Adding traditional and emerging biomarkers for risk assessment in secondary prevention: A prospective cohort study of 20,656 patients with cardiovascular disease.

Abstract

Background: This study aims to explore whether conventional and emerging biomarkers could improve risk discrimination and calibration in secondary prevention of recurrent atherosclerotic cardiovascular disease (ASCVD), based on a model using predictors from SMART2.

Methods: In a cohort of 20,658 UK Biobank participants with medical history of ASCVD, we analysed any improvement in C indices and net reclassification index (NRI) for future ASCVD events, following addition of LP-a, ApoB, cystatin C, HbA1c, GGT, AST, ALT, and ALP, to a model with predictors used in SMART2 for the outcome of recurrent major cardiovascular event. We also examined any improvement in C indices and NRIs replacing creatinine based estimated glomerular filtration rate (eGFR) with cystatin C based estimates. Calibration plots between different models were also compared.

Results: Compared with the baseline model (C index=0.663), modest increment in C indices were observed when adding HbA1c (ΔC=0.0064, p<0.001), cystatin C (ΔC=0.0037, p<0.001), GGT (ΔC=0.0023, p<0.001), AST (ΔC= 0.0007, p<0.005) or ALP (ΔC=0.0010, p<0.001) or replacing eGFRCr with eGFRCysC (ΔC=0.0036, p<0.001) or eGFRCr-CysC (ΔC=0.00336, p<0.001). Similarly, the strongest improvements in NRI were observed with the addition of HbA1c (NRI=0.014), or cystatin C (NRI= 0.006) or replacing eGFRCr with eGFRCr-CysC (NRI=0.001) or eGFRCysC (NRI=0.002). There was no evidence that adding biomarkers modify calibration.

Conclusions: Adding several biomarkers, most notably cystatin C and HbA1c, but not LP-a, in a model using SMART2 predictors modestly improved discrimination.