**4. Discussion**

Cats with underlying myocardial disease are known to be at a greater risk of thromboembolic events which frequently results in euthanasia on humane grounds [6,7]. Given the high prevalence of cardiomyopathy in cats, which for HCM is estimated at up to 17% in outbred animals and even greater in certain pedigree breeds [33], it is important to obtain a greater understanding of the left atrial remodelling process that predisposes to thrombus formation.

This is the first study to quantify and characterise vWF in the LA endocardium and LA thrombi from cats with cardiomyopathy at different stages of their disease process. LA samples from cardiomyopathic cats with LAE and clinical signs (CHF and ATE), but

not those from subclinical cardiomyopathic cats had increased endocardial vWF protein compared to control cats (Figure 5), which suggests an association of endocardial vWF elevation with an advanced stage of cardiomyopathy. The immunohistological evaluation of the thrombi revealed that vWF was extensively involved in thrombus formation and organisation of both microthrombi and the three macrothrombi, which were visible on gross pathology within the LA(A) in cats CHF 8, ATE 9, and ATE 12 shown in Figure S1.

The finding that vWF localised to some vessels in the LA samples is consistent with previous report of differential expression of endothelial vWF in mice where vWF expression varied not just in different organs but even in the vessels in the same vascular tree [42]. The minimal detection of vWF in the endocardium in the control cats is similar to that in a human report where no or only minimal focal immunostaining for vWF was present in the LAA endocardium from patients without cardiac abnormalities [19,43]. It is also striking that there were no female cats in our ATE group, which is consistent with previous reports showing a male predisposition to ATE and to an increase hazard of an ATE associated death. However, the possibility of this finding being due to chance cannot be ruled out [7,44].

Importantly, our finding in cats with advanced myocardial disease are similar to two human studies where patients with hemodynamical disturbance in the LA secondary to various cardiac causes showed elevated endocardial vWF compared to the non-cardiac patients using immunohistological microscopy [19,20]. In the first of these studies, a significant increase in the expression of vWF in the endocardium of atrial appendages was identified in human patients with a variety of congenital and acquired heart diseases irrespective of the presence of atrial fibrillation. Furthermore, increased vWF expression correlated with the degree of platelet adhesion and thrombus formation [19]. In the second study involving patients with valvular and non-valvular atrial fibrillation, expression of vWF protein in the endocardium correlated with the degree of structural remodelling in the atrial wall. In addition, endocardial vWF appeared important for platelet adhesion/aggregation on the endocardium and intra-atrial thrombi formation suggesting that increased endocardial vWF may contribute to LA thrombogenesis [20]. It is recognised that inflammation per se can cause thrombosis via a VWF-mediated mechanism by elevating the level of VWF, enhancing the reactivity of VWF, and modulating the levels and activities of regulatory molecules such as the metalloprotease ADAMTS13 [45]. Heart failure is an established cause of systemic inflammation, and the increased expression of vWF in our cohort of cats with clinical signs may at least in part be related to this pro-inflammatory milieu, however the relationship between circulating concentrations of vWF and heart failure remains unclear in human studies [46–48] and that between endocardial vWF and inflammation is not examined.

The immunostaining of the micro- and macrothrombi supports a pivotal role for vWF in thrombus formation. Thrombosis is a multifactorial process which usually involves an abnormality in one or more of the following factors: endothelial damage and dysfunction, disturbed blood flow, and hypercoagulability. Thrombus formation begins when platelets adhere to the exposed subendothelial matrix by means of vWF that was previously deposited on the sub-endothelium by endothelial cells. Once adhered, the stimulated platelets then secrete more mediators such as ADP and thromboxane A2 that further activate additional platelets leading to propagation of the thrombus [49,50].

Traditionally, two categories of thrombus have been described based on their gross appearance and composition. A red thrombus that forms under low shear stress and is primarily composed of erythrocytes and fibrin, and a white thrombus that forms under high shear stress and contains a large number of platelets and few erythrocytes. The former mostly comprise venous thrombi where stasis of blood is evident and endothelial injury is not an absolute requirement for thrombus formation. Conversely, white thrombi are usually confined to the arterial system and form after the severe endothelial disruption such as plaque erosion or rupture as part of the atherosclerosis process [51]. In the present study, platelets were found to be an important component in the thrombi despite being formed in the environment of slow blood flow and low shear stress found in the enlarged

LA. Different patterns of vWF expression in the thrombi were seen in or around particular cellular components forming different structural patterns around the same cell type, such us bead-like or web-like patterns around erythrocytes. vWF is known to fold or unfold under different conditions of shear stress and can therefore display various properties and functions as a result of exposing certain binding sites for adhesion or uncovering its cleavage site [52,53]. These various structural patterns formed by the thrombi might imply that the local environment within the LA where the thrombus developed altered over time, affecting vWF function and thereby the composition of the thrombus over its development. More research is needed to investigate how and why the feline cardiogenic thrombi showed components of both a traditional red and white thrombus.

There are a number of limitations to the study. First, the origin of the detected endocardial vWF protein in the LA samples remains undetermined. We cannot rule out the possibility that the LA endocardial vWF we detected was originally secreted by platelets or endothelia elsewhere in the body and delivered via the circulation. Second, although all the control cats had structurally normal hearts, a number did have other clinical potentially inflammatory conditions that may have affected endothelial vWF expression. However, most systemic inflammatory diseases are associated with increased level of vWF, which is contrary to the finding of low endocardial vWF in the control cats in our study. Third, in the subclinical group, four out of the nine cats had mild LAE. However, when we further sub-divided these cats into (1) subclinical with LAE, and (2) subclinical without LAE, there was no significant difference in endocardial fluorescence intensity between these subgroups and the controls. Fourth, to better evaluate the structure of the thrombi, sliced-though slides with co-immunostaining for fibrin would be helpful to gain greater understanding of thrombus organisation. However, this was beyond the scope of the study and not performed.
