Cardiac Troponin I and BNP for Predicting Zero Agatston Score in Patients with Diabetes Mellitus

Abstract

Coronary artery calcifications (CAC) scoring assessed by the Agatston score has shown an excellent prognostic value in large studies and particularly in diabetic patients, with a very low rate of cardiovascular events in patients with a zero Agatston score. Moreover, recent studies have suggested that high-sensitive cardiac troponin I (hs-cTnI) and brain natriuretic peptide (BNP) may be useful for detecting subclinical atherosclerosis. However, the link between hs-cTnI/BNP and the Agatston score has not been investigated in patients with diabetes. The aim of this study was to investigate if hs-cTnI and BNP can bring additional value to predict zero Agatston score in patients with diabetes mellitus in addition to usual risk factors. Between 2015 and 2019, CAC score was prospectively performed in consecutive patients with diabetes mellitus. Patients with symptoms, known coronary artery disease or history of atrial fibrillation were excluded. Within 24 h from CT examination, peripheral blood samples were taken to measure hs-cTnI and BNP. The relationship between serum hs-cTnI/BNP concentrations and zero Agatston score was assessed using univariate and multivariate binomial models. The implication of hs-cTnI and BNP in this multivariate model was evaluated using nested models associated with Chi 2 test of independence. A total of 844 patients with diabetes were enrolled (61 ± 7years, 57% men, mean duration of diabetes 18years). In this population, 294(35%) had a zero Agatston score, 253(30%) an Agatston score from 1 to 100, 161(19%) from 101 to 400, and 136(16%) higher than 400. In univariate analysis, hs-cTnI and BNP concentrations were associated with zero Agatston score (respectively OR, 2.63 [95%CI, 1.51-5.01]; P < 0.001 and OR, 1.09 [95%CI, 1.01-1.22]; P = 0.03). In multivariate analysis, hs-cTnI and BNP concentrations were associated with zero Agatston score (respectively OR, 2.38 [95%CI, 1.51-4.76]; P = 0.009 and OR, 1.18 [95%CI, 1.07-1.32]; P = 0.001). The multivariate model included age, gender, smoking, dyslipidemia, duration of diabetes, hypertension, diabetic neuropathy, hs-cTnI and BNP concentrations, significantly discriminated the zero Agatston score (AUC = 0.81; P < 0.001) ( Fig. 1 ). The most discriminant threshold was ≤ 3 ng/l for hs-cTnI and < 17 ng/l for BNP. In nested models, both hs-cTnI and BNP brought information to this multivariate model to predict a zero Agatston score (respectively P = 0.003 and P < 0.001). Moreover, removing hs-cTnI and BNP from the model results in a significant reduction in model performance (AUC = 0.79; P = 0.004). Cardiac biomarkers hs-cTnI and BNP are associated with zero Agatston score, which is correlated with a very low risk of cardiovascular events, in asymptomatic patients with diabetes mellitus.