Correlation of imaging and histopathology of thrombi in acute ischemic stroke with etiology and outcome

Abstract

Mechanical thrombectomy has become the stand of care for patients with large vessel occlusions, yet major improvements in thrombectomy speed, efficacy, and completeness can still be achieved. High rates of clot fragmentation and failure to remove the clot resulting in poor neurological outcomes suggest that in order to further advance the field of stroke intervention; we must turn our attention towards understanding the science of clot. Accurately identifying the composition of the occlusive clot prior to intervention could significantly influence the success of the revascularization strategy used to treat them. Numerous features of thromboemboli could be studied and characterized, including quantitative histomorphometry and diagnostic imaging characteristics. Each of these features might logically predict superior thrombectomy outcomes with one device or another. This article aims to review the current literature on histopathological composition of acute ischemic stroke clots, with a particular focus on the correlation between clot composition and diagnostic imaging, stroke etiology and revascularization outcomes.

Introduction

Mechanical thrombectomy has become the standard of care for many patients suffering from acute ischemic stroke following the publication of multiple randomized controlled trials that have demonstrated that endovascular revascularization, as compared to intravenous (IV) tissue plasminogen activator (t-PA) treatment alone, results in significantly higher rates of good neurological outcome and functional independence at 90 days [1–5]. More recently, the results of two further trials called DEFUSE 3 and DAWN were published, demonstrating the benefit of endovascular treatment in selected patients up to 16 and 24 hours, respectively, after time last seen well [2, 6]. When engaging a clot with a stent-retriever or suction catheter, optimal outcomes are likely best achieved when the entirety of the thrombus is able to be retrieved in one pass. Multiple passes indicate that the clot is adherent to the vessel or fragmented which could put the patient at risk for distal emboli and less than optimal recanalization outcomes [7]. It has been shown that the histological content of the clot influences the mechanical characteristics of thrombi and thus affect the ability of a stent-retriever device or suction catheter to engage it [8].

The advent of mechanical thrombectomy devices has created the unique opportunity to study acute ischemic stroke clot material. Accurate interpretation of the histologically stained slides has remained the foundation of pathological analysis and diagnostic medicine for over a century and studying the histomorphology of retrieved clots is crucial to improving our understanding of the variations in clot compositions of various etiologies. Cardioembolic clots often occur due to a new onset of a heart arrhythmia, typically atrial fibrillation and large artery clots tend to result from the embolization of atherosclerotic plaque from a large artery source, commonly the carotid arteries. The histopathologic signatures of the clots formed as a result of these very different underlying conditions should be recognizable and identifiable, both to histopathologists and also to sophisticated image analysis software programs. Numerous features of thromboemboli can be studied and characterized, including quantitative histomorphometry and diagnostic imaging characteristics [8–12] and each of these features might logically predict superior thrombectomy outcomes with one device or another. This article aims to review the current literature on histopathological clot composition, with a particular focus on the correlation between clot composition and imaging, stroke etiology and revascularization outcomes.

Histopathologic Analysis

Over the past several years, there has been an increase in the number of studies characterizing clots retrieved from patients with acute ischemic stroke [13–24]. However, in general, there is a lack of uniformity in the histopathologic characterization of retrieved thrombi [9]. H&E is the most widely used histology stain and is considered the Gold-standard for the diagnosis of many diseases including cancer and to date most studies analyzing clot composition have used the classical Hematoxylin and Eosin (H&E) stain to characterize AIS clot composition [12, 25, 26]. The major advantage of the H&E stain is that it is a relatively simple stain and results in dissimilar colours of cell nuclei and cytoplasm, allowing easy recognition of cells and cell populations. However, the major limitations of the H&E stain are that the cytoplasmic differentiation is insufficient, reticular fibres, basement membranes and cell borders are not stained and the contrast between cytoplasmic and other extracellular structures is poor. This means that studies that have used H&E for the characterization of AIS clots typically only identify three components of clots (Red Blood Cells, White Blood cells and Fibrin/Other) and have tended to classify the clots into sub-groups such as Red-Blood Cell-Rich, Fibrin-Rich and Mixed [24]. However, this fails to take into consideration other key factors in the coagulation cascade such as Platelets and von Willebrand Factor that have also been shown to be present at high levels in clots. Several more recent studies have acknowledged this and now refer to this sub-group as Fibrin/Platelet-Rich [27, 28]; however, the classification of clot composition using only H&E staining is still inadequate.

