Heart failure is a major global cause of disease and death [1]. B-type natriuretic peptide (BNP) is currently regarded as a clinically 6338useful biomarker for diagnosis of heart failure and monitoring treatment [2]. BNP is a biologically active hormone composed of 32 amino acids produced in the myocardium [2]. BNP increases in response to an increased myocardial wall stress [3]. Plasma BNP levels of 100 pg/mL and 400 pg/mL are helpful for exclusion and consideration of heart failure in untreated patients with symptoms of a heart failure, respectively [1, 3].

Various commercial BNP tests are performed using automated analyzers in many clinical laboratories [2]. The point of care (POC) test continues to be needed because it can reduce the clinical potential decision time in emergency situations by providing test results within approximately one hour [1]. However, the POC test should be verified to have a good correlation and precision compared to those in the clinical laboratory assays [1].

A new SelexOn^TM B-type natriuretic peptide assay (SelexOn BNP) (Osang Healthcare Inc., Anyang, Korea) was recently developed for POC testing. It consists of a single-use, disposable SelexOn^TM cartridge and a portable automated analyzer system. The principle for quantifying BNP is a rapid sandwich immunoassay using gold particles conjugated with a specific antibody on a test cartridge using lateral ow with K₂EDTA-anticoagulated whole blood. The camera inside the analyzer detects the signal line intensity. The integrated software converts the signal intensity to a quantitative result by comparing with an internal calibration curve (Fig. 1).

Figure 1. The Selexon^TM B-type natriuretic peptide (BNP) assay system. (A) The disposable cartridge uses a lateral flow immunoassay to detect BNP. Control and test lines are visible on the nitro-cellulose membrane. The control line indicates that the sample has migrated across the membrane as intended. (B) The camera inside the analyzer detects the signal line intensity. The integrated software converts the signal intensity to a quantitative result and shows it on the display.

We evaluated the analytical performance of the SelexOn BNP compared to two BNP assays based on the Clinical and Laboratory Standards Institute (CLSI) guidelines. This study was approved by the Catholic Medical Center Institutional Review Board (OC16-DISI0012).

1. Imprecision

Imprecision was evaluated according to the CLSI guideline EP15-A3 [4]. Low, medium, and high levels of the quality control materials were provided by the manufacturer. The control materials were measured in five replicates a day for five days. Repeatability and within-laboratory imprecision at each level were assessed. Within-laboratory imprecision was compared to the desirable imprecision specification (15%) recommended by Apple et al. [5]. In addition, two K₂EDTA-anticoagulated whole blood samples near a clinical decision level of 100 pg/mL were measured in 20 replicates a day per sample to obtain repeatability since BNP should be measured within 4 h of collection at room temperature [5].

2. Linearity

Linearity was evaluated according to CLSI guideline EP6-A [6]. The claimed measuring interval ranged from 50 pg/mL to 2,000 pg/mL. Two patient samples with low (L) and high levels (H) were mixed as follows: 0.75*L+0.25*H, 0.5*L+0.5*H, 0.25*L+0.75*H. Each sample was measured three times using the SelexOn, and the mean values were compared to the expected values. Linear fit, second order, and third order polynomial regression analyses were performed. The regression equation was obtained (y, measured concentration; x, expected concentration) using a linear fit model.

3. Method comparison

Method comparison was performed according to CLSI guideline EP9-A3 [7]. Two BNP assays, the ADVIA Centaur^® XPT BNP (Centaur BNP) (Siemens Health Care Diagnostics Inc., Tarrytown, NY, USA) and the Triage^® BNP test (Triage BNP) (Alere, San Diego, CA, USA), were selected. The Centaur BNP is an automated sandwich immunoassay using EDTA plasma for central laboratory testing. The Triage BNP is a uorescence immunoassay using EDTA plasma or whole blood for POC testing. K₂EDTA-anticoagulated whole blood in a plastic tube was used to perform the SelexOn BNP and the Triage BNP, and the corresponding EDTA plasma was used to perform the Centaur BNP. Regression equations between the compared assays were obtained using the Passing-Bablok fit.

4. Reference interval verification

A total of 120 EDTA whole blood residual samples obtained from individuals (aged from 22 years to 75 years; mean, 57 years), including 48 males (aged from 24 years to 75 years; mean, 60 years) and 72 females (aged from 22 years to 74 years; mean, 54 years) who participated in a health examination program without a medical history of cardiovascular disease or renal disease, were included to verify the reference interval (RI) of 100 pg/mL as a clinical decision point claimed by the manufacturer [1, 3, 8].

5. Statistics

Analyze-it version 5.20 (Analyze-it Software, Ltd., Leeds, UK) was used to analyze imprecision, linearity, and method comparison.

