Aetiological blood biomarkers of ischaemic stroke

Each year, over 5 million people die worldwide from stroke, and at least every sixth patient who survives will experience another stroke within five years [1]. We are therefore eager to advance early and rapid diagnosis, prognosis and optimal risk stratification, as well as secondary prevention. In this context, blood biomarkers may improve patient care, as they have already done in other fields in the past, for example, troponin T/I in patients with heart attacks, natriuretic peptides in patients with heart failure or PCT (procalcitonin) [2] in patients with pneumonia. In the setting of acute stroke, a blood biomarker can be any quantifiable entity that reflects the manifestation of a stroke-related process. The most fruitful implementation of stroke biomarkers is in areas where information from traditional clinical sources is limited. There may be markers, for example, to guide risk stratification, reveal stroke aetiology, identify patients who may benefit most from interventions, monitor treatment efficacy, and recognise the risk of shortterm complications or unfavourable long-term outcomes. For this review we focus on blood biomarkers that could help distinguish the underlying aetiology of an ischaemic stroke. Stroke tends to be a much more heterogeneous condition than ischaemic heart disease, which is caused by atherosclerosis in the vast majority of cases. Causes of stroke include small and large vessel disease, cardioembolism, dissections, and rare vasculoand coagulopathies, among others. Because of this heterogeneity among stroke patients, it is clear that a monolithic approach to stroke prevention or secondary prevention is not warranted. Aetiological classification is important specifically because prognosis, risk of recurrence and management options differ greatly between aetiological subtypes. Considering that today up to 30% of stroke patients still cannot be classified into a specific subtype [3], the ability to improve aetiological classification to direct prevention methods at the underlying mechanism would be of greatest interest. For this review we collected data from studies, on aetiological blood biomarkers in ischaemic stroke patients, listed in PubMed up to October 2014. We describe the potential role of 22 selected blood biomarkers in the context of stroke aetiology. Finally we provide the readers with an outlook in this research field.


Epidemiology
Stroke is a leading cause of death worldwide.In the United States of America every year about 795,000 people suffer Examples of stroke aetiologies (imaging material adapted from Servier Medical Art).

Figure 2
Most important mechanisms for the formation of cardiac embolism (imaging material adapted from Servier Medical Art).PFO = patent foramen ovale from a new or recurrent stroke.In Switzerland, stroke incidence has been estimated at 150/100,000 inhabitants [4].In 2010 direct and indirect costs were $315.4 billion in the United States of America for cardiovascular disease and stroke [1,5].Stroke can be categorised into ischaemic and haemorrhagic stroke.Ischaemic stroke accounts for approximately 87% whereas haemorrhagic stroke accounts for 13% of all strokes [6].Ischaemic stroke itself is a heterogeneous disease and different pathological events can lead to a common final ischaemic cascade within the brain (see fig. 1).In a European population of incidence stroke, 30.2% were due to cardioembolism, 25.8% due to small artery/vessel occlusion (SVO) and 15.3% due to large artery atherosclerosis (LAA).The cause of ischaemic strokes remained undetermined in 25% to 39% of patients even after thorough examination [7].Moreover the risk of recurrence differs between the different etiological subtypes.The highest recurrence rate was found in cardioembolic strokes (22%), in the same study the recurrence rate for LAA was 10% [7].

Secondary prevention according to stroke aetiology
In any case of ischaemic stroke it is important to reduce the overall risk factors that led to the disease with, for example, blood-pressure treatment, antidiabetic agents, antiplatelet therapy and lipid-lowering agents in the case of atherosclerosis, heart disease or metabolic disorders.There are some aetiologies where there is, however, a specific treatment indicated.If the stroke is due to a cardioembolic event in patients with atrial fibrillation, it has been shown in several studies that these patients clearly benefit from oral anticoagulation.In a first randomised controlled trial -the WARS trialwarfarin was superior to placebo or aspirin in reducing stroke rates in patients suffering from atrial fibrillation [8,9].Over the years, meta-analysis data confirmed that warfarin reduces the occurrence of stroke in patients with atrial fibrillation by 64%.After that, several recent studies followed, proving a relevant risk reduction of stroke recurrence also for novel oral anticoagulants (direct thrombin inhibitors or factor Xa inhibitors) such as abixaban, dabigatran, rivoroxaban or edoxaban [10][11][12][13].
In the case of large vessel disease, if a patient suffers from a 70% to 99% stenosis without near-occlusion, carotid surgery (carotid endartectomy) showed an absolute risk reduction of 16% (p <0.001) [14].Among patients with symptomatic carotid stenosis, the 4-year rate of stroke or death was clearly reduced to 8.0% with stenting and to 6.4% with endarterectomy [15].Overall, the recurrence rate was reduced compared with "natural history" rates for both intervention types; however, endarterectomy performed slightly better than stenting with the exception of some subgroups.Because of the aetiological heterogeneity and the fact that prognosis, risk of recurrence and management options differ greatly between subtypes, it is important to identify rapidly and accurately the underlying pathological mechanism leading to ischaemic stroke.

