Studying stroke thrombus composition after thrombectomy: what can we learn?

Abstract

The composition of ischemic stroke thrombi has gained an increasing amount of interest in recent years. The implementation of endovascular procedures in standard stroke care has granted researchers the unique opportunity to examine patient thrombus material. Increasing evidence indicates that stroke thrombi are complex and heterogenous, consisting of various biochemical (e.g. fibrin, von Willebrand factor and neutrophil extracellular traps) and cellular (e.g. red blood cells, platelets, leukocytes and bacteria) components. This complex composition may explain therapeutic limitations and also offer novel insights in several aspects of stroke management. Better understanding of thrombus characteristics could therefore potentially lead to improvements in the management of stroke patients. In this review, we provide a comprehensive overview of the lessons learned by examining stroke thrombus composition after endovascular thrombectomy and its potential relevance for thrombectomy success rates, thrombolysis, clinical outcomes, stroke etiology, and radiological imaging.

Introduction

Thrombectomy has in recent years dramatically changed acute ischemic stroke care, following several successful thrombectomy trials in 2015.^1–5 Besides the enormous clinical impact, endovascular procedures have also instigated a novel subfield in stroke research. By mechanically, and usually en bloc, removing the occluding thrombus from the patient vasculature, endovascular thrombectomy is providing the opportunity to collect thrombus material for research purposes.^6,7 Better understanding of thrombus composition may help to overcome the current limitations of both pharmacological and mechanical revascularization therapies. As stroke thrombus material is increasingly available, a growing number of studies is revealing the multifaceted composition of endovascularly-retrieved cerebral thrombo-emboli. Whereas the first reports mainly focused on the presence of red blood cells (RBCs), fibrin and platelets^6,7, subsequent research showed that also other components contribute to the complexity of ischemic stroke thrombi, including leukocytes, von Willebrand factor, neutrophil extracellular traps (NETs) and extracellular DNA.^8–11 The variable composition of such thrombi may present obstacles for recanalization therapies but also new opportunities in the management of ischemic stroke patients. In this review, we cover the most important associations between thrombus composition and endovascular treatment parameters (such as thrombectomy recanalization rates), effect of pharmacological thrombolysis, clinical outcome, stroke etiology, and radiological imaging. This review is based on as good as all reports (until December 2020) that used patient stroke thrombi to study these associations. For a more elaborate and in-depth description of the composition and internal architecture of ischemic stroke thrombi we refer to recently published reviews.^12,13

Thrombus composition and endovascular procedural success

The main goal of endovascular treatment is to establish recanalization of the affected blood vessel by removing the occluding thrombus, which can be achieved by a stent retriever, by aspiration or by a combination of both techniques. Several factors are known to influence thrombectomy success rates, including thrombus location, size and vascular access.^7,14 First-pass complete reperfusion has in recent years become the preferred goal in endovascular therapy since the number of thrombectomy attempts needed to achieve good recanalization is inversely correlated with clinical outcome.^15–18 Such first-pass effect is not always achieved and multiple attempts, often using various devices, are required in 60–75% of patients to achieve complete recanalization.^15,16 In 10–20% of the patients, the attempts remain futile due to failure to remove the thrombus and establish reperfusion.¹⁴ Procedural success rates are likely to be influenced by thrombus characteristics such as stiffness, stickiness, deformability and mechanical friction, all of which may be defined by thrombus composition. Various studies examined whether and how thrombus composition affects thrombus removability and thrombectomy success rates. A summary of these studies is presented in Table 1 and online Supplemental Table I and II (online supplement, please see https://www.ahajournals.org/journal/str).

