Targeted multiplex proteomics for risk stratification in patients with cardiovascular disease

Risk stratification is valuable in patients with cardiovascular disease (CVD). Proteins involved in different pathophysiological processes in atherosclerosis have shown prognostic capacity. A new technology, Proximity Extension Assay (PEA), enables the simultaneous analysis of numerous plasma proteins from a minimal amount of plasma. In this thesis, the overall aim was to identify and study prognostic protein biomarkers that could become useful in risk stratification. We used PEA to identify the biomarkers with the best prediction of long-term mortality among patients with acute myocardial infarction (AMI) and peripheral arterial disease (PAD). The study populations included consecutive patients with AMI from the Västmanland Myocardial Infarction Study (VaMIS), outpatients with carotid or lower extremity PAD from the Peripheral Arterial Disease Study in Västmanland (PADVa), and a population-based control cohort. A set of 92 proteins was analysed with PEA and related to follow-up data in the populations. The biomarkers which were best at predicting all-cause mortality were identified using LASSO regression analyses. The added value of the identified biomarkers to clinical risk markers was evaluated by logistic or Cox regression models. The associations between the plasma concentrations of growth differentiation factor 15 (GDF-15) to clinical risk factors, indicators of atherosclerotic burden, and variables of cardiac geometry and function were analysed with linear regression models. Spearman’s rank correlation coefficients compared the plasma concentrations obtained from conventional immunoassays and PEA for GDF-15 and N-terminal pro-brain natriuretic peptide (NT-proBNP) in the PAD population and the population-based cohort without PAD. The association between the two assay data and outcome was evaluated separately with Cox regression. GDF-15 and tumour necrosis factor-related apoptosis-inducing ligand receptor 2 (TRAIL-R2) had the best prognostic performance for all-cause mortality among individuals with AMI, and PAD, and improved the risk prediction beyond the established risk markers. Circulating GDF-15 levels among individuals with PAD were independently associated with several of the clinical and biochemical risk variables, particularly diabetes and low low-density lipoprotein cholesterol. We demonstrated an excellent correlation and a similar prognostic performance of plasma levels of NT-proBNP and GDF-15 obtained by conventional assays compared with PEA in both cohorts. Except for high levels of NT-proBNP, the PEA reliably reflects the serum levels obtained from the conventional assay. Hopefully, our results can contribute to the finding of potential biomarkers useful in clinical practice in order to better identify subgroups among patients with CVD. In the long run, this might open up possibilities for individualised treatment and to more cost-effective follow-up routines, thus improving quality of life and longevity