to patients with AD (Kettunen et al., 2022). Additionally, soluble platelet-derived growth factor receptor beta, which reflects pericyte injury, is a potential biomarker for subcortical ischemic vascular dementia as it can detect early BBB damage (Nation et al., 2019). It is noteworthy that AD biomarkers are useful for differential diagnosis, if only to exclude AD. Amyloid and tau imaging and CSF measurement of amyloid-β (Aβ) 42/40, phosphorylated tau (p-tau), and total tau are established biomarkers for AD pathophysiology. Recently, plasma and CSF levels of p-tau217 are becoming a widely accepted biomarker for AD (Palmqvist et al., 2020). Another study showed that a biomarker panel encompassing proteins, such as neurofilament light chain and matrix metalloproteinase-9, combined with tau protein and Aβ₄₂ exhibited a high degree of accuracy in differentiating patients with AD from those with subcortical ischemic vascular dementia (Bjerke et al., 2011). In the differentiation of dementia with Lewy bodies, recent advances in alpha-synuclein seed amplification assays can also rule out synucleinopathies as a differential diagnosis (Concha-Marambio et al., 2023).

Extracellular vesicles (EVs) of central nervous system (CNS) origin have also emerged as promising biomarker candidates, often referred to as CNS-enriched EVs. While molecular biomarkers in the blood are degraded by proteases and post-translational modifications, EVs are able to cross the BBB and can communicate cell-state-specific messages and protect their cargo from degradation through a phospholipid-membrane. There has been significant progress in the study of EVs as biomarkers for AD and Parkinson’s disease, but, as noted in a recent meta-analysis, the results show large heterogeneity and the presence of publication bias (Taha and Bogoniewski, 2023), suggesting that larger unbiased studies with more standardization are needed. The study of EVs in VCID has been very limited.

As noted above, abnormal levels of various molecular biomarkers have been reported in VCID (Hosoki et al., 2023), but so far there have been no clinical successes. Thus, the question arises: what underlies the absence of efficacious clinical applications concerning research of VCID biomarkers? The main reason is that none of the biomarkers solely suffices to encapsulate the multifaceted pathologies characterizing VCID which involves heterogeneous etiologies. Furthermore, the investigation of biomarkers for VCID poses considerable challenges in delineating its pathophysiological distinctions from healthy aging, as CVD is common with age. Therefore, a combination of biomarkers appears to be the way forward. When selecting biomarkers to be combined, it is crucial to select biomarkers from each of the multiple pathways, rather than from one pathway alone. Studies of genetic models of VCID are also very important for understanding the pathophysiology of VCID. Of them, it is worth noting that the vascular pathology of cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy (CADASIL) is similar to sporadic VCID and may serve as a useful model. Machine learning using data extracted from blood and CSF assays and neuroimaging has shown superior performance in identifying biomarkers. Several types of machine learning algorithms/methodologies have positive implications for processing large volumes of complex data and are useful for encompassing the different aspects of disease physiopathology (Hosoki et al., 2023).

Recently, transcriptomics utilizing single-cell RNA sequencing has elucidated the presence of segmental heterogeneity within the transcriptome of vascular cells, unveiling variations in the cerebrovascular tree and neurovascular correlations. Transcriptomics has revealed the molecular foundations underpinning the gradual transition in cell phenotype along the arteriovenous axis through phenotypic zoning, and offers potential for the advancement of specific biomarkers tailored to each endothelial cell type (Ihara and Yamamoto, 2022). Furthermore, genome-wide association studies have identified numerous risk loci for stroke and have established causal links between specific genes and gene regions in stroke pathophysiological processes. Recent strides in the domain of omics have significantly expanded horizons, underscoring the significance of multi-omics methodologies. Numerous studies have revealed links between the human microbiota and a wide range of pathologies, including CNS diseases, and the molecular mechanisms involved in the development and exacerbation of diverse diseases (Hosoki et al., 2020). Exploration of intricate multi-omics biomarkers that concurrently integrate genetic and environmental attributes will not merely propel our comprehension of the disease paradigm of VCID forward but will also unveil the potential for groundbreaking therapeutic interventions (Hosoki et al., 2023).

In conclusion, while strides have recently been made in identifying fluid biomarkers for VCID, their specificity and clinical utility are still in question. Future studies should attempt to better understand the pathophysiological basis of VCID, and aim for a panel of biomarkers, using both targeted as well as discovery strategies, that distinguishes VCID from healthy controls as well as other causes of cognitive disorders, detects it at early stages, is useful for prognosis, and serves as a proxy for treatment response.

This work was supported by the Japan Society for the Promotion of Science Overseas Research Fellow, the fellowship of Astellas Foundation for Research on Metabolic Disorders and the Japanese Society of Neurology grants for overseas study “Young Researcher Overseas Training Program” (SH) and NHMRC CRE grant 2006765 (to PSS).