Table 1:

Main associations between thrombus composition and endovascular/thrombolytic treatment success, functional outcome, stroke etiology and radiological imaging

Histological parameter	Endovascular treatment success	Thrombolytic treatment success	Stroke etiology	Radiological imaging	Stroke severity/Functional outcome
RBCs, fibrin and platelets	RBC-thrombi are more easy to retrieve19,20 and have improved recanalization outcomes19,39,40 compared to fibrin/platelet-rich thrombi19,20,39–41 RBC- rich thrombi are prone to preprocedural thrombus migration24,25 Formation of secondary embolisms is more present in RBC-rich thrombi47 versus secondary embolisms are more present in fibrin/platelet-rich thrombi42 Studies showing no association32–38,42,80	RBC-rich areas are more susceptible to thrombolysis compared to platelet-rich areas37 Thrombi contain a dense outer shell of platelets that is resistant to fibrinolysis57	CE thrombi = higher RBC and lower fibrin/platelet content 34,36,39,73–76 versus LAA thrombi = higher RBC and lower fibrin/platelet content20,38,77–79,81 Cryptogenic thrombi resemble CE38,42,77,79 versus LAA thrombi39 LAA thrombi = Inner RBC core with platelets on the surface vs. CE thrombi = platelets interspersed with RBCs throughout the entire thrombus.38,87 No association19,33,44,80,86	A HAS is associated with higher RBC and lower fibrin/platelet content while the absence is associated with a lower RBC and higher fibrin/platelet content. 20,33,34,36,37,39,42,74,79,92 Increased thrombus perviousness is associated with RBC-rich thrombi102 versus perviousness is associated with fibrin/platelet-rich thrombi103	Higher NIHSS score at admission = Higher RBC content37, higher platelet or fibrin bundle content62 Higher NIHSS score 7 days post admission = Higher polyhedral RBC, platelet or fibrin bundle content62 Favorable clinical outcome = Higher RBC content20 Worse clinical outcome = higher polyhedral RBC content62 No association34,35,41
VWF	Unknown	Higher VWF content = increased rt-PA resistance8	No association8,43	Unknown	Higher NIHSS score at admission = Higher VWF content43 No association8
Leukocytes	Higher leukocyte content = improved recanalization outcomes27 Higher leukocyte content = more passes required27 Higher neutrophil content = more secondary embolisms47 No association exists32–34,38,43,78	Unknown	CE thrombi = more leukocytes than LAA thrombi27,77,78 No association19,33 CE thrombi are associated with a higher neutrophil74 and lower T-cell82 content No association (T/B-cells, eosinophils, monocyte/macrophages, neutrophils)10,44,78	Unknown	Higher NIHSS score at admission/discharge = higher leukocyte content27 or monocytes43 No association34,41
NETs	Higher NET content = increasing amount of thrombectomy attempts11,44 and a worse recanalization outcome44	Higher NET content = increasing rt-PA resistance10,11	CE thrombi = Higher NET content10,44,74 and higher overall DNA content81 than non-CE thrombi No association11	Unknown	NETS are associated with a worse NIHSS score at discharge and a worse mRS score.44
Bacteria	Presence of bacteria = more thrombectomy attempts46	Unknown	Presence of bacteria = underlying infectious pathology (e.g. infective endocarditis)46,90	Unknown	Unknown

CE: Cardioembolic, HAS: Hyperdense artery sign, LAA: Large artery atherosclerosis, mRS: modified Ranking scale, NETs: Neutrophil extracellular traps, NIHSS: National Institute of Health Stroke Scale, RBCs: Red blood cells, rt-PA: Recombinant tissue plasminogen activator

Emerging evidence indicates that the amount of RBCs is an important determinant of thrombus removability as RBC-rich thrombi require lower amounts of passes to establish recanalization.^19,20 Interestingly, Duffy et al. showed that the composition of thrombus material retrieved in the first two thrombectomy attempts contains significantly more RBCs and less fibrin compared to thrombus material retrieved in subsequent attempts, indicating that RBC-rich thrombus material is easier to remove than fibrin-rich thrombi.¹⁹ Several factors could contribute to this observation. RBC-rich thrombi have a lower coefficient of friction compared to fibrin-rich thrombi.^19,21 Higher amounts of RBCs also reduce thrombus stiffness and are associated with better stent strut integration into the thrombus and also probably lead to better conformability into an aspiration catheter in case such a technique is applied.^22,23 Of note, such increased deformability and reduced friction potentially also explains why RBC-dominant thrombi are more prone to pre-interventional thrombus migration^24,25. This phenomenon, in which the entire thrombus migrates more distally in the intracranial blood vessel before intervention, was reported to reduce subsequent thrombectomy success rates.²⁶

