Critical mechanical conditions around neovessels in carotid atherosclerotic plaque may promote intraplaque hemorrhage

Abstract

1 Introduction Stroke is the third leading cause of death and the primary cause of disability in the world [1]. Carotid atherosclerotic disease is thought to be the predominant etiology of stroke in Western society [2]. Nevertheless, clinical assessment of stroke risk has not progressed beyond the use of luminal stenosis in spite of evidence to suggest that this is an inadequate predictor of stroke [3]. Recent imaging studies have suggested plaque composition as an independent risk factor for ischemic stroke [4,5]. A typical carotid atherosclerotic plaque is composed of lipid-rich necrotic core (LRNC), calcium and plaque hemorrhage covered by a layer of fibrous cap (FC). High risk characteristics include large LRNC [6], presence of hemorrhage [7], and thin or defective FC [8]. Intraplaque hemorrhage (IPH) may cause complications by promoting vulnerability, luminal occlusion or downstream emboli. Long term plaque progression due to IPH can be captured using high-resolution magnetic resonance imaging (MRI) [9]. It has been observed that IPH has a much wider prevalence in symptomatic patients than asymptomatic individuals [5]. Altaf et al. found that 15 out of 66 recurrent events were associated with IPH while only two recurrent events occurred in its absence in symptomatic patients with high-grade atherosclerotic lesions [10]. Similar results were obtained from 39 symptomatic patients with mild to moderate (30–69%) stenosis [11]. Recently, in a prospective study of 61 acutely symptomatic patients, Sadat et al. found that the presence of plaque hemorrhage was closely associated with the occurrence of future cerebrovascular events [12]. Even in asymptomatic patients, there is a risk conferred by IPH. Takaya et al. found that presence of IPH was associated with the incidence of cerebrovascular events in previously asymptomatic patients (n = 154; stenosis: 50–79%) [13]. Singh et al. confirmed that MR-depicted IPH was associated with an increased risk of cerebrovascular events (n = 91) in asymptomatic moderate carotid stenosis [14]. Histopathological examinations have revealed the association between IPH and the presence of neovessels [15,16]. Neovascularization can be considered a compensatory response to hypoxia present in the deep intimal and medial areas of the artery [17,18]. Due to poorly developed vessel walls, blood components, such as red blood cells (RBC), neutrophils and other proinflammatory cells, may migrate from the bloodstream into the plaque [19,20]. These may release an array of proteases that induce the death of endothelial cells, thereby generating local disruption of microvessels [21] and further promoting IPH. Aside from these inflammatory factors, under physiological conditions, atherosclerotic plaque is subjected to mechanical loading due to blood pressure, as are its associated neovessels. Finite element analysis has been widely used to estimate the stress concentrated within the fibrous cap as a mechanics-based vulnerability assessment [22,23]. This is based on the hypothesis that fibrous cap rupture possibly occurs when the extra loading due to blood pressure and hemodynamic flow exceed its material strength. Being embedded in the atherosclerotic plaque, neovessels are also presumably susceptible to mechanical loading effects from the deformation of entire plaque structure driven by the dynamic blood pressure. Harsh local mechanical conditions, if present, may also contribute to the neovessel damage and further encourage IPH formation. However, this has not yet been examined in detail. This study, therefore, aims to (1) quantify the critical mechanical conditions (stress and stretch) around neovessels based on high-resolution histological images; and (2) characterize the association between these conditions and plaque's pathological features, such as the distribution of red blood cells as a marker of IPH. 2 Materials and methods Four carotid plaques with over 70% stenosis were collected en bloc following carotid endarterectomy. One of the four patients was male; they were 74.3 ± 15.2 years old; the blood pressure was 127.5 ± 26.8 mmHg for systole and 78.0 ± 15.0 mmHg for diastole. The samples were formalin-saline fixed, decalcified, embedded in paraffin and stained using hematoxylin and eosin (H&E), Verhoeff's Van Gieson (EVG), Nile red and Masson's trichrome to visualize various components within plaque. Histopathological slides were digitalized using NanoZoomer (Hamamatsu, Japan) (Fig. 1). Considering the computational