Objectives To evaluate the prognostic importance of transient or persistent elevations of cardiac troponin-I (cTnI) and N-terminal-B-type natriuretic peptide (NT-proBNP) in atrial fibrillation (AF).

Methods Plasma samples were obtained at randomisation and after 3 months in 2514 patients with AF in the RE-LY trial; median follow-up was 2.0 years. Patients were grouped based on levels at the two time points according to detectable cTnI levels (≥0.01 µg/L) or NT-proBNP levels above median (≥778 ng/L). These groups were related to occurrence of stroke or cardiovascular events evaluated with Cox models adjusting for established risk factors.

Results The proportion of patients with detectable cTnI levels at both time points was 48.5%, at one time point 28.5% and at neither time point 21.0%. Patients with detectable cTnI at both time points had substantially higher rates of stroke compared with those with transient elevations and those with no elevation at either time point (p<0.005, effect of cTnI). Persistent elevation of either or both cardiac biomarkers at baseline and 3 months was associated with a higher risk for cardiovascular events and mortality (p<0.0001). Prognostic prediction improved most with the use of repeated measurements of both cardiac biomarkers simultaneously (p<0.05) and achieved C-statistic 0.644 for stroke compared with 0.611 for CHADS₂-score.

Conclusions Persistent elevation of troponin and NT-proBNP indicates a worse prognosis than transient elevations or no elevations of either marker. Prognostication of stroke, death and thromboembolic events is improved by the use of repeated determinations of cardiac biomarkers.

Trial registration number: http://www.clinicaltrials.gov, NCT00262600