Other thrombus components may also influence thrombus removability. Boeckh-Behrens et al. showed that higher amounts of leukocytes in the thrombus have a tendency to require more passes to be needed for removal of the thrombus.²⁷ Along the same lines, the presence of neutrophil-derived NETs, was shown to be associated with a higher number of thrombectomy attempts to remove the thrombus.¹¹ NETs components such as DNA and histones can modify the structure of fibrin and render it resistant to mechanical deformation, which could account for the thrombectomy resistance.²⁸ Similarly, bacteria increase thrombus stiffness by altering the fibrin microstructure²⁹, which could explain why the presence of bacteria negatively affects septic thrombus removability.³⁰ Components of the vascular wall in a thrombus, such as collagen, have also been shown to reduce thrombus removability, although such presence may also be attributed to vascular injury induced by multiple passes required to remove a difficult thrombus.³¹

Besides the number of thrombectomy attempts to achieve recanalization, the degree of reperfusion as indicated by the modified Thrombolysis in Cerebral Infarction (mTICI) scale is also a key indicator of procedural success. Various studies focused on the association between the composition of the thrombus and the mTICI score. Whereas some studies were unable to show such correlation^24,32–38, an increasing number of studies indicate that RBC-rich thrombi are associated with better recanalization outcomes (mTICI >2b) in comparison to fibrin/platelet-rich thrombi.^{19,31,39–41} Similarly, the presence of leukocytes and NETs were not associated with the mTICI score in various reports^{11,27,32–34,38,42,43}, although some studies show that leukocytes, in particular neutrophil elastase positive cells and NETs, contribute to lower rates of complete recanalization (mTICI score < 2b).^26,44,45 In line with the need for multiple passes to remove them, both septic emboli and those containing vascular wall components are associated with worse mTICI scores.^31,46

An important unwanted aspect of thrombectomy procedures is thrombus fragmentation, leading to a shower of small thrombo-emboli that may travel more distally in the brain where mechanical removal is impossible or at least very difficult and risky. Fragmentation is likely influenced by the biochemical and cellular make-up of the thrombus as this can greatly influence the mechanical properties. To date, reports on thrombus composition and thrombus fragmentation are scarce. Low RBC and high fibrin content⁴², and higher amounts of neutrophil elastase-positive cells⁴⁷ have been suggested to be associated with the occurrence of secondary emboli.

In sum, it is clear that the composition of thrombi can influence endovascular procedural success. The growing body of literature indicates that RBC-rich thrombi are more easily retrieved which translates into better recanalization and clinical outcomes compared to fibrin/platelet-rich thrombi. Leukocytes, perhaps specifically NETs, may also influence success rates although more studies are needed to fully confirm this idea. New insights on thrombus characteristics and their impact on thrombus retrieval can help in the development of improved thrombectomy protocols and adapted device technology, such as stent retriever designs for easy to retrieve but fragment-prone thrombi⁴⁸ or to improve retrieval of difficult fibrin/platelet-dominant thrombi.⁴⁹

