Background: Risk assessment of future major adverse events in patients with stable coronary heart disease (CHD) is usually based on clinical risk factors. However, the precision in predicting events in these patients is low. In other diseases, novel biomarkers have shown to enhance risk prediction.
Purpose: To investigate if the associations between 155 potential protein biomarkers and cardiovascular death (CVD), improve risk assessment of patients with stable CHD on optimal secondary prevention therapy, in addition to established biomarkers and clinical risk factors.
Methods: In the Stabilization of Atherosclerotic Plaque by Initiation of Darapladib Therapy (STABILITY) trial, 15,828 patients with stable CHD and at least one clinical risk factor were included. During a median follow-up time of 3.7 years, 605 patients had CVD. Baseline plasma samples from all cases with CVD and 2,789 randomly selected control cases from the same cohort, underwent analysis for normalized protein expression (NPX) levels with proximity extension assay (PEA) technology. An unbiased selection of variables with CVD association (proteins and clinical risk factors) was performed with random forest analysis. Confirmation of variable importance for CVD was determined by Boruta analysis. Independent association between variables and CVD was evaluated by weighted multivariate Cox regression analysis adjusted for baseline characteristic and established cardiorenal biomarkers: cystatin C, N-terminal prohormone of brain natriuretic peptide (NT-proBNP) and high sensitivity troponin T (hs-TnT). Cox regression analysis used NPX levels modeled as continuous by standard deviation and because of multiple testing, threshold of significance was Bonferroni adjusted (p<0.00039).
Results: Median age was 69 years for cases and 65 years for controls. At baseline 92% and 97% were treated with acetylsalicylic acid and statin, respectively. Out of 155 analyzed proteins, 85 differed in NPX levels when comparing cases and controls (p<0.0001). After selection of variables associated with CVD, variable importance was confirmed for 28 proteins. Further, 4 of them passed the threshold of significance with independent associations (hazard ratio, 95% confidence interval) with CVD in adjusted models comparing cases with controls: NT-proBNP 2.35 (2.05–2.69), hs-TnT 1.48 (1.32–1.66), TRAIL receptor 2 (TRAIL-R2) 1.25 (1.15–1.35) and hepatocyte growth factor (HGF) 1.22 (1.10–1.36).
Conclusions: In this well-managed cohort of patients with stable CHD and high cardiovascular risk, the established cardiac risk markers NT-proBNP and hs-TnT were independently associated with CVD. The novel biomarkers TRAIL-R2 and HGF added prognostic value beyond that of conventional variables but need further validation. Furthermore, there was a sizable number of potential protein biomarkers associated with outcome before adjustment for established cardiorenal biomarkers, which need further evaluation.