Masson’s trichrome is a three-colour staining protocol that has occasionally been used for the identification of AIS clots components [29, 30]. Masson’s trichrome is best suited to distinguishing cells from surrounding connective tissue and many of the components that it specifically identifies such as muscle fibres and bone are not commonly found in AIS clots. Hence it has more usefulness in the characterization of atherosclerotic plaque and could potentially be incorporated in the histological characterization of AIS clots when a Large Artery aetiology is suspected. Additionally, several studies have used other histological stains to study clot composition, such as Elastica von Gieson (Connective tissue), Prussian Blue (Iron) and Gomori trichrome stain (Muscle), but have not reported any significant correlations with imaging, etiology or outcome using these stains [13, 21, 31].

Martius Scarlett Blue (MSB) stain is another commonly used histological stain that has been used for the characterization of clot composition previously [10, 30, 32]. MSB staining allows for the identification of Red Blood cells (Yellow), Fibrin (Red), White Blood Cells (Purple/blue) and Collagen (Blue). The major advantage of the MSB stain is that the distinctive colour separation is much better than H&E and therefore enables more accurate quantification of clot components, as can be seen in Figure 1. The MSB stain also identifies more components that the Traditional H&E stain, with the identification of Collagen and consequently MSB seems to be the optimal histology stain for the identification of the major components of AIS Clots.

Figure 1: Commonly used histological stain in acute ischemic stroke clot assessment.

A&E are gross photographs of clots removed from two different patients that had suffered an acute ischemic stroke with an area of suspected calcification in E highlighted with a red box. B&F are examples of H&E stained slides from the corresponding clots showing Red Blood Cells (Red), White Blood Cells (Purple) and Fibrin/Platelets (Pink). C&G are examples of MSB stained slides showing Red Blood Cells (Yellow), White Blood Cells (Purple), Fibrin (Red) and Platelets/Other (Grey). D&H are examples of Masson’s Trichrome stained slide showing Red Blood Cells (Red), White Blood Cells (Purple) and Fibrin/Platelets (Purple). I is an example of a von Kossa stained slide confirming the presence of calcification (Brown).

The presence of calcification in AIS clots has been reported previously [30] and cases in which calcification is present are likely due to the embolization of atherosclerotic plaque from a large artery source [33, 34]. Calcified cerebral emboli are frequently overlooked or misinterpreted [35]. They are composed of large amounts of calcium phosphate which influences their mechanical properties and therefore, render them stiffer and less accessible for stent retrievers [33]. Areas of suspected calcification can often be identified using H&E, Masson’s Trichrome and MSB, however, the von Kossa stain is used to specifically identify mineralization in tissue and should be used to confirm the presence of calcification when suspected, as demonstrated in Figure 1. To date, no studies have shown a correlation between the presence of calcification as identified by von Kossa staining and large artery aetiology.

The advantages and disadvantages of histological stains commonly used to characterize the composition of stroke thrombi are summarized in Table 1.

Table 1:

Summary of the major advantages and disadvantages of histological staining techniques commonly used in the characterization of acute ischemic stroke clots.