Thrombus composition and thrombolysis

At present, rt-PA is the only FDA-approved thrombolytic drug to pharmacologically dissolve the thrombotic cerebral occlusion. Use of rt-PA is, however, limited to less than 15% of patients due to the short therapeutic time window of 4.5 hours after stroke onset.⁵⁰ In addition, recanalization after rt-PA is only successful in less than half of patients with a proximal artery occlusion.⁵¹ The reasons for the latter are not well understood but it seems that thrombus length plays a major role as thrombi >8 mm respond badly or not at all to intravenous thrombolysis (IVT).^52,53 Recent evidence indicates that thrombolysis reduces the size of a thrombus retrieved by thrombectomy, but this effect is not associated with recanalization outcome.⁵⁴ Most likely, also thrombus composition influences the response to IVT and studies examining retrieved stroke thrombus material might shed some light on this so-called rt-PA resistance (summarized in Table 1 and online Supplemental Table III; online supplement, please see https://www.ahajournals.org/journal/str). The thrombolytic mechanism of rt-PA is based on the activation of plasminogen into plasmin, which degrades fibrin in the thrombus. Fibrin is an important constituent of RBC-rich as well as platelet-rich stroke thrombus material, but platelet-dominant thrombi have other specific structural features that could impair rt-PA mediated fibrinolysis. Indeed, whereas RBC-rich material mainly consists of RBCs and fibrin, platelet-dominant thrombus regions also contain various other extracellular scaffold molecules such as dense fibrin, von Willebrand factor (VWF), extracellular DNA and NETs.^8–11,43 Such non-fibrin components may contribute to rt-PA resistance by providing additional mechanical stabilization of the thrombus, by altering the structure of fibrin or by decreasing the thrombus permeability^8,10,11, which is in line with the observation that RBC-dominant thrombi are more efficiently dissolved by rt-PA than platelet-rich thrombi.^8,37,55,56 Of note, Di Meglio et al. recently described a fibrinolysis-resistant outer thrombus shell composed of platelets, VWF and extracellular DNA, forming a barrier that hampers rt-PA-mediated thrombolysis.⁵⁷ Interestingly, this shell also contained inhibitors of fibrinolysis, such as plasminogen-activator inhibitor 1 (PAI-1).⁵⁸ Thrombus contraction, a common phase of thrombus formation mediated by contractile forces of platelets on fibrin, might also influence thrombolytic success.⁵⁹ Thrombus contraction facilitates the redistribution of platelets and RBCs into separate areas and mediates the compression of RBCs into tightly-packed polyhedrocytes^59,60, which can reduce thrombus permeability and thus the degree of thrombolysis.⁶¹ Intravital thrombus contraction was recently demonstrated in stroke thrombi, resulting in a compact structure with a limited porosity.⁶²

Taken together, insights on thrombus composition and architecture may reveal novel therapeutic avenues that can lead to improved thrombolysis. Future pharmacological treatment could include the VWF-degrading substances ADAMTS13 (a disintegrin and metalloprotease with thrombospondin type-1 repeats, member 13) and N-acetylcysteine, the DNA-cleaving enzyme DNAse1 or inhibitors of PAI-1, all of which show promising results in pre-clinical studies.^{8,10,11,63–65} Novel fibrin-targeting fibrinolytics could further add to the efficacy and safety of thrombolysis, such as Tenecteplase, which has a higher specificity, longer half-life and improved resistance to endogenous inhibitors compared to alteplase.⁶⁶

Thrombus composition and stroke etiology

Due to the mainly embolic nature of acute ischemic stroke, thrombo-emboli can originate from different locations in the body.⁶⁷ The original hemodynamic conditions in which thrombus formation took place, such as blood flow rate, shear stress, turbulence and vasculature most likely influence the composition of the thrombus and thus the embolus causing the ischemic stroke.⁶⁸ Arterial conditions with high shear stress are typically associated with platelet-rich thrombi, whereas venous, low shear stress conditions are assumed to promote the development of more coagulation-driven thrombi. Stroke etiology is classified according to the TOAST criteria (Trial of ORG 10172 in Acute Stroke Treatment), identifying the origin as cardioembolic, large artery atherosclerotic (LAA), small vessel occlusion, other (e.g. carotid dissection or paradoxical embolisms) or a cryptogenic origin.⁶⁹ To prevent recurrent stroke, good knowledge of the underlying risk factors and potential pathogenesis is crucial in patient follow-up and treatment. Recurrent strokes still occur in approximately 25% of all stroke patients within 5 years after the initial event, highlighting the need for improved primary and secondary prevention.⁷⁰ Cardioembolic strokes are commonly caused by atrial fibrillation and are mainly treated using anticoagulants, while large artery atherosclerotic strokes are mostly treated using antiplatelet agents (e.g. aspirin).⁷¹ Cryptogenic strokes, which comprise approximately a third of all ischemic strokes, pose a significant problem as the appropriate secondary prevention strategy is difficult to select in the absence of a known underlying pathogenesis.

To better understand the variable stroke thrombus pathogenesis, numerous studies have investigated the link between the histological composition of retrieved thrombi and the origin of the thrombus.⁷² A summary of these studies is shown in Table 1 and online Supplemental Table IV (online supplement, please see https://www.ahajournals.org/journal/str).

