Incremental utility of circulating biomarkers for cardiovascular risk prediction beyond the updated SCORE2 model

Abstract

Abstract Background Accurate risk prediction for future cardiovascular disease (CVD) is crucial for timely initiation of preventive measures in high-risk individuals. Most risk scores, such as the recently updated SCORE2 risk-prediction model supported by the European Society of Cardiology, consider only traditional cardiovascular risk factors. Whether the addition of circulating biomarkers to the existing SCORE2 model may improve risk prediction is unclear. Purpose We aimed to evaluate the incremental utility of four widely available circulating biomarkers to improve the prediction of 10-year CVD-risk beyond SCORE2. Methods Data from ten prospective population-based cohorts from seven countries across Europe were collected if information on SCORE2-variables and at least one of the following four investigational biomarkers was available: high-sensitivity cardiac troponin I (hs-cTnI), NT-pro-B-type natriuretic peptide (NT-proBNP), high-sensitivity C-reactive protein (hs-CRP) and creatinine-based estimated glomerular filtration rate (eGFR). Primary outcome was incidence of CVD at 10 years, defined as the composite of cardiovascular mortality, non-fatal myocardial infarction and non-fatal stroke. We used Fine and Gray models adjusted for competing-risk and SCORE2-variables as well as penalized cubic splines to assess and visualize the association of individual biomarkers with incident CVD. In a multimarker approach, we performed backward selection to identify biomarkers providing independent predictive value beyond SCORE2-components. C-indices and category-free net reclassification index (cfNRI) were used to compare the performance of the original SCORE2 model to the biomarker-extended model. Results In 78'507 individuals, median age was 50 years and 50.3% were females. NT-proBNP, hs-CRP and hs-cTnI but not eGFR showed strong associations with 10-year CVD-risk when adjusted for SCORE2 and provided incremental predictive value when individually added to SCORE2 (Figure 1). In a multimarker approach, all three biomarkers remained independently associated with CVD beyond SCORE2 with strongest association of NT-proBNP, followed by hs-CRP and hs-cTnI (Table 1). The simultaneous addition of these three biomarkers to the SCORE2 model significantly increased discrimination (C-index; 0.782 [95% CI, 0.757, 0.806] versus 0.793 [95% CI, 0.768, 0.817], Delta 0.011 [95% CI, 0.005, 0.016]) and risk reclassification, driven by an improvement in non-events (cfNRIoverall 0.17 [95% CI, 0.12, 0.22], cfNRIevents 0.06 [95% CI, 0.02, 0.11], cfNRInon-events 0.11 [95% CI, 0.10, 0.11]). Conclusion NT-proBNP, hs-CRP and hs-cTnI but not eGFR provide incremental predictive value when added to the SCORE2 risk-prediction model and may help to further improve personalized CVD-risk prediction. Funding Acknowledgement Type of funding sources: Public grant(s) – EU funding. Main funding source(s): European Community's Seventh Framework Programme (FP7/2007-2013)