1. Imprecision

The repeatabilities were 15.8%, 9.0%, and 8.1% for three mean BNP control levels of 89 pg/mL, 481 pg/mL, and 1,202 pg/mL, respectively (Table 1). The total CV was 20.3% at 89.44 pg/mL, exceeding the desirable imprecision specification (15%), 9.0% at 480.71 pg/mL and 8.1% 1201.84 pg/mL, which met the desirable imprecision specification advised by Apple et al. [5]. However, mean BNP levels in two clinical samples were 91 pg/mL and 116 pg/mL, with repeatabilities of 4.6% and 6.9%, which were lower than 15.8% of the 89 pg/mL for the control material.

Table 1 . Precision of the SelexOn^TM B-type natriuretic peptide (SelexOn BNP) assay

Material	Mean BNP (pg/mL)	Repeatability		Within-laboratory
		SD	CV	SD	CV
Control 1*	89	14	15.8%	18	20.3%
Control 2	481	43	9.0%	64	13.3%
Control 3	1,202	98	8.1%	124	10.3%
Sample 1†	91	4	4.6%	-	-
Sample 2	116	8	6.9%	-	-

^*CLSI guideline EP15-A3 (five repeats a day for five days) for control materials.

^†20 repeats a day for the K₂EDTA-whole blood samples.

2. Linearity

Linearity was observed from 56 pg/mL to 1544 pg/mL (Fig. 2). The regression equation between the measured (y) and the assigned values (x) was y=16.38 + 0.9728x based on a linear fit (R²=0.992, recovery rate: 95.8%-112.6%). The expected bias for non-linearity between the linear fit and the nonlinear (third order polynomial) fit was -20.6% (95% CI, -78.3% to 37.1%), 8.5% (0.1% to 16.8%), -1.1% (-4.6% to 2.4%), -3.9% (-6.9% to -0.8%), and 1.9% (-0.2% to 4.0%) at 56 pg/mL, 428 pg/mL, 800 pg/mL, 1,172 pg/mL, and 1,544 pg/mL, respectively.

Figure 2. Linearity of the SelexOn^TM B-type natriuretic peptide assay. The red and green regression lines represent the linear and polynomial fits, respectively. The regression equation between the measured (y) and the assigned values (x) was y=16.38+0.9728x based on a linear fit (R²=0.992).

3. Method comparison

A total of 100 patient specimens within the measuring range instructed by the SelexOn BNP manufacturer (50 pg/mL to 2,000 pg/mL) were available for method comparison (Table 2 and Fig. 3). The SelexOn BNP (y) showed correlation coefficients of 0.930 with the Centaur BNP (x) and 0.845 with the Triage BNP (x). The regression equations were y=0.9706x-21.68 with the Centaur BNP and y=0.7600x+0.0506 with the Triage BNP, respectively. The predicted mean difference (%) ((y–x)/y×100) of the SelexOn BNP (y) was -24.63% at 100 pg/mL and -8.63% at 400 pg/mL, respectively, compared to the Centaur BNP (x). The predicted mean difference (%) of the SelexOn BNP (y) was -24.0% at 100 pg/mL and -24.0% at 400 pg/mL, respectively, compared to the Triage BNP (x).

Table 2 . Method comparison of the SelexOn^TM BNP with the Centaur^® BNP and the Triage^® BNP assays (N=100)

Methods compared (y vs. x)	Intervals measured (y, pg/mL)	r	Passing-Bablok fit equation	y-intercept (95% CI)	Slope (95% CI)
SelexOn vs. Centaur	52-1995	0.930	y=0.9706x-21.68	-49.27 to -5.654	0.8969 to 1.108
SelexOn vs. Triage	52-1995	0.845	y=0.7600x+0.0506	-22.60 to 30.62	0.6217 to 0.8576

Figure 3. Scatter plot of method comparison for the SelexOn BNP, the Centaur BNP, and the Triage BNP assays. The red line represents the linear regression equation by the Passing-Bablok fit (left) and difference plots (right). The SelexOn BNP (y) regression equation was (A) y=0.9706x-21.68 with Centaur BNP (x) (r=0.930), and (B) y=0.7600x+0.0506 with Triage BNP (x) (r=0.845).

4. Verification of reference interval

Out of the 120 samples, 95.0% were less than 50 pg/mL, 96.3% were less than 75 pg/mL, and 97.5% were less than 100 pg/mL. The distribution of BNP results by age and gender have been summarized in Table 3. Of the 120 individuals across all ages, 45 males and 69 females showed BNP results below 50 pg/mL.