workload, one slide located at the most stenotic site was chosen for analysis. The digitalized image was segmented manually using NDP Viewer (Hamamatsu, Japan) to identify the lumen contour, fibrous tissue, lipid and hemorrhage, etc. The contours of lumen and outer wall of each neovessel were carefully traced at 40× magnification. About 100 neovessels were identified for each slide. All contours were exported and processed using an in-house developed package in Matlab (MathWorks, USA). All components were assumed to be non-linear hyper-elastic, piecewise homogeneous and incompressible materials governed by the modified Mooney–Rivlin strain energy density function, W = c 1(I 1 − 3) + D 1exp[D 2(I 1 − 3) − 1]where I 1 is the first strain invariant and c 1, D 1 and D 2 are material parameters, derived from earlier studies [24] with the following details: vessel material: c 1 = 36.8 kPa, D 1 = 14.4 kPa, D 2 = 2; fibrous cap: c 1 = 73.6 kPa, D 1 = 28.8 kPa, D 2 = 2.5; lipid core: c 1 = 2 kPa, D 1 = 2 kPa, D 2 = 1.5: calcification, c 1 = 368 kPa, D 1 = 144 kPa, D 2 = 2.0; fresh IPH: c 1 = 1 kPa, D 1 = 1 kPa, D 2 = 0.25 and for chronic IPH: c 1 = 9 kPa, D 1 = 9 kPa, D 2 = 0.25. The blood pressure of each patient was used as the loading condition applying on the plaque as a whole and the pressure in the neovessel was assumed to be 10 mmHg (as it was not directly measurable). This value was chosen because it approximately reflects blood pressure in the venous environment. However, our experimental conclusions did not change when the value was lowered to 5 mmHg. Considering the small size of an individual neovessel, a very fine mesh was used around the local region with about 0.5 μm on each element edge. Each model consists of over 100,000 elements. Maximum principal stress (Stress-P1) and stretch (Stretch-P1) were computed using finite element method (FEM) in ADINA8.6.1 (ADINA R&D, Inc., USA). The region of interest (ROI) for each neovessel was defined as the region within four times of the corresponding lumen area (The number could be changed to 2.5, 6 and 8 and the results and conclusions remained the same). The maximum value of Stress-P1 and Stretch-P1 within ROI was extracted from the simulation. The value at systole and the difference across the cardiac cycle were used to quantify the critical mechanical condition. The change of lumen area of the neovessel during the cardiac cycle was also computed to quantify the deformation. The locations of red blood cells were recorded to quantify their distribution. Therefore, the neovessels were divided into two groups (without-RBC and with-RBC) depending on the presence of red blood cell within the ROI. The association between this distribution and critical mechanical condition was further analyzed. The statistical analysis was performed in Instat3.06 (GraphPad Software Inc., USA). A two-tailed Mann–Whitney test was used for the statistical analysis if the data did not pass the normality test (Shapiro–Wilk test); otherwise, two-tailed student t test was used. A significant difference was assumed with a p-value <0.05. 3 Results In total, 379 neovessels were identified in four histological slides. Red blood cells were found within the region of interest of 146 of them (38.5%). As it can be seen from Fig. 1, neovessels appeared throughout the plaque structure and many were adjacent to the lumen (Fig. 1A&B), some were located in the middle of the plaque (Fig. 1C) surrounded by a cluster of red blood cells and some were located in the periphery of the plaque (Fig. 1D) with various lumen sizes and wall thicknesses. The corresponding band plot of Stretch-P1 was shown in Fig. 2. As depicted in the amplified thumbnails, large deformations were found around the neovessel when it was close to the lumen (Fig. 2A, B & C). Fig.3 visualizes the location-dependent mechanical parameters, stress concentration at systole (Stress-P1; Fig.3A), stress variation during one cardiac cycle (Diff-Stress-P1; Fig.3B), large local deformation at systole (Stretch-P1; Fig.3C) and the stretch variation (Diff-Stretch-P1; Fig.3D), in the ROI of each neovessel. These parameters decrease greatly when the neovessel is located away from the carotid lumen. The harsh mechanical environment around neovessels may be associated with the leak of red blood cells, which were found around those neovessels that had undergone a large deformation, shown in Fig. 4. Further analysis indicated that there was no significant difference (p = 0.087) in terms of Stress-P1 at systole between the groups without (without-RBC) and with (with-RBC) red blood cells (Table 1); however, during one cardiac cycle, with-RBC underwent greater Stress-P1 variation (Diff-Stress-P1) than without-RBC (p < 0.0001). The deformation of neovessels in with-RBC group at systole was about 8.9% which was much greater than the one in the without-RBC group (3.4%; p < 0.0001). During one cardiac cycle, the stretch variation (Diff-Stretch-P1) of with-RBC was about 2.82%, while the value of without-RBC was only 0.87% (p < 0.0001). Furthermore, the lumen contour deformed (Diff-Area) much less in the without-RBC group than that in the with-RBC group (0.565% vs. 2.024%; p < 0.0001). 