Aetiological classification systems
Various classifications of ischaemic strokes have been developed.The TOAST (Trial of Org 10172 in Acute Stroke Treatment) classification [16] is the most commonly used classification system.Based on the patient's neurological symptoms, imaging of the brain using computed tomography or magnet resonance imaging (CT/MRI) and other diagnostic tests, including standard 24-hour electrocardiography (ECG), echocardiography and ultrasound of extraand intracranial arteries, the most likely aetiology is determined.
The TOAST classification divides ischaemic strokes into the following subtypes: Ischaemic stroke due to large artery atherosclerosis, cardioembolism, small vessel occlusion, ischaemic stroke of undetermined aetiology and ischaemic stroke of other determined aetiology [16].Currently, at least two newer sub-classification systems, the SSS-TOAST [17] and the A-S-C-O classification [18], are available.These newly developed classification systems are advancing the accuracy of ischaemic stroke subtype diagnosis by taking into account the level of diagnostic evidence in order to devise the "most likely mechanism" in the presence of multiple potential overlapping causes (i.e.SSS-TOAST classification) or the "phenotype", which categorises stroke patients according to a combination of their aetiological characteristics (ASCO).These advanced systems, however, are not yet widely used in clinical routine, they are more complex (though more precise) and so far no data have been published on their impact on implementation of secondary prevention and clinical management in clinical routine.Recently, a new approach [19] was suggested by scientists to how the former group of "cryptogenic strokes" should be defined and treated.Originally, patients were categorised as having "cryptogenic strokes", a subgroup within the "undetermined" group according to the TOAST criteria, if even after extensive evaluation no aetiology was determined.They proposed that most of these types of stroke are of an embolic nature, they called this embolic stroke of undetermined source (ESUS).Per definition an ESUS has to be a nonlacunar brain infarction without a proximal arterial stenosis or a major-risk cardioembolic source that would present a clear indication for anticoagulation.They furthermore recommended anticoagulation therapy in these patients and currently there are several trials ongoing, which aim to prove that these patients might benefit from the novel anticoagulants.

Ischaemic stroke due to large artery atherosclerosis
In the initial phase of atherosclerosis the endothelium is injured.This leads to a dysfunctional endothelium.More and more low density lipoproteins (LDLs) are deposited in the intima part of the artery.The accumulation of LDL induces endothelial cells to produce chemokines.These chemokines (e.g.monocyte chemotactic protein-1 [MCP-1]) attract monocytes and change them into macrophages.The macrophages enter into the intimal part of the vessel.Endothelial cells and macrophages oxidise LDL.
Macrophages which take up oxidised LDL then turn into foam cells.The foam cells increase in size, unable to metabolise the lipoproteins, and finally die.This in turn leads to inflammation of the lipid plaques.The inflammatory reaction finally leads to destabilisation of the plaque and potentially to plaque rupture, inducing an occlusion of the blood vessel further upstream, and consequently to ischaemia of brain tissue [20].Moreover the diameter of the atherosclerotic vessel locally decreases and this stenosis can then lead to hypoperfusion of the upstream cerebrovascular tissue.According to the TOAST classification, an ischaemic stroke is due to large artery atherosclerosis if either a major brain supplying artery or major cortical artery shows a >50% stenosis or occlusion and the infarction is located in the corresponding vessel territory.Other competing reasons have to be unlikely.These pathological changes (>50% stenosis) in an intra-or extracranial artery has to be clearly shown by ultrasound or imaging to consolidate the diagnosis of ischaemic stroke due to large artery atherosclerosis [16].The beginning of the pathophysiological process in ischaemic strokes due to large artery atherosclerosis is of an inflammatory nature.As a consequence, blood biomarkers indicating inflammatory processes in vessels might be promising aetiological biomarker candidates.

