Local atherosclerotic plaque composition and plaque rupture

Abstract

Introduction: Carotid atherosclerosis is characterized by the build-up of the so-called plaque tissue in the wall of the carotid artery. The rupture of carotid plaques can lead to acute manifestations, including ischemic stroke and transient ischaemic attack (TIA). The mechanism responsible for rupture is not yet unravelled, complicating the medical management of the disease. There are indications that plaque morphology plays a role in plaque rupture risk. Previous histopathological studies found that ruptured plaques and non-ruptured plaques tend to differ in their structural composition. These studies did not perform a detailed analysis of the local plaque features at the rupture sites, even though this might increase the insight in the rupture event. The aim of this thesis was to investigate the relation between local plaque composition and rupture in carotid plaques. Methods: Digitized histological sections of 21 ruptured carotid endarterectomy samples (obtained from fourteen different patients) were segmented for the following nine tissue structures: cholesterol crystals,plaque, calcium, necrotic core, fresh haemorrhage, foam cells, neovessels, macrophages and old haemorrhage. The rupture site on the sections was identified as a luminal disruption in the presence of red blood cells. The digitized sections were then divided into 8 bins (pie slices), resulting in a total of 168 bins (21 sections x eight bins). Of these observations, 23 contained rupture, whereas the other 145 were rupture-free. The plaque area in each bin was measured and the other tissue structure sizes were calculated relative to the plaque area in the bin. Spatial heatmaps demonstrated the location of the tissue structures relative to the rupture site. They were created for the three variables cholesterol crystals, calcium and macrophages, as these tissue structures were often recognized close to the rupture sites. Kendall Tau-b correlation coefficients were calculated to determine the relation between tissue structures in close proximity. The two-sided clustered Wilcoxon rank sum test (with clusters at the patient level) was used to investigate differences between the composition of ruptured and non-ruptured bins. Additionally, the association of tissue structures with rupture was tested by a generalized estimating equations (GEE) logistic regression model (working correlation matrix:independence). Clusters were assigned at the patient level. A sensitivity analysis was carried out (using 2-,4- and 16-bins) to determine the effect of the bin size on the outcome of the statistical analyses. Results: Visual analysis demonstrated that ruptures tended to extend to (or start at) areas of calcification or cholesterol crystals and that the rupture sites could be markedly inflamed. The spatial heatmaps showed that areas of calcification were closer to the rupture sites than areas with cholesterol crystals. Moreover, the prevalence of macrophages near the rupture sites was lower when compared to the prevalence of calcium and cholesterol crystals. In ruptured bins, the strongest tissue structure correlation was found between calcium and cholesterol crystals (tau_b= -0.44, p ≤ 0.05). This correlation was less evident when ruptured and non-ruptured bins were analysed together (tau_b= -0.07, p > 0.05). The clustered Wilcoxon rank sum tests howed that calcium (p=0.05) and fresh haemorrhage (p=2.17x10-11) tended to take on higher values in the ruptured bins, when compared to the non-ruptured bins. Lastly, with the GEE logistic regression model an association was found between rupture and the predictor variables calcium (p= 0.00094) and plaque (p=0.049).The odds ratio of rupture was 1.05 [1.02-1.07] for a plaque size increase of 0.34 mm2, if the other plaque characteristics in the bin were held fixed. The odds ratios of rupture belonging to plaque-bins that contained10% calcium and 50% calcium were 1.96 [1.60-2.40] and 3.01 [2.16-4.20], respectively, relative to plaque-bins with 0% calcium content and equal other plaque characteristics in the bin. The sensitivity analysis showed that the choice for an 8-bin approach was reasonable. Conclusion. The rupture sites of the carotid sections showed an enhanced calcium content. Moreover, the calcium content and the plaque size showed to be significantly associated with plaque rupture. The developmental progress of non-invasive imaging techniques, using magnetic resonance imaging and ultrasound, is promising for the characterization of carotid plaques in vivo. Understanding the relation between plaque composition and plaque rupture at a local level can contribute to rupture risk assessment and eventually to the medical management of atherosclerosis.