4.2 Inflammatory Markers

Low-grade systemic inflammation underlies the pathophysiology of HT. C-reactive protein (CRP) and interleukin-6 (IL-6) are probably the most extensively studied inflammatory biomarkers in relation to CSVD, although results are not consistent among all populations [114–116].

CRP is an unspecific biomarker reflecting a systemic inflammatory state, and it has been associated with the risk and severity of ischemic stroke in healthy and in prehypertensive populations [117–120].

These results show that the interrelation between these two biomarkers (CRP and IL-6) and brain MRI lesion load is very controversial through the different investigations. WMH were the most frequently evaluated lesions; nevertheless, the results are not clear. Many factors could be responsible: study design, methods employed for biomarker determinations, sample size, age, and different ethnic composition, among others.

Elevations of the inflammatory markers lipoprotein-associated phospholipase A2 and myeloperoxidase, but not CRP, were found to be associated with a greater burden of WMH [131]. On the other hand, Shoamanesh’s study in a large middle-aged, community-based sample is perhaps one of the most extensive concerning the possible role of inflammation in the pathogenesis of CSVD [77]. Assessing the association of a panel of 15 systemic inflammatory markers with MRI findings they reported increased levels of tumor necrosis factor (TNF) receptor 2 and myeloperoxidase in relation with cerebral microbleeds, but not with WMH or lacunar infarcts. Higher levels of osteoprotegerin, intercellular adhesion molecule-1 (ICAM-1) and lipoprotein-associated phospholipase A2, as well as lower myeloperoxidase, were observed in participants with greater WMH volumes and silent cerebral infarcts (as opposed to the results obtained by the previously mentioned work), while no associations were observed between cerebral microbleeds and osteoprotegerin, ICAM-1, and lipoprotein-associated phospholipase A2. Additionally, no changes in CRP or TNF-levels accompanied brain MRI lesions (Table 1).

Leukocyte count was employed as a marker of systemic inflammation in a study including 1586 asymptomatic individuals. Elevated blood leukocyte count was associated with moderate to severe WMH, independently of age, hypertension, and DM [124].

Lower serum levels of soluble receptors for advanced glycation end-products (sRAGE) were associated with higher prevalence of silent brain infarcts and WMH in stroke-free Hispanic and African-American subjects in the USA, suggesting that sRAGE may be predictive of asymptomatic CSVD, particularly in ethnic/racial groups with increased risk for cerebrovascular disease [127].

ICAM-1 participates in inflammatory endothelial activation and is also involved in the pathogenesis of cerebral small and large vessel diseases. An association of high levels of ICAM-1 with more severe deep subcortical and periventricular WMH was reported by Han et al. in 175 elderly subjects without neurological impairment [126]. Six years later, a community-based study of 1763 subjects supported these results and also added an association with the presence of lacunar infarcts [77].

Hyperhomocysteinemia is considered an independent risk factor of cardiovascular disease, including myocardial infarction and atherosclerosis, as well as arterial and venous thrombosis. Homocysteine can lead to endothelial damage, thus contributing to vascular pathology, and it is known to be associated with MRI lesions and cognitive status in symptomatic CSVD (transient ischemic attack, subcortical stroke) [132]. Plasma homocysteine levels measured in a middle-aged cohort from the Framingham Offspring study, showed a strong inverse association with total cerebral brain volume, while higher homocysteine related with silent lacunar infarcts but not with WMH volumes [133]. The Rotterdam Scan study obtained similar results, evidencing a continuous and graded association with silent lacunar infarcts, although homocysteine levels were also higher in those with more WMHs [134]. Homocysteine was found to be a stronger predictor for diffuse CSVD than for isolated lacunar infarction [135], while an association with progression of WMHs was reported in a cohort with symptomatic atherosclerotic disease [65].

4.3 Oxidative Stress Markers

Endothelial dysfunction appears as a result of persistent ischemia–reperfusion and inflammation generating oxidative stress from unbalanced free radical formation that leads to peroxynitrite formation, lipid peroxidation, protein modification, matrix metalloproteinase (MMP) activation, and DNA damage [114].

It has been suggested that oxidative stress is also related with CSVD and stroke risk. Nevertheless, this association still remains unclear, especially due to the overlap between systemic inflammation, endothelial dysfunction, and oxidative stress. To our knowledge, direct biomarkers of oxidative stress (oxidative damage products and antioxidant enzymes) have not been investigated in relation to asymptomatic CSVD.