Ischaemic stroke due to cardioembolism
Virchow's triad, including hypercoagulability, endothelial injury and stasis of blood flow, is commonly known to lead to blood clotting.Prothrombin is activated by tissue factors and then changed into thrombin.Thrombin cuts fibrinogen into fibrin.The fibrin molecules react with each other and build cross-linked fibrin.Cross-linked fibrin then interacts with thrombocytes, which themselves adhere to the fibrin, building a thrombus.A thrombus moved by blood flow is called an embolus.In ischaemic stroke due to cardioembolism, the source of the ischaemic event is an embolic occlusion of a brain vessel and the origin of the embolus is located in the vicinity of the heart.There are mainly three mechanisms of embolism coming from the heart (fig.2): 1. local haemostasis in a cardiac chamber which leads to thrombosis (e.g.left atrial septal aneurysm and left ventricular aneurysm or, most important, atrial fibrillation); 2. material released from an abnormal cardiac surface (e.g.calcific degeneration); 3. abnormal passage between the venous and arterial system in the heart which allows venous material to end up in the arterial circulation (e.g.patent foramen ovale).Major sources for cardioembolism are atrial fibrillation, mitral stenosis, prosthetic valves, infective endocarditis, marantic endocarditis, atrial myxoma, acute myocardial infarction, left ventricular thrombosis, left ventricular aneurysm and left ventricular failure.Minor sources are patent foramen ovale, atrial septal aneurysm, atrial or ventricular septal defects, calcific aortic stenosis, mitral annular calcification and fibroelastoma [21].In the case of atrial fibrillation -the most common source of cardioembolism -supraventricular tachyarrhythmia causes unco-ordinated muscle contractions leading to a sluggish blood flow in the atrium, with the consequence that coagulation thrombi are produced, consisting of mostly fibrin and platelets [22].Ischaemic stroke due to cardioembolism usually comes with greater mortality and morbidity because of the occlusion of mainly large intracranial arteries [22].Moreover the highest recurrent rate in ischaemic stroke among all subgroups is found in cardioembolism [7].According to the TOAST criteria a cardiac source of embolism must be detected to verify whether it is an ischaemic stroke due to cardioembolism.In case there are no such findings, clinical signs and imaging of the patient are very similar to those patients suffering from large artery atherosclerosis.In that peculiar case blood biomarker may add important information to better identify a cardioembolic source.
Ischaemic stroke due to small vessel disease / small vessel occlusion Ischaemic stroke due to small vessel disease (SVD) is mainly due to arteriosclerotic SVD and sporadic and hereditary cerebral amyloid angiopathy [23].Arteriosclerotic SVD can be attributed to several pathomechanisms: fibrinoid necrosis, lipohyalinosis, microatheroma, microaneurysms and segmental arterial disorganisation.
In the case of arteriosclerosis as the cause of small vessel disease, the number of smooth muscle cells in the tunica media is reduced, the blood vessel wall thickens through deposition of fibro-hyaline material and consequently the vessel lumen decreases, leading to an undersupply to the surrounding cerebrovascular tissue.Arteriosclerotic SVD is most often associated with systemic hypertension [23].In sporadic and hereditary cerebral amyloid angiopathy, a progressive accumulation of βA4 immunoreactive amyloid proteins in the wall of small and medium arteries and arterioles can lead to occlusion of the vessel [23].The most important risk factors for a small vessel stroke are hypertension, diabetes and smoking [24].According to the TOAST criteria, a stroke is due to small vessel disease, if a cardioembolic source has been excluded and the absence of a carotid or vertebral stenosis of >50% of the vessel diameter has been shown.Occlusion of small vessels causes small infarcts, termed lacunes, most commonly located in the caudate, putamen, external capsule, internal capsule, corona radiata, pontine tegmentum and thalamus.Furthermore, the diameter of the typical subcortical infarction has to be ≤1.5 cm in diameter [16].

Ischaemic stroke due to "other" determined origins
There are several rather rare diseases, which also lead through distinct mechanisms to cerebral ischaemia (table 1).According to the TOAST criteria, they belong to the huge subgroup classified as "other determined aetiologies".According to the TOAST criteria [16], patients suffering from an ischaemic stroke in this subgroup should have a CT or MRI imaging finding of ischaemic stroke, furthermore the specific cause of the stroke should be revealed by blood tests or arteriography, and cardioembolism or large artery atherosclerosis should be excluded by diagnostic testing.The ischaemic stroke can be of any size and at any location in the brain.Some examples of this category are strokes due to systemic vasculitis as it is, for example, the case in patients with lupus erythematosus, other causes are cervical artery dissections or some underlying haematological disorders.Since this aetiological subgroup contains a plethora of different diseases, including some very rare causes, we provide the reader with only an overview of some important exponents: Undetermined ischaemic stroke Unfortunately, the aetiology of ischaemic stroke remains undetermined in up to 25-39% of patients [3], thus up to at least ¼ of ischaemic stroke patients lack potential adequate secondary prevention treatment.According to the TOAST criteria the following patient groups are classified as such when even after extensive evaluation no aetiology is determined (cryptogenic) or the cause has not been found because the evaluation was only cursory or the physician determined two or more possible causes, making a final diagnosis impossible [16].If we could reduce the proportion of undetermined strokes with the help of biomarkers, recurrence rate might be reduced substantially.

Common final ischaemic cascade
The final common pathway of all ischaemic strokes is the obstruction of arterial flow to the brain.Brain tissue depends on a constant supply of oxygen and glucose.In the case of ischaemic stroke this supply is decreased or even abolished.The impact on neurones depends on the size, location and duration of the ischaemia.After 5 to 10 minutes of local depletion of glucose and oxygen, neurones are harmed irreversibly.The inadequate supply of energy can lead to a deterioration of the ion gradient.The neuronal and non-neuronal cells depolarise.Thereby the brain cell loses potassium in exchange for sodium, chloride and calcium ions.This electrolyte imbalance leads to a water inflow into the cell and results in rapid swelling of the neurones and glia cells -cytotoxic oedema.The harmed cell now releases excitatory neurotransmitters such as glutamate.Glutamate interacts with several cell membrane proteins and, as a result of excessive cell depolarisation, many enzymes are activated, which now release oxygen free radicals.These oxygen free radicals react with intra-and extracellular proteins and thereby finally lead to cell necrosis.Oxygen free radicals may even induce programed cell death (apoptosis) in damaged cells [25].

