Characterization and diagnosis of atherosclerosis: imaging and urine proteomics

The concept of vulnerable plaques has been long described since some atherosclerotic lesions rupture suddenly causing myocardial infarctions and strokes while others remain quiescent or stable for many years. Two potentially feasible approaches were used here in an attempt to identify these vulnerable plaques: imaging and urine proteomics. Imaging: Several intravascular imaging techniques have been investigated to identify vulnerable plaques without definitive success yet, demanding better understanding of pathophysiology of these lesions and more reliable imaging methods. We described the near infrared range (NIR) intrinsic fluorescent activity to be a property unique to the unstable plaques, using a well-established mouse model of tandem stenosis as well as human carotid endarterectomy samples. The source of NIR autofluorescence is shown to be intraplaque haemorrhage where haem degradation products were intermingled with various chemicals of the necrotic core. We also demonstrated that changes in the plaque burden were reflected by the changes in NIR fluorescent intensity using haem oxygenase enzyme modulation which played a complex role in the pathophysiology of plaque progression and vulnerability. NIR autofluorescence in the areas of intraplaque haemorrhage, a critical element of plaque vulnerability, provides a much needed new foundation in the field of intravascular imaging for unstable plaques. Although NIR fluorescent imaging is still in the pre-clinical stage, it should be further explored with the aim of developing intravascular probe to be applied in clinical studies. Urine proteomics: A pilot proteomic study aimed at the identification of secreted urinary peptide biomarkers and the modelling of a prognostic classifier for acute coronary syndrome (ACS) is reported in this thesis. The urinary proteome profile data of 126 individuals who had suffered from ACS up to 5 years post urine sampling and proteome data of 126 controls without ACS were analysed. The initial statistical comparison of proteome profile data of 84 individuals with an ACS and 84 matched controls resulted in the discovery of 75 potential ACS-specific prognostic peptide biomarkers. Based on these peptide biomarkers we established the support vector machine-modelled prognostic ACS classifier ACSP75. The performance of the classifier was assessed using sensitivity, specificity and discrimination (c-statistics) and was compared the performance of the Framingham risk score (FHS), and to an algorithm combining the classifier, age and BMI. In the validation data set, the classifier identified individuals with an ACS with a sensitivity of 73.8% and demonstrated reasonable discrimination (c statistic=0.664). The classifier showed similar performance compared to FHS (C-statistics: 0.664 vs 0.644 [p=0.692] for classifier and FHS, respectively). In a model where we combined the classifier with other traditional risk factors (BMI and age), the algorithm showed good discrimination (c-statistic=0.707), but was not significantly better than the classifier itself (p=0.213). The sensitivity (83.3 %) and specificity (78.6 %) of the composite classifier was better than the classifier on its own. We demonstrate that the proteomic classifier ACSP75 based on urinary peptide biomarkers has the potential to predict future ACS events. The classifier and the composite classifier should be validated in a large cohort