Table 3 . Distribution of BNP results from non-heart failure samples participating in the health examination program (N=120)

Age group	All (%)	All ages (N=120)		<45 years (N=22)		45-54 years (N=22)		55-64 years (N=37)		65-75 years (N=39)
BNP (pg/mL)		Male	Female	Male	Female	Male	Female	Male	Female	Male	Female
<50	114 (95)	45	69	3	18	9	10	14	22	19	19
50-74	1 (0.8)	1	0	-	-	-	-	-		1
75-100	2 (1.7)	1	1	1	-	-	1	-
101-160	3 (2.5)	1	2	-	-	1	1	-	1

We evaluated the analytical performance of the SelexOn BNP for imprecision, linearity, method comparison, and verification of reference interval based on CLSI guidelines.

The total imprecision in the control material ranged from 10.3% to 20.3%, and the CV (20.3%) at 89 pg/mL exceeded 15% of a desirable specification for imprecision [5, 9]. Although repeatability was only evaluated in consideration of BNP stability in whole blood [5], similar low concentration clinical samples had lower repeatability (4.6%-6.9%) than control (15.8%). This may be because the substrate of the manufacturer-provided control material was not the same as that of the clinical sample. On the basis of the intra-individual biological variation (CV_I), the goal for imprecision should be less than or equal to half of the CVI [5, 10]. BNP has a very wide intra-individual biological variation (CVI, 30%-50%) like other hormones, and the total imprecision of the SelexOn BNP is considered less than the minimum requirement of analytical imprecision goal (25%) [9-12, 14].

BNP usually increases in heart failure, and patients with a higher BNP value (≥100 pg/mL) in the acute setting should be evaluated via further work-up to rule out heart failure [1, 3, 13]. If the BNP level rises above 35 pg/mL in patients with a suspected heart failure in the non-acute setting, additional echocardiography should be recommended for heart failure [13]. BNP can also increase in various cardiac and non-cardiac conditions like acute coronary syndrome, acute myocardial infarction, pulmonary embolism, hyperthyroidism, renal failure, or septic shock [11, 13]. Therefore, BNP is reportedly a useful biomarker for ruling out rather than diagnosing heart failure because BNP at clinical decision levels has high negative predictive values in both the acute and the non-acute settings [13]. In addition, High BNP during heart failure follow-up and monitoring is also reportedly associated with adverse outcomes [13].

Here, the correlation between the SelexOn BNP with the other two BNP assays was not high (r=0.930 for the Centaur BNP; r= 0.845 for the Triage BNP), and the predicted mean difference (%) of the SelexOn BNP at clinical decision levels the 100 pg/mL and 400 pg/mL was up to 25% lower than that in the two comparable assays. The substantial differences between methods agree with the previous BNP immunoassay comparison studies [1, 2, 10, 14, 15]. Similar results have been reported in previous studies comparing the Triage and the Centaur BNPs [10, 14]. The main putative cause of the non-harmonized BNP assays is that there are no primary reference materials for BNP measurements; many manufacturers use different antibodies and calibrators for BNP immunoassays [1, 2, 16]. Therefore, it is important to identify and harmonize the systematic differences in BNP assays, and more appropriate clinical decision thresholds may be necessary for the individual BNP assay [2, 14]. If the BNP assay has a negative systematic bias, it may be interpreted as a false negative in disease differential diagnosis at low concentrations near 100 pg/mL. Our results suggest that calibration adjustments may be necessary to improve the negative systematic bias of the SelexOn BNP. Variations between lots can have a significant impact on the test results for cartridge-type POT tests, but the lot-to-lot variability was not evaluated in the present study.

In the reference interval verification, most (95% of the 120 individuals) with the SelexOn BNP exhibited values below 50 pg/mL. There was no difference observed between gender and age in the ratio outside the reference interval for the clinical decision point, but further study will be needed for stratified reference intervals. The BNP reference intervals for the healthy elderly are higher than the clinical decision level used for the heart failure guidelines [13, 17, 18]. Therefore, for the elderly, data should be interpreted in relation with the clinical symptoms [17, 18].

In summary, the SelexOn BNP using EDTA whole blood was developed as a POC test for differential diagnosis or treatment monitoring for acute heart failure. However, harmonization should be further strengthened to ensure that clinical decision values are compatible with other BNP methods.

This research was supported by a grant of the Korea Health Technology R&D Project through the Korea Health Industry Development Institute (KHIDI), funded by the Ministry of Health & Welfare, Republic of Korea (grant number: HI16C0821).

This research was supported by a Grant of Translational R&D Project through the Institute for Bio-Medical Convergence, Incheon St. Mary’s Hospital, The Catholic University of Korea.

We are grateful to Sang Gil Bae, Hyeji You, and Seon-Hwa Kim for excellent technical assistance.