Definition of aetiological blood biomarkers
Aetiological blood biomarkers are associated with physiological or pathophysiological processes in the human body leading to an ischaemic stroke.Ideally, they should be reliable and simple to measure; they should help scientists to gather further meaningful information about the aetiology of ischaemic stroke.Specifically, they should improve the identification of patients who need specific secondary preventive treatment and thereby allow faster initiation of treatments, ultimately leading to a better outcome.

Literature search for aetiological blood biomarkers
The markers we describe in the following paragraphs were identified by a PubMed search.We collected data from studies, written in English, on aetiological blood biomarkers in ischaemic stroke patients, listed in PubMed up to October 2014.Search terms were: stroke / ischaemic stroke, cerebrovascular disease, etiology/aetiology, blood biomarker / biomarker.Among a total of 1,268 manuscripts, we selected 117 possible relevant papers according to their heading in the title.The title of the paper had to be a relevant hint to our paper topic.After reading the abstracts of these 117 papers, we finally selected 31 manuscripts with detailed information about blood biomarkers and their significant association with a specific ischaemic stroke aetiology compared with other aetiologies.We then categorised the biomarkers by each aetiology according to the TOAST classification system.Since the field of biomarkers in the setting of stroke aetiology is rather a new site of interest for scientists, so far there are only a few reviews on this topic.Currently, an interesting, comprehensive review has been published in this field [26].

Large artery atherosclerosis (LAA) biomarkers
Table 2 summarises the characteristics of the following aetiological blood biomarkers of LAA strokes.

Decorin
Decorin is a matrix proteoglycan.It is a component of connective tissue.Decorin binds to type 1 collagen fibrils and is involved in matrix assembly.It is associated with the ability to modify plaque phenotype by reducing inflammation.Decorin reduced the activity of macrophages, the activity of gelatinase and the collagen plaque content in a mouse model [27].

CD62P / P Selectin
Cluster of differentiation (CD) 62P, or P selectin as it is called, is a cell adhesion protein found on blood vessel endothelium and on activated platelets.It plays a role in the initial recruitment of leucocytes in an inflammatory process such as atherosclerosis.
In a case control study [29], the percentage of platelets expressing CD62P in patients with ischaemic stroke of

CD63
CD63 (cluster of differentiation 63) is a cell surface glycoprotein and marker.CD63 levels are elevated in inflammatory processes.To enable platelets to aggregate, they have to be activated thus CD63 might contribute to atherothrombosis and atherothrombotic stroke [30].
In a small case control study [29] CD63 levels in patients with acute ischaemic stroke of noncardioembolic origin (n = 54, median percentage of platelets expressing CD63: 2.33%) and healthy controls (n = 28, median percentage of platelets expressing CD63: 1.15%) were measured.CD63 levels were 1.52 times higher (p <0.05) in patients with acute ischaemic stroke due to large artery atherosclerosis (n = 22, median percentage of platelets expressing CD63: 3.2%) than in cases due to small vessel disease (n = 32, median percentage of platelets expressing CD63: 2.1%) if measured 48 hours after symptom onset.Afterwards (7, 30 and 90 d after symptom onset) there was no significant difference detectable.

Soluble CD40L (cluster of differentiation 40 ligand)
The soluble CD40 ligand belongs to the tumour necrosis factor superfamily (TNFSF) family.It binds to the CD40 receptor on antigen-presenting cells.It promotes inflammatory activation on endothelial cells [31].Thus it is prominently involved in vessel inflammation [30].
In a small cohort study [32] levels of soluble CD40L were significantly higher in patients with large artery atherosclerotic stroke (n = 11, 165.01 ± 73.76 pg/ml; p <0.047) compared with patients suffering from stroke of cardioembolic origin (n = 17, 75.33 ± 30.72 pg/ml), but levels did not differ significantly between patients suffering from stroke due to large artery atherosclerosis and patients suffering from stroke due to small vessel disease (n = 13).
In another small study [31], scientists compared ischaemic stroke patients (n = 82) with normal healthy individuals (n = 20) and patients who had carotid artery stenosis without stroke (n = 17

CXCL16
CXCL16 (C-C-C motif chemokinreceptor 16), also known as SR-PSOX (scavenger receptor that binds phosphatidylserine and oxidised lipoprotein), is a chemokine and works as a scavenger receptor [33,34].In atherosclerotic plaques high levels of CXCL16 were found.It is seen to be involved in the uptake of oxidised low-density lipoprotein (LDL) in monocytes and thereby enhancing atherosclerosis [33,35,36].
In a small cohort study [37], scientists compared CXCL16 levels in patients who suffered from ischaemic stroke due to LAA (n = 86) or SVD (n = 67) and healthy controls (n = 74).Mean CXCL16 levels were higher in SVD (c = 2.13 ng/ml, p <0.05) and LAA (c = 2.36 ng/ml, p <0.01) strokes compared to healthy controls(c = 2.04 ng/ml).When mean CXCL16 levels in LAA and SVD strokes were compared, mean CXCL16 levels were 1.1 times higher in patients suffering from LAA stroke compared with SVD strokes (p <0.05).