The majority of reports mainly focused on the quantity of RBCs and fibrin and show inconsistent results. Whereas several studies found that cardioembolic thrombi are characterized by higher amounts of RBCs and lower amounts of fibrin compared to LAA thrombi^{34,36,39,73–76}, other studies reported the opposite^{20,38,77–79} or found no association at all.^19,33,44,80 Forming the main target in antiplatelet therapy, platelets are an important factor in thrombosis and are thought to play a particular role in high-shear conditions. Two studies found that cardioembolic thrombi contain higher amounts of platelets in comparison to LAA thrombi.^38,76 whereas the opposite was reported by others.^32,81 Various studies showed no association between etiology and platelet content.^34,44,73,79 Similarly, VWF has been shown to be present in all thrombi regardless of their origin, with amounts ranging from 0.1% to 95%.^8,41,43 Initial reports, using low sample sizes, showed no link between VWF content and stroke etiology.^8,43 Data on leukocytes also remain inconsistent. Various reports indicate that leukocyte content is not related to stroke etiology^{19,20,33,34,44,78}, whereas several other studies did find an association between higher leukocyte content and cardioembolic origin.^27,77,78 Different leukocyte subtypes, including neutrophils, eosinophils, monocytes/macrophages, T-cells and B-cells, have also been linked with stroke etiology, but the overall findings remain fragmentary and not conclusive at this moment (Table 1 and online Supplemental Table IV; online supplement, please see https://www.ahajournals.org/journal/str).^{10,43,44,74,78,82} NETs or extracellular DNA seems to be particularly present in cardioembolic thrombi.^10,44,74,81 While it is common to see calcium deposition in many LAA lesions using imaging modalities, radiological studies indicate that only 1.3% of thrombi are calcified.^83,84 At this point, only one study, using a specific histological staining method, has evaluated the presence of calcifications in a limited subset of stroke thrombi⁸⁵, highlighting the need for additional large scale studies to evaluate this aspect.

Besides the quantitative determination of thrombus components, various studies have also evaluated the internal architecture of thrombi from different etiologies. The presence of serpentine and layered fibrin was not linked with etiology^80,86, but LAA thrombi were found to more frequently consist of an inner RBC-rich core surrounded by platelets along the thrombus surface, in contrast to cardioembolic thrombi in which platelets were typically found to be interspersed with RBCs.^38,87

Taken together, various studies attempted to link thrombus composition and stroke etiology, but the overall outcome remains largely inconclusive. The reported inconsistencies are most likely related to the low sample sizes used in the majority of studies, as underlined by a recent meta-analysis.⁶ Nevertheless, two of the largest patient thrombus cohorts, indicate that cardioembolic thrombi contain higher amounts of fibrin/platelet aggregates and lower amounts of RBCs compared to other etiological subtypes.^77,78 Based on the histological analysis, these large studies suggest that the majority of cryptogenic thrombi most likely originate from a cardioembolic etiology.^77,78 Additional large-scale studies will be needed to further clarify if and how thrombus composition, organization and structure can reveal information on stroke etiology and guide treatment using anticoagulant therapy, antiplatelet therapy or other strategies to prevent secondary events.^7,88 Another uncertainty in this context is that it is currently unknown if the composition of the original thrombus, mostly located in the heart or the carotid bifurcation, is reflected in the composition of the emboli found in the brain vasculature. For instance, no evidence is available whether certain parts of the parental thrombus are more prone to embolize, highlighting the potential differences between the parental and embolized thrombi. Stroke etiology has typically been classified according to the TOAST criteria. Currently, newer classification methods are available such as the ASCOD criteria that assign a degree of likelihood to a patient-specific etiological classification.⁸⁹ Potentially, such improved etiology classifications will further strengthen the link between thrombus characteristics and stroke etiology. Finally, histological thrombus analysis might also be used to identify less common etiologies such as septic emboli and atrial myxomas. Using Gram-staining, two studies revealed the presence of Gram-positive bacteria in thrombi from patients suffering from infective endocarditis or other infectious diseases.^46,90 Since diagnosis of stroke due to an infectious disease is often not straightforward, early identification of a septic embolus might help to initiate early antibiotic treatment.

