A high resolution scanning electron microscopy analysis of intracranial thrombi embedded along the stent retrievers

Endovascular treatment with stent retriever thrombectomy is a major advancement in the standard of care in acute ischemic stroke (AIS). The modalities through which thrombi embed along stent retriever following mechanical thrombectomy (MTB) have not yet been elucidated. Using scanning electron microscopy (SEM), we analyzed the appearance of thrombi retrieved by MTB from AIS patients, when embedded into the stent retriever. We observed that the organization and structural compactness vary for compositionally different thrombi. The modalities of attachment onto the stent vary according to thrombus composition and organization.


Red blood cells rich thrombus.
are examples illustrating an RBCs rich thrombus incorporated into the stent. As thrombus spirals along the stent, its segments are anchored on single or multiple stent struts at a time- Fig. 1a,b. The RBCs rich thrombus segments display a compact core of polyhedrally shaped red blood cells and an outer layer formed of fibrin, characteristic for cerebral arterial thrombi 23,24 - Fig. 1c,d and Supplementary Fig. S1. The polyhedral shape of red blood cells is acquired due to compressive forces in vivo, during thrombus formation, and is considered a marker for intravital thrombus contraction [25][26][27] . The porosity of thrombus increases towards its periphery where platelets and white blood cells are also present along with fibrin. The segments of thrombus are interlinked by fibrin strings- Fig. 1e,f. The various modalities through which thrombus is incorporated into the stent are illustrated in Fig. 2. Protrusion of the stent through the thrombus (Fig. 2a,b; also Supplementary Fig. S2) occurs at sites with loose cellular packing, where the red blood cells are biconcave in shape. The thrombus can deform, wetting the stent surface- Fig. 2c,d. We also observed, between the double struts of the stent mesh, the existence of films, or bridges, of fibrin, with or without cellular content- Fig. 2e,f. It is unlikely that such fibrin bridges are native to the original thrombus that caused the stroke. Most probably, the bridges across the stent struts are formed during the retrieval process, and they potentially aid in securing the thrombus attachment. Table 1. Characteristics of retrieved thrombi. *Thrombi were situated within segments of middle cerebral artery (MCA), for all the investigated cases, and retrieved with one pass MTB. **TREVO Striker, Kalamazoo, Michigan, USA; Solitaire Medtronic, Minneapolis, USA; Catch-Mini, Balt, Montmorency, France. ***According to Thrombolysis in Cerebral Infarction (TICI) scale, used for grading the result of recanalization therapy. ****RBCs rich: red blood cells rich core wrapped in a thin outer layer of fibrin. Intermediate: regions of red blood cells agglomerates are relatively scarce and encased in large regions of compact fibrin and platelets. Fibrin rich: thrombus composed mainly of fibrin, with overall sheet like aspect. *****Protrusion: stent struts are penetrating through segments of thrombus. Wetting: Thrombus conforms with the surface of the stent strut. Trapped: thrombus trapped between adjacent stent struts. Wrapping: thrombus material is wrapped around stent struts. Bridges: fibrin, interlaced with various amounts of cellular material, in the form of a film anchored on adjacent stent struts. ******Fragments of sheet like structures at thrombus surface, indicative for possible vascular tissue removal during thrombectomy. *******Vascular occlusion at a distal MCA segment following thrombus fragmentation and migration of thrombus fragments.      www.nature.com/scientificreports/ Figure 6a depicts fragments of the same thrombus after being sectioned along perpendicular directions. The fibrin is the main and ubiquitous structural element, while red blood cells content is scarce and non-uniformly distributed. The larger thrombus section reveals folding of a dense fibrin sheet (100-200 µm), with fibrin fibers being visible in higher magnification micrographs. The cross section indicates that the fibrin fibers are organized in bundles with a preferred parallel orientation to each other and perpendicular to the stent direction ( Fig. 6a,b). White cells are incorporated into the dense fibrin sheet, with an estimate density of 2-to-4 cells per 1000 µm 3 , or roughly 15% of the volume, while the fibrin fibers occupy an estimated 70-80% of the local volume. The folding of fibrin sheet forms a cavity, which was inconspicuous prior to sectioning. Infrequent clusters of biconcave red blood cells and white blood cells are scattered through the cavity (Fig. 6c). A closer view within the cavity and examination of the inner walls also shows that fibrin fibers aggregate in large bundles which are cross linked and aligned in parallel to each other (Fig. 6d). After fragmenting the thrombus, we had a closer look at the part of thrombus originally wrapped around the stent strut-the smaller fragments in Fig. 6a. We observed that loosely packed microregions of fibrin with no cellular content (Fig. 6e) alternate with microregions of fibrin mesh encasing polyhedral red blood cells and white cells (Fig. 6f). The estimated cellular content, based on microscopy images in the latter case, is 10-20% of the local volume. Most importantly, in the region of thrombus wrapped around the stent strut the porosity is considerably higher, compared with the part distant from stent.
We summarize our findings for the fibrin rich thrombus as follows. The fibrin rich thrombus displays a sheet like morphology with two distinct types of structural organization. One, which constitutes most of thrombus volume, is compact and textured, as it consists in fibrin bundles connected to each other and uniformly aligned, in aggregates with high aspect ratio (100-200 µm thick, several hundreds of microns wide). In the direction of fibrin bundles, the bending angle of thrombus sheet is large, which suggests resistance to deformation. In other directions than along the fibrin bundles, the fibrin sheet proves more flexible, as undergoes bending at smaller angles and torsional deformation. The other type of structural organization consists in porous and randomly oriented micro-regions, tens of micrometers in size, of fibrin, with and without cellular content. The overall porosity, the randomness, and the size of the heterogeneous microregions allow deformation in multiple directions and thrombus wrapping around the stent strut. www.nature.com/scientificreports/