4 Discussion To our knowledge, this is the first study to quantify the mechanical conditions around neovessels within atherosclerosis (Fig. 2). We highlight possible associations between intraplaque hemorrhage and these mechanical conditions (Fig. 4 and Table 1). We found first, that mechanical stress and stretch decreased significantly as the distance between the neovessel and the main arterial lumen increased (Fig. 3). Second, those neovessels with surrounding red blood cells, presumably evidence of fresh hemorrhage, underwent much larger deformation at systole and stress and stretch variations during one cardiac cycle than those without red blood cells close by (Table 1). Several studies have shown a pathological effect of vessel stretch on the cellular and genetic environment of the plaque. The large cycle deformation may impede endothelial cell survival and tubulogenesis through the NAD(P)H subunit p22phox pathway [25]. Pathological stretch can dysregulate cytoskeletal gene expression, such as filamin A [26], affecting cell attachment and encouraging programmed cell death [27] and therefore preventing healing in the carotid plaque following acute events [28]. On a tissue level, the risk of elevated strain/deformation on plaque destabilization has been also recognized by various computational and clinical studies [28–30]. Although there are likely several biological processes at work in the promotion of intraplaque hemorrhage, here we suggest a possible contribution from the mechanical conditions around the neovessel. The association between alterations in mechanical stress and plaque hemorrhage was suspected by Lusby et al. in early 1980s [31]. Recently, intraplaque hemorrhage has been recognized as one trigger of plaque vulnerability [32]. Monitoring the development of neovascularization within plaque might be important clinically. Non-invasive imaging techniques [33] such as contrast-enhanced magnetic resonance imaging (MRI) [34] and microbubble-targeted ultrasound [35], have been developed to quantify it. In vivo high-resolution elastography approaches, such as intravascular ultrasound [36], optical coherence tomography [37] and B-mode ultrasound elastography [38], have shown the capacity in quantifying the local tissues deformation in the atherosclerotic plaque. Further development of these techniques could lead to a more accurate plaque vulnerability assessment by integrating plaque compositional features and critical mechanical conditions. Despite the interesting findings reported in our paper, some limitations exist: (1) the small number of plaques analysed (n = 4) means the pathological conclusions, such as the distribution pattern of neovessels and extravasated red blood cells ought to be repeated. However, this limitation does not completely negate our conclusion that large deformations around the neovessel might promote hemorrhage, because those four plaques yielded approximately 400 neovessels for analysis; (2) the origin of neovessel could be various. It may be from the vasa vasorum in the adventitia or due to the thrombus healing [39]. They are not differentiated in this study; another consideration is that (3) this study was a two-dimensional simulation, and the effect of the blood flow was not taken into account in this model. Since the neovessels were located within the plaque structure, high velocity blood flow in the main arterial lumen should have minimal impact on the prediction of critical mechanical conditions around the neovessel; lastly, (4) despite rigorous attention to detail, some distortion of the plaque samples may have occurred during processing for histopathological examination. Our segmentation, therefore might not represent the true in vivo configuration of the plaque. In conclusion, we suggest that there are large degrees of deformation and high variation in the mechanical loading around plaque neovessels during the cardiac cycle. These factors might damage the vessel walls and, in conjunction with inflammatory and other factors, promote intraplaque hemorrhage. Conflict of interest None.