Prediction of thrombus composition by radiological imaging

Computed tomography (CT) and magnetic resonance imaging (MRI) are the primary imaging modalities used to exclude cerebral hemorrhage, to assess the extent of infarction and the at-risk penumbra, to grade the collateral circulation and to identify the location of the arterial occlusion. Apart from identifying thrombus location and its size, imaging also has the potential to allow early characterization of thrombus composition and permeability, which could guide procedural decisions such as selection of thrombus-specific retrieval protocols or device technologies. A summary of studies addressing the link between thrombus composition and radiological imaging is given in Table 1 and online Supplemental Table V (online supplement, please see https://www.ahajournals.org/journal/str).

At the site of the cerebral occlusion a hyperdense artery sign (HAS) and a susceptible vessel sign (SVS) is detected in approximately 50% of ischemic stroke patients using CT or MRI, respectively.⁹¹ The vast majority of studies indicate that both the presence and the density, typically measured in Hounsfield units, of HAS on CT is associated with RBC-dominant thrombi, while the absence of this radiological sign is indicative of fibrin/platelet-rich thrombi.^{20,27,92,33,34,36,37,39,42,74,79} The correlation of HAS and SVS with the presence of RBCs can be explained by the concentration of hemoglobin in the thrombus.^93–96 Current imaging modalities, however, can only discriminate between RBC-dominant and fibrin/platelet-dominant thrombi and are unable to accurately identify mixed thrombi. Brinjikji et al. recently demonstrated in vitro that dual energy CT can be used to improve the characterization of thrombus composition, but this remains to be evaluated in patients.⁹⁷ Apart from CT and MRI-based imaging, intravascular optical coherence tomography (OCT) can also be used to determine thrombus composition. Intravascular OCT uses a fiber optic wire that both emits and records the reflection of light while simultaneously being rotated and pulled back from the artery, giving rise to an image by measuring the backscattering of light from the vessel wall and thrombus.⁹⁸ While currently in use to evaluate the morphology of coronary plaques during coronary endovascular interventions, recent in vitro studies showed that OCT can be used to discriminate between RBC-dominant, fibrin-dominant and mixed blood clots as well.^99,100

Apart from identifying thrombus composition, radiological imaging can also be used to assess thrombus permeability, also termed ‘thrombus perviousness’. Thrombus perviousness is defined as the degree in which a contrast agent is able to flow through the structure of the thrombus and is measured by comparing thrombus attenuation on non-contrast CT with that on CT angiography, thereby giving an idea about residual blood flow through the thrombus.¹⁰¹ An increase in thrombus attenuation between the two respective imaging techniques implies a higher thrombus perviousness as contrast media enters the thrombus.¹⁰¹ Higher thrombus perviousness is associated with better functional outcome, smaller infarct volumes and improved recanalization outcomes with both thrombolytic and endovascular therapy.¹⁰¹ Indeed, pervious thrombi have a porous structure that allows passage of residual arterial flow or thrombolytics. Current reports with regard to the histology of the thrombus and thrombus perviousness are conflicting.^102,103 Benson et al. have shown that RBC-dominant, fibrin/platelet-poor thrombi are associated with more permeable thrombi on CT imaging.¹⁰² On the other hand, Berndt et al. have shown that higher amounts of fibrin/platelet conglomerates and lower amounts of RBCs are associated with more permeable thrombi.¹⁰³ As described earlier, the degree of thrombus contraction, a process that is dependent on platelets, most likely contributes to permeability and differences in contraction might explain the conflicting results.

In the future, it will be interesting to further establish the link between radiological signs and thrombus composition and to use such insights to develop pre-treatment decision-making strategies to increase first pass recanalization success rates.