Histological Stains:	Advantages:	Disadvantages:
Hematoxylin and Eosin	Widely used Cheap and relative easy to run Suitable for automatic staining machines	Can only identify 3 major constituents (RBC’s, WBCs and Fibrin/Other) Insufficient discrimination between fibrin and platelets
Martius Scarlett Blue	Has good color differentiation making characterization more accurate Can Identify platelets-rich regions as distinct from fibrin	More complex histological stain than H&E requiring skilled histopathologists The differentiation step is critical to the outcome of the staining
Masson’s Trichrome	Can Identify Collagen, Muscle and Bone	The color discrimination between the major components of the clot is not sufficient
Von Kossa	Can identify the presence of calcification	No other major clot components can be identified

Immunohistochemical Analysis

In addition to the use of basic histological stains, many studies are now using immunohistochemistry to identify specific components of AIS clots. Platelets play a key role in response to endothelial injury as platelet plug formation is associated with activation of the coagulation cascade and resultant fibrin deposition. Platelet-rich clots have long been known to be more resistant to thrombolytic therapy [36, 37] and platelets have been shown to be present at similar levels to RBCs and Fibrin in AIS clots [14]. As shown in Table 2, various antibodies have been used to investigate platelet composition in clots. Von Willebrand factor (vWF), a large multimeric plasma glycoprotein, plays a major role in blood coagulation and it is important in platelet signaling promoting their adhesion to vascular injured sites. Importantly, vWF can be cleaved by the metalloprotease ADAMTS13 into smaller and less reactive multimers leading to thrombus dissolution. Neutrophil extracellular traps (NETs) promote thrombus formation generating a scaffold for platelets and RBCs thereby influencing the coagulation cascade [38]. NETs are fibrous networks of extracellular DNA released by neutrophils under the form of de-condensed chromatin associated with histones and neutrophil granule proteins such as myeloperoxidase and neutrophil elastase. Antibodies against myeloperoxidase (MPO) and citrullinated histones H3/H4 as well as neutrophils markers (CD66b and neutrophil elastase) have been used to assess thrombus composition [23, 38].

Table 2.

Summary of the main antibodies previously used for the assessment of acute ischemic stroke clot composition.

Target:	Antibody Used:	References:
Platelets	GPIIb/IIIa (CD61), GPIb	[19, 41, 42]
	CD31 (PECAM-1)	[43, 44]
	CD42b	[32]
Von Willebrand Factor	vWF	[39, 45]
Fibrin	FibII	[19]
Endothelial Cells	CD34	[30, 42]
Neutrophils	Neutrophil Elastase	[38, 40]
	MPO	[23]
	CD66b	[38]
NETs	H3Cit	[35]
	Histon H4	[23]
T-Cells	CD3	[22, 46]
B-Cells	CD20	[22]
Macrophages	CD68	[22, 39]

Immune cells are known to be involved in thromboinflammation during stroke, including CD3+/CD4+ T cells and CD68+ monocytes [39] and it has been suggested that the degree of infiltration by inflammatory cells may impact the mechanical properties of the thrombus including its stability and degradation [40]. The interaction of immune cells with platelets (e.g., via cluster of differentiation 40 (CD40) and CD40 ligand or P-selectin (CD62-P) and P-selectin glycoprotein ligand 1 generating platelet-leukocytes complexes) and with endothelial cells (e.g., via intercellular adhesion molecule 1 and lymphocyte function-associated antigen 1) may play a role in thrombus formation [39]. Early endothelialization may occur over and within the thrombus affecting the clot dissolution by t-PA. Although endothelial cells covering the surface of thrombi is part of their normal evolution, t-PA will not penetrate the thrombus depending on the degree of endothelialization [30]. Therefore, endothelialization and calcification of clots are among the factors that may reduce t-PA efficacy.