LP PLA2
Lipoprotein-associated phospholipase A2 (LP PLA2) is an enzyme produced by inflammatory cells.It plays an important role in the development of atherosclerosis.LP PLA2 hydrolyses oxidised phospholipids, thus producing lysophosphatidylcholine and oxidised nonesterified fatty acids (OxNEFA).OxNEFA have chemotactic activity towards monocytes and lysophoshatidylcholine is able to upregulate inflammatory mediators and thereby enhance the process of evolving atherosclerosis [38].
In a cohort study [39] higher levels of LP PLA2 were associated with carotid stenosis (n = 19) as compared with patients with ischaemic stroke without showing carotid stenosis (n =111) and healthy controls (n = 20).In a large cohort study [40] (n = 1,946) LP PLA2 was found to be associated with ischaemic stroke due to large artery atherosclerosis in white non-Hispanic men (n = 381, hazard ratio 3.05).This association remained even after adjusting for confounders.In another study [41], scientists compared patients suffering from transient ischaemic attack (TIA) with healthy controls.Patients suffering from TIA due to large artery atherosclerosis had higher LP PLA2 activity (207 nmol/ml/min) compared with patients suffering from TIA due to nonlarge artery atherosclerosis (184 nmol/ml/min, p = 0.044).

LIGHT
The lymphotoxin-like inducible protein that competes with glycoprotein D for binding herpesvirus entry mediator on T lymphocytes (LIGHT) is a member of the TNFSF, the tumour necrosis factor super family.LIGHT has proinflammatory and prothrombotic effects.It is an inducer of matrix metalloproteinase (MMP) in macrophages and is involved in platelet adhesion to endothelium.When macrophages are costimulated with oxLDL (oxidised low density lipoproteins) and LIGHT at the same time, the macrophage sig-nificantly increases the uptake of neutral lipids (≈20% increase) which results in quicker building of foam cells [42] and thus atherosclerosis.
In a small case control study [43] scientists compared levels of LIGHT in patients suffering from atherosclerotic ischaemic stroke (n = 20) to patients with asymptomatic carotid stenosis (stenosis >50%, n = 19) and healthy controls (n = 23).LIGHT levels were 4.1 times higher in patients suffering from ischaemic stroke due to large artery atherosclerosis (LIGHT: 350 pg/ml, p <0.05) than in normal healthy patients (LIGHT: 85 pg/ml, p <0.05), but LIGHT levels were not significantly different compared with patients with asymptomatic carotid stenosis.

High serum total cholesterol
Cholesterol is a hydrophobic sterol molecule.It is the base substance of cell membranes and of many hormones.Particularly high levels of cholesterol in the blood circulation, depending on how they are transported within lipoproteins, are strongly associated with the progression of atherosclerosis.
In a large Japanese cohort study with 612 ischaemic stroke patients, scientists found out that high serum total cholesterol levels in Japanese men (>6.21 mmol/l) were associated with large artery atherosclerosis (multivariable hazard ratio 2.86).They did not find such differences for ischaemic stroke due to large artery atherosclerosis in Japanese women [44].C-reactive protein C-reactive protein (CRP) is an acute-phase protein produced in the liver.CRP is an inflammation marker.It is part of the innate immune response, and it is capable of several immunological actions that may contribute to the chronic inflammatory process that constitutes atherosclerosis.In the acute phase, the systemic immune-inflammatory reaction is more enhanced in cardioembolic stroke then in other stroke subtypes, which makes CRP an ambivalent marker for strokes.

Serum lipoprotein A
In a small study [46], scientists examined levels of CRP in noncardioembolic ischaemic strokes (n = 99).They studied the differences in CRP levels between patients suffering from ischaemic stroke due to large artery atherosclerosis (n = 52) and patients suffering from small vessel disease (n = 47).When CRP levels were measured on the day of onset, levels were 1.85 times higher in LAA patients compared with SVD patients (p <0.05) [46].
In another study [47], CRP levels were measured in ischaemic stroke patients (n = 600) and healthy controls (n = 600) at two different time-points, at acute phase and after 3 months.Highest levels of CRP were measured in cardioembolic strokes (n = 98) when measured at acute phase (1.5 times higher in cardioembolism than in LAA strokes) but after a 3-month follow-up the highest levels of CRP were found in the subgroup of LAA (n = 73) (1.3 times higher in LAA than in cardioembolic strokes).