A study including 131 “stroke-free” individuals who were followed during an average of 13.8 years measured 8-iso-prostaglandin F2-α, 4-hydroxinonenal conjugated with mercapturic acid, 8-hydroguanosine, and 8-nitroguanine, and found that only higher 8-iso-prostaglandin F2-α levels were associated with the risk of stroke and increased risk stratification for stroke [136].

4.4 Hemostatic Biomarkers

Due to the close relation between CSVD, endothelial dysfunction, and platelet activation, during the last years several studies have evaluated markers of coagulation and fibrinolysis in relation to CSVD. The most studied markers are those related to changes in coagulation/fibrinolysis ratio, promoting a prothrombotic status, and some of these mediators are fibrinogen, plasminogen activator inhibitor (PAI), tissue plasminogen activator (t-PA), D-dimer, thrombomodulin, and von Willebrand factor.

Fibrinogen is an indicator of increased blood coagulation, plasma viscosity, and platelet activation, and raised levels have been associated with WMH and lacunar infarct [137], with leukoaraiosis and periventricular hyperintensity in patients with stroke and atrial fibrillation [138, 139], and with WMHs in nondiabetic patients with lacunar noncardiogenic ischemic stroke [140]. In the Rotterdam Study the authors found that fibrinogen levels are associated with WMHs in subjects over 65 years of age in the general population [141]. Plasminogen activator inhibitor (PAI) was independently associated with the presence of WMH and/or lacunar infarcts in some studies [125, 142]. The authors hypothesized that PAI-1 may play a role in maintaining the BBB through positive actions on tight junctions in cerebral endothelial cells [143]. t-PA is locally expressed within the brain and release of t-PA in the neurovascular unit increases permeability of the BBB [144, 145]. Van Overbeek et al. found an association between plasma t-PA-activity and progression of periventricular WMH [142]. Higher t-PA serum levels have been reported in lacunar stroke patients with WMH compared to patients without WMH [146], and in patients with lacunar stroke compared to non-stroke individuals [147]. D-dimer levels were independently associated with subclinical lacunar infarcts in the Atherosclerosis Risk in Communities cohort [148] and with total cerebral volume, but not with WMHs or silent brain infarct in the Framingham study [34]. Other studies employing small cohorts demonstrated increased D-dimer levels in CSVD, for example in lacunar infarcts and clinically manifest Binswanger disease, but the relation was not confirmed after multivariate analyses [149, 150]. Serum thrombomodulin has been associated with the presence and number of lacunar infarctions and with WMHs, especially with the coexistence of microalbuminuria in multiple studies, indicating an interesting association with renal impairment [151, 152]. High levels of von Willebrand factor (vWF)—an endothelial activation marker and an important performer in thrombosis—have been related with the presence and number of silent deep lacunar infarcts, with periventricular but not deep WMHs, and with WMH occurrence and progression in several population studies [148, 149, 153].

Other hemostatic markers have been less studied in CSVD, with non-concluding results. Results with tissue factor pathway inhibitor (TFPI) are controversial: higher [151] and lower [154] levels have been reported in patients with lacunar stroke compared with controls. In one study, the levels of TFPI were independently associated with the extent of WMH after multivariate analysis [151]. The soluble complex thrombin–antithrombin (TAT) was independently associated with the presence of WMH only in one small-sized study [149] and was found to be higher in a group of patients with subcortical vascular dementia and severe leukoaraiosis [150] Factor VII has been found to be increased in CSVD, whereas antithrombin III, a plasma protein that inactivates thrombin, was found to be reduced [147, 155].

Differences in the methodological conception of studies could influence the heterogeneity observed in the collected data. Different groups of subjects have been employed in studies: population-based and hospital cohorts, which differ in median age of enrollment, clinical status, and vascular comorbidities. Specifically, in hospital-based cohorts, the clinical status varies from asymptomatic and hypertensive subjects to previous symptomatic lacunar stroke patients [156]. In addition, the detection of MRI findings could be visual or volume-rated, where the expertise of the technician or the machine software could make major differences. Finally, different technical approaches employed in multiple laboratories for blood marker quantification, ranging from manual assays to automatic platforms and from turbidimetric/nephelometric methods to enzyme-linked immunosorbent methods, limit the reproducibility of the results.