Quantification Methods

Quantification methods of clot histologic characteristics are highly variable, with many studies relying on visual inspection rather than more robust techniques such as computer-aided quantification [9]. Whilst manual interpretation of histological and immunohistochemical stained slides remains the cornerstone of diagnosis in many diseases such as cancer, manual quantitative of composition has high inter-observer variability and low reproducibility [47]. The field of digital pathology has been around for some time and many image analysis software packages have been used with some success to quantify the components of clots from scanned H&E stained slides including ImageJ, Aperio and Adobe [24, 27, 28]. We are now entering the era of artificial intelligence and we believe that the use of Machine Learning will become commonplace in the analysis and quantification of histological and immunohistochemically stained images. Machine-learning based Image analysis software packages such as QuPath and Orbit Image Analysis have been used previously and allow for quick and accurate quantification of tissue components by using automated segmentation algorithms combined with trainable cell/tissue classification [48, 49]. The use of Machine-learning techniques will undoubtedly increase the accuracy and reproducibility of quantitative histopathology which will be crucial to determining the cellular composition of acute ischemic stroke clot components.

Clot Composition and Stroke Etiology

Initial studies examining correlations between stroke etiology and clot composition determined using H&E staining were conflicting, with studies showing no correlation, a correlation between RBC-Rich emboli and cardioembolic stroke and conversely a correlation between RBC-Rich emboli and large artery atherosclerosis [12–14, 16–19, 21, 25, 42]. These studies had relatively small cohorts of patients ranging from 17–54. A previous systematic review article on these studies by the authors found that there was no significant difference in the proportion of RBC-Rich thrombi between cardioembolic and large artery atherosclerosis etiologies [50]. However, a recently published study by Sporns et al that included 187 patients again suggested that Cardioembolic thrombi had significantly fewer erythrocytes and higher proportions of fibrin/platelets than non-cardioembolic thrombi [22], a finding supported by another more recent study [24]. Boeckh-Behrens et al performed a quantitative analysis of 145 H&E stained thrombi collected from stroke patients with large-vessel occlusion and found that cryptogenic strokes strongly overlap with cardioembolic strokes but not with non-cardioembolic strokes in terms of thrombus histology as well as interventional and clinical outcome parameters [51]. Interestingly a recent study has shown that a higher percentage of white blood cells (WBCs) in the thrombus was associated with cardioembolic etiology and hypothesize that WBC-mediated immunological and coagulatory processes may play a key role in thrombus formation and pathogenesis of stroke [12].

The amount of Platelets does not appear to be related to the aetiology as they appear to be present at similar levels in both cardioembolic and large-artery atherosclerotic patients [14, 19, 52]. However the distribution of platelets within the clot might be varied; Ahn et al observed that in arteriogenic clots platelets covered the fibrin layers or were localized at the edge or periphery of the clot whereas in cardiogenic thrombi, fibrin was most abundant and platelets were clustered within the fibrin-rich regions [32]. vWF-positive areas co-localize with regions of fibrin and collagen and inversely correlate with red blood cell content, thus vWF and platelets are suspected of being major components of ‘white’ clots [53, 54]. Schuhmann et al revealed that the number of CD4+ and CD68+ cells was increased in erythrocytic and red clots compared to white thrombi rich in vWF+ cells compared and mixed clots. However, the mechanisms by which the immune cells contribute to the pathogenesis of stroke are not completely understood [39]. It has also been demonstrated that neutrophils are abundantly present in thrombi of all pathogenesis [38], however, there are again conflicting studies on correlations between the number of NETs in AIS thrombi and stroke etiology. Laridan et al found a significant correlation between a higher amount of NETs in thrombi from cardiac origin compared with non-cardiac thrombi, whereas Ducroux et al showed that NETs are important constituents of thrombi irrespectively of their aetiology [23, 38].

The inconsistencies in the reported findings to date are likely due to differing interpretations and implementation of the TOAST criteria, diverse quantification methods and to the aforementioned limitations of the H&E stain. The authors suggest that studies including larger patient cohorts, the use of a more comprehensive histological stain such as the MSB stain and standardized quantification techniques are required before definitive correlations between histological composition and stroke aetiology can be confirmed.