Adiponectin
Adiponectin is secreted by the adipose tissue.It is involved in lipid and glucose metabolism, it stimulates glucose uptake and metabolism in muscle cells and decreases glucose production in the liver, and it is involved in energy homeostasis.Adioponectin improves the insulin resistance level of the tissue.Adiponectin might play a protective role in case of atherosclerosis [48].

Cardioembolic biomarkers
Brain natriuretic peptide Brain natriuretic peptide (BNP) is a vasoactive peptide hormone, produced in heart and brain tissue.Its function is of a natriuretic and diuretic nature.BNP is mostly secreted from myocardium in response to increased intracardiac pressure or increased wall tension [50][51][52][53].Thus it represents underlying cardiac disease, which might be associated with thrombus formation.
In a cohort study with 223 ischaemic stroke patients, a BNP cut off level of 90 pg/ml had a sensitivity 85% and a specificity of 78% to identify a cardioembolic origin, blood samples were taken >7 days after symptom onset (cardioembolic stroke: n = 69).When blood samples were taken within 24 hours after symptom onset in another cohort study (n = 707), a BNP cut-off of 76 pg/ml identified a cardioembolic stroke with a specificity of 68% and a sensitivity of 72% (cardioembolic stroke: n = 259).In a further independent cohort of 362 ischaemic stroke patients, those suffering from cardioembolism (n = 144) had the highest median BNP levels (n = 144, 179.5 pg/ml) compared with the median level in other subtypes (57.2 pg/ml) (undetermined n = 144, LAA n = 76, SVD n = 61, other n = 16).Similar results were found in a study [58] with 142 ischaemic stroke patients.In this study it was possible to distinguish cardioembolicfrom noncardioembolic stroke with a cut-off value of 66.50 pg/ml, with a sensitivity of 75.0% and a specificity of 88.7%.

Review article: Current opinion
Finally, in a recent small cohort study [59]  In a relatively small study [61] (n = 93 patients, 66 with ischaemic stroke [28 due to cardioembolism]) NTproBNP levels were 3.9 times (p <0.001) higher in patients suffering from ischaemic stroke due to cardioembolism (491.6 pg/ ml) than noncardioembolic ischaemic stroke (124.7 pg/ml) when measured <72 hours after symptom onset.In this study the cut-off value was 265 pg/ml NTproBNP.It was possible to distinguish cardioembolism from other subtypes with a sensitivity of 94.4%, a specificity of 72.9%, and a positive predictive value of 56.6% and a negative predictive value of 97.2%.Interestingly, NTproBNP levels were not associated with the size of stroke.
In a slightly larger cohort study [62] (n = 125 patients, 57 patients with stroke of cardioembolic origin), a NT-proBNP level ≥342 pg/ml, was associated with a cardioembolic stroke with a sensitivity of 93% and a specifity of 75%.
In another study [63] scientists tried to indentify the cause of ischaemic strokes in patients with cryptogenic ischaemic stroke.Among patients with cryptogenic strokes NT-proBNP levels above 360 pg/mL were attributed to an underlying cardioembolic source of the stoke with a sensitivity of 87% and a specificity of 83%.BNP and NTproBNP seem to be quite reliable markers for cardioembolism in ischaemic stroke, measured early after stroke onset.All studies showed that BNP or NTproBNP levels in ischaemic stroke of cardioembolic origin are significantly higher than in ischaemic stroke due of noncardioembolic origin.Ten studies were found showing that BNP and NTproBNP are able to distinguish accurately a cardioembolic stroke from other subtypes.However no direct comparison between BNP and NTproBNP were made, and thus so far we are not able to state which marker might perform better.
A list of the above mentioned studies which contain cutoff values and AUC (area under the curve) (eight out of ten studies) for either BNP or NTproBNP is presented in table 3: Midregional proatrial natriuretic peptide Midregional proatrial natriuretic peptide (MRproANP) is a stable fragment of the precursor hormone of ANP.It is produced in an equimolar amount to ANP.ANP has a natriuretic, diuretic and vasodilating function and thereby is able to regulate blood pressure.It is mainly produced by cardiomyocytes of the atrium and secreted during increased atrial wall tension [64,65].The midregional fragment of proANP is more stable than the N-or C-terminal part of the precursor in vivo and in blood ex vivo, which might render it generally more applicable to clinical practice then other fragments of natriuretic peptides [66].
In a prospective observational study [67], MRproANP levels were 1.66 times higher (p <0.0001) in patients with ischaemic stroke due to cardioemolism (n = 131, 206 pmol/ l) compared with patients with noncardioembolic strokes (n = 228, 124 pmol/l).If MRproANP levels were >180 pg/ml the ischaemic stroke was most probably due to cardioembolism (specificity 60.3%, sensitivity 71%, positive likelihood ratio 2.0, negative likelihood ratio 0.6).This association was independent of known risk factors for cardioembolic strokes such as known atrial fibrillation or congestive heart failure.

