Performance characteristics of the Architect brain natriuretic peptide (BNP) assay: a two site study

INTRODUCTION: Brain natriuretic peptide (BNP) is produced by the ventricles of the heart and is a biomarker for heart failure. Several commercial assays are now available. We evaluated the performance characteristics of the ARCHITECT BNP assay. METHODS: We evaluated the limit of blank, limit of detection, linearity and imprecision. Method comparison studies were performed with 3 other automated BNP assays including the ADVIA Centaur, AxSYM, and UniCel DxI 800 methods. RESULTS: The mean LOB and LOD of the Architect assay were 3.5 and 5.8 ng/L, respectively. Imprecision studies yielded within run CVs of 1.1 to 5.1% and total CVs of 2.3 to 5.3% using human plasma based multi-constituent controls at concentrations of 92, 500, and 3500 ng/L. The maximum deviation from the target recovery for dilution linearity was 9.6%. Concordance with other BNP assays at a 100 ng/l cutoff was 91 to 98% and kappa statistics were 0.78 to 0.96. The mean difference between the Architect and Advia Centaur methods was positive. For the other methods, the mean difference with the Architect was negative. CONCLUSIONS: The Architect BNP assay shows good performance characteristics with total imprecision < or =5.3%. It agrees well with the Advia Centaur, AxSYM, and UniCel DxI BNP assays.