Abstract / Summary
N-terminal pro-B-type natriuretic peptide (NT-proBNP) is widely used for prognostic assessment in cardiovascular disease, but the optimal timing of its measurement for risk stratification after acute myocardial infarction (AMI) remains unclear. Among patients who survived the first year after acute myocardial infarction, we analysed how strongly NT-proBNP measured at the index admission and NT-proBNP measured at the one-year visit were each associated with cardiovascular events that occurred thereafter. We analyzed 848 patients from the nationwide KAMI registry who had NT-proBNP measured both at admission and at the one-year follow-up visit and who were free of the primary outcome up to that visit (a one-year landmark cohort). The time origin was set to each patient’s actual one-year NT-proBNP measurement date. The primary endpoint was a composite of cardiac death, recurrent myocardial infarction, and repeat percutaneous coronary intervention (Re-PCI) that occurred after the landmark. Prespecified covariates were all recorded at the index admission, and only the exposure was measured at the one-year visit. Sensitivity analyses used a stricter endpoint that held only cardiac death and recurrent myocardial infarction, a reduced set of covariates, and a binary ejection fraction below 40%. The value that each measurement added to a clinical model was tested by the likelihood ratio test. Over a median follow-up of 4.1 years after the landmark, 61 composite events occurred. Stratification by admission NT-proBNP (cohort median 160 pg/mL) showed a non‑significant difference in event rates (8.5% vs. 5.9%, log-rank p = 0.121), whereas stratification by the one-year NT-proBNP (median 151 pg/mL) did (10.1% vs. 4.2%, p < 0.001). Analyzed separately, the one-year NT-proBNP discriminated risk for cardiac death (4.7% vs. 0.2%, p < 0.001) and recurrent myocardial infarction (3.8% vs. 1.4%, p = 0.031) but not Re-PCI ( p = 0.636). In a multivariable Cox model adjusted for nine prespecified covariates (834 patients, 60 events), the continuous log₂-transformed one-year NT-proBNP remained independently associated with the composite outcome (adjusted hazard ratio 1.34 per doubling, 95% CI 1.18–1.54, p < 0.001). The association with the dichotomized form was of similar magnitude but did not reach significance (adjusted hazard ratio 1.86, 95% CI 0.98–3.53, p = 0.059). The one-year value added prognostic information to a clinical model that held neither measurement (likelihood ratio χ² = 17.03, p < 0.001), whereas the admission value did not (χ² = 0.04, p = 0.845). With the stricter endpoint, 41 events occurred and the adjusted hazard ratio per doubling of the one-year value was 1.43 (95% CI 1.21–1.69, p < 0.001). Among patients who survived the first year after acute myocardial infarction, a higher NT-proBNP at the one-year visit was associated with a higher long-term risk of cardiac death and recurrent myocardial infarction, whereas the admission value was not associated with the composite outcome. This association was independent of established clinical risk factors. Because the analysis was observational and the number of events was small, these findings identify a prognostic association rather than a management strategy. Whether measurement of NT-proBNP at the one-year visit improves clinical decisions or patient outcomes requires prospective evaluation.