Thrombus composition, stroke severity and functional outcome

In the past decade, several studies have identified various clinical, interventional and blood parameters as independent predictors of clinical outcome. Some examples of these parameters include a higher National Institute of Health Stroke scale (NIHSS) score at admission¹⁰⁴, diabetes mellitus¹⁰⁵, multiple thrombectomy attempts^15,16,104, higher neutrophil counts¹⁰⁵, higher neutrophil/lymphocyte ratio¹⁰⁶ and a higher VWF/ADAMTS13 ratio¹⁰⁷. The question whether the composition of the occluding thrombus is directly associated with functional outcome has also been addressed. Various studies have attempted to correlate thrombus composition with functional parameters such as stroke severity (NIHSS) and clinical outcome (modified Rankin Score (mRS)) (Table 1 and online Supplemental Table VI; online supplement, please see https://www.ahajournals.org/journal/str). Whereas not all studies could confirm strong associations^{8,34,41,11,35,43,57}, some interesting correlations have been reported. For example, stroke severity, based on admission NIHSS scores, is positively correlated with the content of RBC, platelets, fibrin, VWF and monocytes in the occluding thrombus.^37,43,62 In particular, polyhedral RBCs, a morphological marker of clot contraction, were associated with more severe strokes, most likely because high amounts of polyhedrocytes render thrombi more compact, less deformable and less porous.⁶² Thrombus composition was also found to be associated with stroke outcome. Apart from lower amounts of RBCs²⁰, especially higher amounts of leukocytes and NETs and in the thrombus have been linked with a poor outcome.^27,44,62

Limitations and future perspectives

Since the arrival of endovascular thrombectomy, stroke thrombi have been collected and studied. It has become clear that stroke thrombi are complex and heterogenous, consisting of various cellular and molecular components that affect endovascular/thrombolytic success rates and that are associated with stroke etiology and radiological signs. Current imaging techniques can to some extent be used to characterize the thrombus prior to therapy, providing an early sense of how the thrombus will respond. Increased knowledge on thrombus composition has instigated refined treatment strategies to improve thrombectomy first-pass recanalization rates and to increase the efficiency and safety of pharmacological thrombolysis. Yet, research on thrombi retrieved via thrombectomy is not without limitations, which should be considered. First, only thrombi from large vessel occlusions that did not dissolve spontaneously or after infusion of rt-PA and that can be successfully retrieved via mechanical thrombectomy, are available for study. Thus, a selection bias exists, excluding rt-PA-susceptible or thrombectomy-resistant thrombi. The improvement in radiological characterization of thrombi could potentially be used in the future as a surrogate to estimate the composition of these inaccessible thrombi.⁹⁷ Second, thrombus characteristics could be influenced by patient-specific variables such as pre-stroke anti-thrombotic treatment, pharmacological thrombolysis or the technique of mechanical thrombectomy itself. Little information is currently available on these aspects and should be addressed in future studies.

While early studies may have been limited by the low sample sizes, they provided proof-of-concept for thrombus-driven stroke research and laid the foundation for larger scale studies. Various national and international initiatives have in the meantime established large-scale thrombus registries, such as the EXCELLENT (NCT03685578)¹⁰⁸, STRIP^32,41 and the THRAPS (MR CLEAN) registries. It will be interesting to see how results from these large studies will further our understanding of ischemic stroke thrombi and potentially inspire novel ideas for optimized stroke treatment.

Non-standard Abbreviations and Acronyms

ADAMTS13

A disintegrin and metalloprotease with thrombospondin type-1 repeats, member 13)

CD31

Cluster of differentiation 31

CE

Cardioembolic

CT

Computed tomography

HAS

Hyperdense artery sign

IVT

Intravenous thrombolysis

LAA

Large artery atherosclerosis

MRI

Magnetic resonance imaging

mRS

Modified Ranking scale

mTICI

Modified Thrombolysis in Cerebral Infarction

NETs

Neutrophil extracellular traps

NIHSS

National Institute of Health Stroke Scale

OCT

Optical coherence tomography

PAI-1

Plasminogen activator inhibitor-1

RBC

Red blood cell

rt-PA

Recombinant tissue plasminogen activator

SVS

Susceptible vessel sign

TOAST

Trial of ORG 10172 in Acute Stroke Treatment

Data availability

The authors declare that all supporting data are available within this review manuscript and in the supplemental files.