D-dimer
D-dimer is a fraction of fibrin, produced when plasmin is activated.It is a marker of haemostatic imbalance.D-dimers are a product of the degradation of fibrin by plasmin, thus they are involved in thrombus formation.In cases of ischaemic stroke due to cardioembolism, thrombi mainly consist of fibrin and platelets [22].It follows that the degradation of the thrombus is associated with a higher systemic D-dimer concentration.
Several studies [55,68,69] showed that D-dimer levels are significantly higher in ischaemic strokes of cardioembolic origin than in those of noncardioembolic origin.
In a relatively small cohort study [68] (patients n = 98, patients with ischaemic stroke of cardioembolic origin n = 53), D-dimer levels were 1.9 times higher in cardioembolic strokes compared with other subtypes of stroke.When Ddimers were measured within 6 hours after symptom onset, a level below 300 µg/l ruled out a stroke due to cardioembolism (sensitivity 100%, specificy 53%).

Soluble thrombomodulin
Soluble thrombomodulin (sTM), or CD141 antigen as it is called, is an endothelial-specific type tumour necrosis factor-alpha (TNF-α), interleukin-6 (IL-6) and interleukin-1beta (IL-1β) All of these three molecules are inflammatory markers.Inflammation plays a role in thrombus formation or degradation.An alternative association with cardioembolic stroke might be explained by the fact that the systemic immuneinflammatory reaction is more pronounced in cardioembolic stroke then in other stroke subtypes.
Aetiological blood biomarkers of cardioembolic strokes are listed in table 4.

Small vessel disease biomarkers
Aetiological blood biomarkers of SVD stroke are listed in table 5.

Homocysteine
Homocysteine is an amino acid that is produced via demethylation of methionine [73].Higher levels of homocysteine can lead to endothelial cell injury [74].Evidence suggests that homocysteine may further promote fatty deposits in blood vessels by damaging the inner lining of arteries and promoting blood clots.
In a case control study [75] with 172 patients suffering from SVD and 172 healthy controls, mean homocysteine levels were 1.21 times higher in patients with ischaemic stroke due to SVC compared with healthy controls (p <0.0005).However the study did not compare homocysteine levels with other subtypes such as LAA or CE.

Platelet CD40 ligand
After platelet activation, platelet CD40 ligand (CD40L) is presented on the cell surface.It then binds CD40 on endothelial cells and thereby induces the production of adhesion molecules, inflammatory cytokines and procoagulant tissue factor [76], all factors with a causal relevance to atherogenesis, most probably also in small vessels [30].

Intercellular adhesion molecule 1
Intercellular adhesion molecule 1 (ICAM1) is a glycoprotein expressed on the surface of endothelial cells, macrophages and lymphocytes.ICAM1 is induced by IL-1 and TNF.ICAM 1 is involved in adhesion and transendothelial migration of lymphocytes [77], which may contribute to small vessel atherosclerosis.
In a cohort study [78], scientists followed up 267 healthy individuals and compared baseline ICAM1 levels in subjects who developed white matter lesions on MRI with subjects who did not develop white matter lesions after 3 and 6 years.Patients who showed MRI findings of white matter lesions after 3 years (n = 23, mean ICAM1 levels: 372.6 ng/ml) had 1.19 times higher ICAM1 levels in their blood compared with patients who had no such findings (n = 219, mean ICAM1 level: 312.8 ng/ml; p = 0.005).
When comparing MRI findings and ICAM1 levels after 6 years, ICAM1 levels in patients with MRI findings (white matter lesions, n = 32, mean ICAM1 levels: 347.4 ng/ml) were still 1.12 times higher in these patients than in healthy subjects (n = 152, mean ICAM1 levels: 309.0 ng/ml).