4.5 Specific Brain Damage Markers

Blood-based brain-specific biomarkers have been employed to assess degree of brain injury and as outcome predictors in different clinical conditions (stroke, traumatic brain injury, and cardiac arrest, among others) [157, 158], and also as end points for evaluating the effect of new therapies [159].Among these biomarkers, brain-specific proteins (neuronal or glial in origin) have received growing attention in clinical neurological research [157–159], as well as autoantibodies to brain-specific proteins [160, 161]. The brain, chronically affected by CSVD, could be leaking brain-specific molecules into the blood stream at a rate much lower than in acute brain injuries (stroke, brain trauma, and global ischemia due to cardiac arrest). Studies employing autoantibodies to n-methyl-d-aspartate (NMDA) and α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptors have shown promise as biomarkers of brain damage for detection of asymptomatic CSVD [34, 160, 162–164].

In CSVD the permeability of the BBB is compromised as a result of the slight and maintained injury to brain parenchyma caused by sustained HT, DM, and other vascular comorbidities, facilitating the slow release of cytosolic contents of brain cells to the bloodstream. Several studies have measured brain-derived proteins in cerebrospinal fluid as markers of CSVD (published elsewhere); however, fewer studies have evaluated the same proteins in blood. Brain natriuretic peptide (BNP), neurofilament light chain (NfL), neuron-specific enolase (NSE), S100B protein, and β-amyloid are the most studied brain-derived markers in subjects with asymptomatic CSVD.

N-terminal pro-brain natriuretic peptide (NT-proBNP) is released together with BNP from cardiomyocytes in response to myocardial wall stress and is considered a marker of subclinical cardiac injury. BNP and NT-proBNP have been reported to be associated with both WMHs and silent lacunar infarcts [165]. In a study including 278 hypertensive patients free of stroke or dementia, NT-proBNP and cardiac troponin-T—another cardiovascular biomarker—were found to be independently associated with silent cerebrovascular lesions [129]. Reinhard et al. had previously reported very similar results with proBNP in a large sample of stroke-free diabetic patients (Table 1), thus supporting the notion that NT-proBNP could be a surrogate marker of vascular brain damage in hypertension [128].

A recent study demonstrated a twofold increased level of serum-NfL in CSVD subjects compared with healthy controls, and an association was observed with both imaging and clinical features of CSVD, suggesting a possible utility for assessing CSVD burden in asymptomatic subjects [166]. Higher NfL serum levels have also been associated with the occurrence of small subcortical infarcts and with new-CSVD-related MRI lesions, thus also presenting it as a putative marker of active CSVD [167].

A cross-sectional study conducted by our group included measuring serum levels of NSE and S-100B in 101 hypertensive patients and seeking association with brain MRI lesions indicative of CSVD. Higher levels of NSE but not of S100B were associated with more severe WMH [35, 168]. Furthermore, antibodies against brain-specific proteins have also been measured in blood as indicators of CSVD. Serum autoantibodies against the subunit NR2 of the NMDA receptor (NR2Ab) were explored in hypertensive subjects, and higher serum NR2Ab levels were related with more severe brain MRI lesions, particularly WMH denoting CSVD [164]. Serum NSE concentration could be a filter in identifying asymptomatic hypertensive patients with putative subclinical brain damage for subsequent brain MRI scanning and a prognostic marker for the occurrence of acute vascular events in the central nervous system.

S100B, a family of Ca²⁺ binding proteins, abundantly expressed in brain glia, have been independently associated with CSVD; Gao et al. suggest that increased S100B does not simply reflect the degree of brain injury but also plays an important role in the pathogenesis of CSVD [169]. Moreover, higher levels of S100B were independently associated with the presence and number of CMBs in patients with first-ever acute lacunar stroke [170]. A recent report demonstrated that microRNAs (miRNAs) could be used to differentiate patients with CSVD, specifically miR-409-3p, miR-502-3p, miR-486-5p, and miR-45 [171]. miRNAs are a class of small noncoding RNA molecules that are highly stable in biofluids because they are protected by membranes in exosomes and other microparticles and by binding to specific proteins [172].

A small cross-sectional study associated high plasma levels of Aβ1–40 with increased WMH and lacunes in cerebral amyloid angiopathy [173] and high plasma Aβ1–40 and Aβ1–42 levels with increased WMH and lacunes [173] but only in APOE e4 carriers [174].