Clot Composition and Outcome

As highlighted earlier, AIS clot histological compositions are diverse often as a result of their differing etiologies. It is therefore logical to assume that such diversity in composition will have a significant impact on revascularization outcome; yet, the neurovascular community still lacks a complete understanding of the distribution of clot phenotypes and their putative association with revascularization outcomes. Red Blood Cell-rich clots are associated with significantly higher recanalization rates, reduced number of maneuvers and a shorter mean recanalization time than fibrin-rich clots. This is likely due to the fact that Fibrin-rich clots have a significantly higher coefficient of friction than Red Blood Cell-rich clots and therefore have a stronger interaction with the vessel wall and are harder to remove from the vessel wall [55]. Additionally, the viscoelastic properties of clots change according to their composition, with Red Blood Cell-rich clots being more viscous, as water is the main constituent of RBCs, and fibrin-rich clots being more elastic [56]. Calcified thromboemboli are occasionally encountered (1% of patients) and calcified specimens are stiffer than arteriogenic and cardiogenic emboli, resulting in poorer revascularization outcomes [8, 33]. The amount of WBCs has also been shown to influence the success of the procedure [12]. A higher percentage of WBCs in the thrombus has been shown to be associated with an extended mechanical recanalization time, a less favorable recanalization rate (TICI score) and a poorer clinical outcome (NIHSS post treatment) suggesting that leukocytes are a notable factor in thrombus formation and development of acute cerebral ischemia [13].

The platelet cell adhesion molecule CD31 (PECAM-1) is a receptor protein expressed by multiple cell types involved in coagulation and immunological processes and is thought to have a neuroprotective effect. Boeckh-Behrens et al demonstrated a significant correlation between CD31+ cells and an improved outcome in 86 patients treated with mechanically thrombectomy [44]. The platelet-to-lymphocyte ratio (PLR) was introduced as a potential marker to determine increased inflammation and a high-PLR ratio is associated with a poorer recanalization rate and a larger infarcted area [57]. Similarly, high levels of vWF and low ADAMTS13 levels are correlated with an increased risk of stroke and worse outcome. Consequently it has been suggested that pre-treatment with anti-platelet therapy or targeting vWF directly by ADAMTS13 action or by inhibition of the vWF-platelet glycoprotein Ib (vWF-GPIb) interaction may become promising thrombolytic strategies and potentially improve the odds of a successful recanalization in patients treated with MT [58]. High amounts of neutrophil elastase-positive cells are also related to an increased risk of periprocedural thrombus fragmentation [40]. It is obvious that the composition of AIS clots affects the success of revascularization strategies and therefore in order to increase the rate of successful revascularization, we must first understand more about the histological composition of clots from different etiologies.

Imaging Analysis and Clot Characterization

Diagnostic Imaging, including non-contrast computed tomography (NCCT) computed tomography angiography (CTA) and magnetic resonance imaging (MRI), of AIS has evolved from a diagnostic tool to a quantitative and qualitative assessment tool that can help to predictive response to t-PA and mechanical thrombectomy. The presence of a Hyperdense Artery Sign (HAS) on non-contrast CT has been shown to indicate a Red Blood Cell-rich phenotype [25, 59] and is associated with improved recanalization rates [25, 50, 60–67]. Niesten et al performed CD31 immunohistochemistry and found a non-significant negative correlation between CT attenuation and the proportion of platelets [23].

A Susceptibility Vessel Sign (SVS) on MRI corresponds to a localized hypo intense signal at the site of the thrombus and is also related to the amount of red blood cells within the thrombus [25, 29]. There appears to be a consensus in the literature that the presences of an SVS sign is associated with an improved functional outcome but not an improved revascularization outcome [68–70]; Bourcier et al assessed patients from the Contact Aspiration vs Stent Retriever for Successful Revascularization (ASTER) and THRombectomie des Artères CErebrales (THRACE) trials and found that SVS was associated with lower disability at 3 months but mTICI scores did not differ between groups [68]. The proportion of platelets was significantly higher in clots with a negative susceptibility vessel sign compared to those with a positive susceptibility vessel sign. SVS identification is reliable and reproducible, however a major limitation is that the prevalence of the SVS+ varies significantly among MRI machines [71].

