Head-to-head comparison of 10 natriuretic peptide assays

The aim of the study was to compare NT-proBNP and BNP levels in fresh samples from heart failure (HF) patients measured using 10 immunoassays and to assess their agreement.NT-proBNP (CobasH232(®), Elecsys(®), Vidas(®), Vista(®), XPand(®), Vitros(®)) and BNP (Triage(®), Access(®), CentaurXP(®), Architect(®)) levels were measured in 39 heparin and 19 EDTA samples, respectively.The Pearson correlation coefficient ranged between 0.929 (Triage(®-)Centaur(®)) and 0.994 (Access(®)-Architect(®)) for BNP assays and between 0.972 (Vidas(®)-Cobas H232(®)) and 0.999 (Vitros(®)-Vidas(®)) for NT-proBNP assays. Passing Bablok regression analyses showed a significant difference in the slopes [0.80 (Centaur(®)-Triage(®)) to 1.84 (Architect(®)-Centaur(®))] and intercepts [-55 ng/L (Architect(®)-Centaur(®)) to 48 ng/L (Access(®)-Triage®)] for BNP assays, and a lower heterogeneity between NT-proBNP assays [0.83 (Vidas(®)-Elecsys(®)) to 1.20 (Vitros(®)-Vidas(®)) and -97 ng/L (XPand(®)-CobasH232(®)) to 51 ng/L (CobasH232(®)-Elecsys(®)) for slopes and intercepts, respectively]. The concordance correlation coefficient revealed a poor (ρc<0.90) to moderate (ρc=0.90-0.95) agreement in 4/6 pairs of BNP assays and an almost perfect (ρc>0.99) agreement in 5/15 pairs of NT-proBNP assays. The acceptable difference limit reflecting the number of individual discrepant results between two assays, ranged between 15.1% (Access(®)-CentaurXP(®)) and 34.5% (Architect(®)-Triage(®)) for BNP assays, and between 10.9% (Vidas(®)-Vitros(®)) and 55% (CobasH232(®)-Xpand(®)) for NT-proBNP assays.This study stresses the lack of transferability of the results obtained using different techniques to measure BNP and NT-proBNP levels in fresh samples. Individual reference ranges and HF diagnostic cut-offs should be assessed for each commercial NP immunoassay. We recommend to systematically monitoring HF patients using the same assay (BNP or NT-proBNP) over the time.