A recent report concluded that higher serum levels of cystatin C, a cysteine proteinase inhibitor, are independently associated with the long-term progression rate of the cerebral WMH volume, independently of kidney function [175]. Previous studies have proposed that serum cystatin C levels might reflect the functional status of cerebral penetrating arterioles and the activity of neuronal degeneration processes, as cystatin C is mainly secreted from neurons, astrocytes, and microglia, it is very concentrated in the brain [176, 177] and is deposited in brain parenchyma and the walls of microvessels, inducing further neuronal and vascular degeneration [178, 179].

The use of brain-derived markers could be a promising option, due to their higher accuracy as a neurochemical expression of brain damage considering that they are secreted and/or released from specifically damaged brain tissue. As blood-based biomarkers, they represent a more accessible and less expensive tool, enabling physicians to take action earlier against the increased burden of stroke. However additional issues should be considered in order to obtain a precise result, such as the presence of the anatomical barrier, the reactive response of the brain to injury and the existence of extracerebral sources of the brain-derived markers.

5 Predicting Risk of Stroke in Asymptomatic Individuals

Potential blood biomarkers for predicting the risk of future strokes in patients with asymptomatic disease have been studied, as their introduction in clinical practice would be helpful in guiding management decisions, especially for individuals at highest risks.

During a 12-year follow-up of 1462 subjects in the Framingham sub-study of healthy elderly subjects, CRP in the top quartile predicted an increased risk of stroke [180]. Similar results were also reported in the Rotterdam Scan Study [181], while the risk for silent lacunar infarct was found to be 1.85 and 2 times higher with increased levels of high sensitivity CRP (hsCRP) and IL-6 respectively [122]. Particularly in this sense, hsCRP seems to be a very useful biomarker. An investigation conducted in a cohort of 10,456 healthy “cardiovascular disease-free” men (65 ± 8.9 years of age), that was followed during 15 years, individuals with hsCRP >3 mg/L showed a 40% higher stroke risk than those with hsCRP > 1 mg/L, after adjusting for blood pressure and cardiovascular risk factors [117].

A very large prospective study was conducted in postmenopausal women (n = 972; incident first ever stroke cases), where plasma CRP, IL-6, TNF-α, neopterin, E-selectin, VCAM, Factor VII, prothrombin fragment 1 + 2, tPA, plasminogen activator inhibitor-1 antigen (PAI-1), fibrinogen, and homocysteine, as well as fasting plasma glucose and lipids were measured at inclusion [182]. Of all the individual biomarkers examined, CRP was the only independent single predictor of ischemic stroke after adjustment for other biomarkers and standard stroke risk factors. The Biomarker Risk Score identified a gradient of increasing stroke risk with a greater number of elevated inflammatory/hemostasis biomarkers. No evidence for an association between stroke and levels of E-selectin, fibrinogen, tumor necrosis factor-alpha, vascular cell adhesion molecule-1, prothrombin fragment 1 + 2, Factor VIIC, or plasminogen activator inhibitor-1 antigen was encountered. This study supports the need for further evaluation of multiple-biomarker panels to develop approaches for stratifying an individual’s risk of stroke [182]. Very recently in a 2-year follow-up study of 123 elderly persons (age: 72.2 ± 8 years) Staszewski et al. reported that IL-6, platelet factor-4 (PF-4), CD40 ligand (sCD40 L), and homocysteine were associated with progressive brain MRI lesions of CSVD, suggesting that endothelial dysfunction modulates the radiological progression of CSVD through different inflammatory mechanisms [183].

Midlife systemic inflammation was found to be associated with late life brain MRI lesions indicative of CSVD in a bi-ethnic prospective cohort study (1485 participants, baseline age, 56 years—28% African ancestry). A strong association was demonstrated between CRP and periventricular white matter microstructural integrity, but CRP was not associated with the presence of cerebral infarcts or microbleeds. The authors concluded that midlife systemic inflammation may promote the development of chronic microangiopathic structural white matter abnormalities in the elderly [184].

