Prediction of Cardiovascular Disease Risk by Cardiac Biomarkers in 2 United Kingdom Cohort Studies: Does Utility Depend on Risk Thresholds For Treatment?

We tested the predictive ability of cardiac biomarkers N-terminal pro B-type natriuretic peptide (NT-proBNP), high-sensitivity troponin T, and midregional pro adrenomedullin for cardiovascular disease (CVD) events using the British Regional Heart Study (BRHS) of men aged 60 to 79 years, and the MIDSPAN Family Study (MFS) of men and women aged 30 to 59 years. They included 3757 and 2226 participants, respectively, and during median 13.0 and 17.3 years follow-up the primary CVD event rates were 16.6 and 5.3 per 1000 patient-years, respectively. In Cox models adjusted for basic classical risk factors, 1 SD increases in log-transformed NT-proBNP, high-sensitivity troponin T, and midregional pro adrenomedullin were generally associated with increased primary CVD risk in both the studies (P<0.006) except midregional pro adrenomedullin in MFS (P=0.10). In BRHS, QRISK2 risk factors yielded a C-index of 0.657, which was improved by 0.017 (P=0.005) by NT-proBNP, but not by other biomarkers. Using 28% 14-year risk as a proxy for 20% 10-year risk, NT-proBNP improved risk classification for primary CVD cases (case net reclassification index, 5.9%; 95% confidence interval, 2.8%–9.2%), but only improved classification of noncases at a 14% 14-year risk threshold (4.6%; 2.9%–6.3%). In MFS, ASSIGN risk factors yielded a C-index of 0.752 for primary CVD; none of the cardiac biomarkers improved the C-index. Improvements in risk classification were only seen using NT-proBNP and high-sensitivity troponin T among cases using the 28% 14-year risk threshold (4.7%; 1.0%–9.2% and 2.6%; 0.0%–5.8%, respectively). In conclusion, the improvement in treatment allocation gained by adding cardiac biomarkers to risk scores seems to depend on the risk threshold chosen for commencing preventative treatments.