Myeloperoxidase and C-reactive protein augment the specificity of B-type natriuretic peptide in community screening for systolic heart failure

Abstract

N-terminal pro-B-type natriuretic peptide (N-BNP) is elevated in left ventricular systolic dysfunction (LVSD) and may be cost-effective for screening in the community but is relatively nonspecific. We sought to improve specificity using inflammatory markers such as C-reactive protein (CRP) and myeloperoxidase (MPO), which have been implicated in cardiovascular disease. A total of 1360 subjects (45-80 years) were invited in this prospective screening study for undiagnosed LVSD (defined as wall motion score >1.8 [ejection fraction < or = 40%]), and 1331 had analyzable echocardiographic scans and plasma specimens. Peptides were measured using immunoluminometric assays. Twenty-eight patients with LVSD had elevated plasma N-BNP, CRP, and MPO levels compared with healthy subjects (P < .0005). Receiver operating characteristic curve areas for N-BNP, CRP, and MPO were 0.839, 0.824, and 0.909, respectively. All tests had high negative predictive values (>99%). Specificity was maximized to 88.4% in a logistic model with all 3 markers (all independent predictors, accounting for 44.8% of the variance). This reduced the number of cases to scan to detect 1 case of LVSD from 29.7 (using N-BNP alone) to 6.6. Using plasma MPO (at 33.9 ng/mL) or urinary N-BNP (at 10.7 fmol/mL) as initial screening tests, combinations of plasma N-BNP, MPO, and CRP can achieve specificities up to a maximum of 94.3%. Costs were minimized by using urinary N-BNP as the initial screening test, followed by plasma biomarkers. Plasma CRP and MPO increased the specificity of N-BNP in LVSD screening. Screening is optimized by urinary N-BNP as an initial test, followed by plasma CRP, N-BNP, and MPO.