In addition to NCCT, most patients also undergo a CTA as part of their diagnostic imaging. The rate of uptake of the contrast agent by the clot, known as the permeability of the clot, can be quantified by the level of contrast penetration and has been shown to be related to the composition of the clot [72]. Thrombus Attenuation Increase from NCCT to CTA has previously been shown to be associated with improved functional outcomes in patients treated with rtPA and patients treated using mechanical thrombectomy [73, 74]. A recent study has suggested that permeable thrombi correlate with lower fractions of red blood cells counts and more fibrin/platelets conglomerations, concurrent with an association with cardioembolic origin [82]. The assessment of perviousness measures is easy to implement during admission imaging of patients with stroke that consists of non-contrast CT (NCCT) and CTA imaging and has the potential to give additional information about the occluding thrombus.

Future Directions

Proteomics promises to be a useful tool to identify blood-based biomarkers in acute ischemic stroke [75–77]. Targeted proteomics chips examining certain proteins have been found to predict incident ischemic stroke in two independent Swedish cohorts of adults aged over 70 years[78]. Some groups have used a brain proteomics approach to detect specific patterns of protein expression for different areas of brain tissue affected by stroke (core, penumbra and non-ischemic areas) [79].

The specific molecular composition of the retrieved clots is still largely unknown and therefore, clots retrieved by thrombectomy represent a resourceful material for proteomic characterization that may allow a better understanding of the molecular mechanisms of thrombus formation. Two recent studies have reported the use of mass spectrometry on frozen samples in the attempt not only to establish the peptide composition of emboli but also to discover potential biomarkers for stroke etiology [80, 81]. Rao et al performed mass spectrometry analysis of 20 thrombi retrieved from patients with different stroke etiologies in an attempt to identify specific peptides in clot composition corresponding to the tissue of origin. The study detected 81 common proteins in all 20 samples and that proteins associated with inflammation were found in emboli[80]. Munoz et al defined a reference proteome of 339 proteins detected in four studied thrombi. Further studies with larger cohorts of patients are needed and extensive validation must be carried out in order to implement these findings into the clinical practice. However, the preliminary results are encouraging and the availability of preserved clinical samples suggests that the use of mass spectrometry to identify potentially novel biomarkers of stroke subtype should be considered in the future. The ultimate utility of such analyses will be to provide insight into stroke etiology, particular in the group of patients with stroke of undetermined source.

Future Directions

Notwithstanding the many potential benefits of mechanical thrombectomy, major improvements in thrombectomy speed, efficacy, and completeness must be achieved before the full benefit of the procedure is realized by patients. Revascularization outcome is critical to achieving good neurological outcomes [82–84], however, data from the randomized controlled trials published in 2015 show that the rate of TICI 2b/3 recanalization is highly variable ranging, from 59%−88%, with most studies showing rates lower than 80% [85–89]. High rates of clot fragmentation and failure to remove the clot resulting in poor neurological outcomes suggest that accurately identifying the composition of the occlusive clot prior to intervention could significantly influence the success of the thrombectomy procedure. In order to further advance the field of stroke intervention, we must turn our attention towards the science of clot, by gaining a greater understanding of the distribution of clot phenotypes and their association with recanalization outcomes, we can arrive at a better understanding of why it is that successful recanalization is only achieved in 70–80% of cases and develop technologies and techniques that can be used to successfully retrieve these difficult clots.

Figure 2: Histologic and Immunohistochemical Analysis of an acute ischemic stroke clot.

(A) is a gross photograph a ‘White’ clot retrieved from a patient using a mechanical thrombectomy procedure. (B) is an example of an MSB stained slide from the same clot showing the presence of Red Blood Cells (Yellow), White Blood Cells (Purple), Fibrin (Red) and Platelets/Other (Grey). (C) is an example of an immunohistochemically stained slide demonstrating the presence of Platelets in the clot (CD42b=Red). (D) is an example of an immunohistochemically stained slide demonstrating the presence of von Willebrand factor (Red).