Asymmetric dimethylarginine (ADMA), an inhibitor of endothelial nitric oxide synthase, is a marker of endothelial dysfunction. Pikula et al. [130] included 2013 stroke-free Framingham offspring (age, 58 ± 9.5 years) in whom they measured plasma ADMA levels, followed during approximately 5 years and performed subsequent brain magnetic resonance imaging measures of subclinical brain injury. They found that higher concentrations of ADMA were associated with a greater risk of silent lacunar infarcts after adjustment for traditional stroke risk factors but not of WMHs and proposed that ADMA may be a potentially useful new biomarker of subclinical vascular brain injury [130]. A few years later this group evaluated an eight biomarker panel comprising inflammatory (CRP), hemostatic (D-dimer and plasminogen activator inhibitor-1), and neurohormonal activity (aldosterone/renin, natriuretic peptide, and N-terminal pro-atrial natriuretic peptide), and endothelial function (homocysteine and urinary albumin–creatinine) in 3127 “stroke-free” subjects, who were followed for an average period of 9.2 years. The biomarker panel was associated with the incidence of stroke/TIA and with total brain volume but not with extensive WMH [34].

von Willebrand coagulation factor has also been associated with the risk of stroke. The Rotterdam Study demonstrated that the risk of stroke increased with the elevation of this biomarker in 6250 “stroke-free” individuals ≥55 years of age, who were followed during 5 years [185].

Serum enzyme activities of CK and GGT have also been evaluated in relation to stroke risk. In an urban Japanese population-based cohort study including 5026 initially healthy subjects (mean age 54.5 years), that were followed for 11.8 years on average, CK levels were useful to predict first-ever myocardial infarction in the future, but no relationship was observed with the risk for stroke [111]. GGT was evaluated in 456,100 eligible participants and was found to be independently correlated with increased risk of stroke after adjustment for alcohol consumption and stroke risk [113].

A longitudinal study of the cohort including 101 hypertensive patients, where our group had reported that higher blood levels of NSE, but not S100B were associated with more severe WMHs, provided additional information suggesting that increased NSE could also be a useful predictor for the occurrence of acute central nervous system vascular events in these previously asymptomatic hypertensive patients [35].

In the Three-City Dijon Study, with over 1600 participants, lower β-amyloid (Aβ) serum levels were associated with the progression of WMH volume. Lower levels of Aβ1–40 and Aβ1–42 were related with increased rate of progression of total and periventricular WMHs in CSVD [186]. The association of low plasma Aβ protein levels with increased WMH observed in this study could indicate an increased deposition of this protein in cerebral vessel walls, resulting in impaired cerebral blood flow [187].

6 Future Perspectives

Presently in clinical practice blood biomarkers have not gained use for stroke diagnosis and prognosis, given the more extensive availability of neuroimaging technology in the clinical settings that receive these patients and the many factors that confuse interpretation of most serum biomarker levels [188]. Nevertheless, very promising results have been obtained in the differentiation between TIA and stroke measuring blood levels of autoantibodies to NR2 peptides of the NMDA receptors [160, 161].

The potential application of blood biomarkers for the detection of subclinical brain damage and for predicting long-term stroke risk in asymptomatic individuals with vascular risk factors opens a new perspective that has attracted many investigators. The main challenge is to detect asymptomatic brain lesions with noninvasive and cost-effective techniques that can be easily performed and interpreted for widespread screening in the community. This would undoubtedly have a very positive effect for primary stroke prevention.

In this line of thought, numerous blood biomarkers have been investigated with very controversial results. In spite of this, there seem to be some unspecific blood biomarkers which have accumulated evidence as predictors of asymptomatic CSVD in cross-sectional studies or of incident TIA or stroke in longitudinal studies. According to Vilar et al. [114], considering the following selection criteria (two or more replications within independent studies and more than 1000 patients tested), to date, the best performing blood biomarkers in association with asymptomatic CSVD are CRP with WMH and IL-6, homocysteine, urine albumin–creatinine ratio, and amyloid β (1–42) peptide (all associated with silent brain infarcts and WMH).

Notwithstanding, blood-based brain-specific biomarkers will probably be able to offer much more in the detection of asymptomatic CSVD and in the long-term prediction of acute cerebrovascular events than measures of degenerative or inflammatory changes. The latter mainly correspond to unspecific peripheral processes, many of which are undoubtedly related to vascular changes in the central nervous system but do not directly represent what is occurring in the brain. This imposes a great challenge for researchers working in the field of brain-specific biomarkers and subclinical brain damage, who have to make a transition from current exploratory studies to large-scale, well-designed, cross-sectional, and prospective research in order to integrate them into clinical practice.