Plasma Protein Biomarkers and Long-Term Cardiovascular Mortality Risk in Patients With Chronic Coronary Heart Disease

Abstract

Background: Protein biomarkers that reflect different pathophysiological pathways have been associated with the risk of adverse cardiovascular events. However, it is uncertain whether these associations are sustained with increasing years after the biomarkers are measured.

Methods and results: In this cohort study, 7745 patients with coronary heart disease who participated in the LIPID (Long-Term Intervention With Pravastatin in Ischemic Disease) trial, BNP (B-type natriuretic peptide), troponin I, cystatin-C, C-reactive protein, d-dimer and midregional proadrenomedullin were measured at baseline and after 1 year. Discrimination of plasma biomarker concentrations for cardiovascular death were evaluated in landmark analyses from 1 year for the next 5 years of the randomized trial, and for 10 additional years after trial completion. All 6 biomarkers were associated with risk of cardiovascular death (n=1903) both during and after the clinical trial (each P<0.001). C-statistics for BNP were 0.706 and 0.704; cystatin-C, 0.686 and 0.693; troponin I, 0.686 and 0.689; C-reactive protein, 0.655 and 0.684; d-dimer, 0.670 and 0.679, and midregional adrenomedullin, 0.686 and 0.688, respectively. In multivariable models, adding all 6 biomarkers to models with clinical risk factors increased the C-statistic for cardiovascular death from 0.709 to 0.775 during the clinical trial, and from 0.713 to 0.751 during 10-year follow-up after the randomized trial (P<0.001 for both).

Conclusions: In patients with chronic coronary heart disease, biomarkers that reflect different pathophysiological pathways are associated with the risk of cardiovascular death for at least the next 15 years.

Keywords: biomarkers; cardiovascular risk; coronary heart disease; death; long‐term survival.

Figure 1. Relationship between the most recent biomarker concentration in deciles and cardiovascular death during the next 15 years.

Relationship with cardiovascular death over deciles of year 1 biomarkers on log₂ scale, in a landmark analysis at 1 y for (A) B‐type natriuretic peptide in pg/mL, (B) C‐reactive protein in mg/L, (C) cystatin‐C in mg/dL (D) d‐dimer in ng/mL, (E) midregional proadrenomedullin in mmol/L, and (F) troponin I in ng/mL. Censored data are given with square boxes, and circles give measured data. For troponin I ≈33% of data fall below the detectable limit. The first square denotes all censored data, and the other 5 circles each denote ≈11% of the measured data.

Figure 2. Hazard ratios for cardiovascular death for each biomarker during early and late follow‐up after the clinical trial.

Early follow‐up refers to the randomized clinical trial period. Late follow‐up refers to follow‐up after the randomized trial. Hazard ratios for cardiovascular death by plasma concentration of 6 biomarkers, adjusted for age, sex, and randomized to pravastatin. The most recent biomarker measurement is used. Distributions shown on log₂ scale underneath. A, B‐type natriuretic peptide in pg/mL, (B) C‐reactive protein in mg/L, (C) cystatin‐C in mg/dL, (D) d‐dimer in ng/mL, (E) midregional proadrenomedullin in mmol/L, and (F) troponin I in ng/mL. The distribution of plasma concentrations for each biomarker are displayed on the x‐axis.