Discussion
The RBCs rich thrombus incorporates multiple segments which at the core are compact and consist in aggregates of red blood cells shaped as polyhedrocites. The compact aggregates evade anchoring onto the stent. In contrast, peripheral to the compact core and interconnecting thrombus segments, there are loosely packed volume regions of biconcave red blood cells and fibrin. The loosely packed regions of RBCs rich thrombus are the parts engaged in attachment onto the stent struts: they allow the struts to pass through, they are deformable and show surface affinity for the stent material. The intermediate thrombi, although they are compact, can deform. In this case, the extent of grain boundaries between thrombus components, the RBCs aggregates and the fine fibrin/platelets matrix, can be a contributor to thrombus behavior. The outer layers of fibrin strings can aid, in case of intermediate thrombi, the adhesive attachment to the stent. The incorporation of the fibrin sheet thrombus into the stent is non-adhesive and relies on foldability, or in other words on the ability of fibrin to wrap around the stent strut. We show that variations in composition and compactness accompany the modalities of thrombus attachment onto the stent. The polyhedrocites core in RBCs rich thrombi, which is also known as a marker for intravital contraction, is of particular importance for response to treatment in AIS. When present, compact red blood cells regions make thrombi less susceptible to external fibrinolysis 28 . In addition, from the point of view of thrombus mechanical properties, intravital contraction is a contributor to stiffness 17 . Our study hints that the polyhedrocites core does not engage in thrombus attachment onto the stent. We found that RBCs rich thrombi embed onto the account of non-compact regions. Intermediate compositions, in terms of RBCs and fibrin content, lead to overall more compact but also deformable thrombi. Nevertheless, regardless of thrombus composition, the extent of non-compact volume regions, the ability to deform, and the affinity for the stent surface are the features that favor thrombus embedding into the stent.
The advantages of endovascular treatment are beyond doubt, as MTB improves recanalization rates in AIS patients and decreases associated morbidity and mortality 29,30 . It is becoming recognized that the use of stent retrievers is associated with vascular injury [31][32][33][34] . It is therefore not surprising to find in our study remnants of vascular tissue attached to the extracted thrombi.
A limitation of our study is the heterogeneous collection of stent retrievers used for MTB. This is partially due to the limited access to retrieved thrombi shortly after MTB, and limited possibilities in selecting samples according to the stent type. Most of the retrieved thrombi detach from the stent retriever shortly after intervention. It is also recognized that thrombi undergo structural changes, during and after retrieval process with MTB 35 . Also, thrombi are sometimes only partially removed from the occluded arteries. Even within these limitations, examining how thrombi embed along the stent retrievers offers an understanding of the underlying structural features which are responsible for thrombus capture into the stent. Standard procedures are needed, indeed, as  Table 1). (a) Optical micrograph. (b) Low magnification SEM view. (c) Closer SEM view at the thrombus-stent interface (the stent strut was cut, to allow better viewing). There is no adhesion between the thrombus and the stent strut (arrow).  www.nature.com/scientificreports/ should link clinical neuroimaging with the biomechanical factors governing thrombi incorporation into the stent retriever, and nevertheless with thrombi structural organization. Microstructural examination of human thrombi can help, in perspective, fill the existing knowledge gaps.

Conclusions
The incorporation of thrombi along the stent retrievers engages thrombus regions that are deformable on the account of enhanced porosity or adequate composition. The capture can be adhesive, when thrombus is wetting the stent, and non-adhesive, when thrombus folds around the stent struts without making a close contact. In RBCs rich thrombi, stent struts can protrude through non-compact volume regions. The extent of compact and non-compact regions in intracranial thrombi, the composition of thrombus, and the adhesive affinity for the stent surface are important features for thrombus capture into the stent.

Methods
The followed procedures were in accordance with the ethical standards of the responsible committee on human experimentation (institutional and national) and with the Endovascular technique. The thrombi were retrieved from patients suspected of suffering an acute ischemic stroke (AIS), who received whole brain stroke CT protocol 5 and were referred for MTB intervention. Thrombectomy procedures were performed under general anesthesia using a bi-plane C-arm (Allura Clarity FD20, Philips Healthcare, Best, the Netherlands) via a common femoral artery approach. A guiding catheter was placed in the concerned artery at neck level. A large-bore aspiration-catheter was advanced over a microcatheter and a microwire up to the thrombus, while the microcatheter was advanced beyond the thrombus. Subsequently, a stent retriever (Table 1) was advanced through the microcatheter and unsheathed across the thrombus. Hence, the stent retriever was gently retrieved inside the large bore aspiration catheter while a negative pressure was applied through it by using a vacuum system. Sample preparation and microscopy technique. Upon retrieval, thrombi integrated onto the stent were first immersed in formalin (4%), and immediately after were transferred in glutaraldehyde (2.5%), where they were kept overnight at 4 °C. Subsequently, samples were washed in phosphate buffer solution (PBS) 10X three times for 20 min each, dehydrated in solutions of ascending concentrations of ethanol (50, 60, 70, 80, 90, 100%) for 15 min each time, and dried using critical point drying. The samples were mounted on SEM stubs using carbon tape and carbon paint and sputtered with a 5 nm AuPd (80%/20%) coating. The microscopy observations were performed with an ultra-high-resolution field-emission Zeiss Merlin SEM, equipped with Smart-SEM 6.06 service pack 6 software and a Gemini II column, using the Everhart-Thornley secondary electron detector, 5 kV acceleration voltage and 500 pA probe current.

Data availability
The data that support the findings of this study are available on request from the corresponding author.