Conclusion
The use of rapidly measurable blood biomarkers for aetiological assessment on admission may advance the identification of patients who need specific secondary preventive treatment such as oral anticoagulation, and thus accelerate the onset of optimal secondary prevention, improve patient outcome and reduce subsequent costs.However, as the research interest in blood biomarkers in the field of stroke aetiology has evolved only over the last few years, most of the above data on aetiological markers reside still in an exploratory phase.The studies are rather small and only few are controlled for potential confounding factors.A new aetiological blood biomarker, before it can be used in clinical routine, generally needs to provide additional information to established clinical features.But most of those studies which have been published so far were very small, did not include all statistical analyses that have been newly proposed to assess the additive discriminatory value of blood markers in vascular research, and they mostly involved post-hoc analyses.There is clearly a need for validation of these results in larger well-designed studies.Only few markers were evaluated and validated in rather large, welldesigned studies, and the most promising potential was shown for the family of the natriuretic peptides in identifying cardioembolic strokes.
To find the ideal marker, the pathological mechanism underlying the stroke should be well understood.In the case of cardioembolic strokes, biomarkers that have shown associations with cardiac dysfunction, specifically atrial fibrillation, are the most promising.In this case, natriuretic peptides show a great deal of potential for aetiological relevance.In terms of large artery atherosclerosis, inflammatory markers, especially those known to be associated with carotid atherosclerosis, intima media thickness or intracerebral atherosclerosis, are interesting candidates.LP PLA2 seems thus to be an interesting candidate marker, it is involved in the process of atherosclerosis via hydroxylation of oxidised phospholipids and thereby starts the process of attraction of inflammatory cells on the atherosclerotic site.
Currently no marker has yet been identified specifically for small vessel disease in a large enough stroke cohort, thus for this aetiological subtype of stroke, further derivation studies are needed before a marker for further validation can be recommended.For example, markers reflecting a state of chronic high blood pressure or markers related to amyloid deposition in small cerebral vessels would be promising candidates for further evaluation.However, since there are overlapping pathophysiological mechanisms for each aetiology, such as inflammation and thus in-flammatory markers, it is most likely that panels of blood biomarkers will be used in the future since no single marker will be sensitive and specific enough.Newer studies should provide the answer to the question as to whether some aetiological biomarkers will help to improve identification of patients for known effective secondary prevention treatments after acute ischaemic stroke, and if patients with abnormal levels of the selected biomarkers are at increased risk for recurrent events as a hard and clinically relevant endpoint.If one or more of the selected biomarkers are confirmed to be associated with stroke aetiology and recurrent stroke, interventional studies may be undertaken to apply in these patients secondary preventive therapies guided by clinical information and biomarker

Figures (large format)
Examples of stroke aetiologies (imaging material adapted from Servier Medical Art).

1 ,
200 patients, 73 LAA Cohort study APC = antigen presenting cell; CD = cluster of differentiation; CXCL16 = C-C-C motif chemokinreceptor 16; LIGHT = lymphotoxin-like inducible protein that competes with glycoprotein D for binding herpesvirus entry mediator on T lymphocytes; LP PLA2 = lipoprotein-associated phospholipase A2; oxLDL = oxidised low density lipoprotein; TIA = transient ischaemic attack Serum lipoprotein A is a lipoprotein consisting of apolipoprotein A and apolipoprotein B-100.It belongs to the group of LDL (low density lipoprotein).Serum lipoprotein A might be involved in the process of atherosclerosis.In a large cohort study[45], serum lipoprotein A levels were measured in patients suffering from ischaemic stroke or TIA (n = 1,012).Highest levels of serum lipoprotein A were found in patients suffering from LAA (n= 281, median c = 34.6 mg/dl) (CE: n = 204, median c = 29.2mg/dl, SVD: n = 261, median c = 24.2mg/dl, other determined: n = 74, median c = 24.1,undetermined: n = 192, median c = 27 mg/dl)

Figure 2
Figure 2Most important mechanisms for the formation of cardiac embolism (imaging material adapted from Servier Medical Art).PFO = patent foramen ovale

Table 3 :
Table of BNP/NTproBNP studies.n = 54, median percentage of platelets expressing P selectin: 5.4%) were compared with healthy controls (n = 28, median number of platelets expressing P selectin: 3.77%).They then compared patients with ischaemic stroke due to large artery atherosclerosis with patients suffering from ischaemic stroke due to small vessel disease.P selectin was measured at three different time-points (<48 h, 7 d, 30 d, 90 d after symptom onset).
atherosclerosis (8.41 ± 2.89 ng/ml) or small vessel disease (7.41 ± 3.49 ng/ml), compared with healthy controls (0.56 ± 0.40 ng/ml) and compared with patients with carotid stenosis without stroke (3.71 ± 1.19 ng/ml).In the same study, the mean fluorescence intensity of CD40L expression on peripheral blood monocytes was significantly high-er in patients scientists compared BNP levels among aetiological subtypes in ischaemic stroke patients (n = 131).Patients suffering from cardioembolism could be identified with a cut-off BNP value of 77 pg/ml, with a sensitivity of 75.8% and a specificity of 76.8%, compared with all other aetiologies.
1 membrane receptor.When bound to thrombin protein C is activated.Protein

Table 5 :
Aetiological blood biomarkers of small vessel disease (SVD) stroke.Swiss Med Wkly.2015;145:w14138 Swiss Medical Weekly • PDF of the online version • www.smw.chlevels.These markers might in the future even serve as surrogate endpoint for new secondary prevention measures.To specifically address the question on aetiological biomarkes in acute stroke, the BIOmarker SIGNAture of Stroke AetioLogy Study (BIOSIGNAL) -a multicenter international study funded by the Swiss National Science Foundation (PZ00P3_142422) -was designed and has begun enrolling patients across Europe.PD Dr. med.Mira Katan Kahles, MSc, Department of Neurology, University Hospital of Zurich, Frauenklinikstrasse 26, CH-8091 Zürich, Switzerland, mira.katan[at]usz.ch Correspondence: