Reference Interval for the Axis-Shield Clinical Chemistry Heparin-Binding Protein Assay

The newly developed Axis-Shield clinical chemistry heparin-binding protein (HBP) assay (Axis-Shield Diagnostics Ltd., Dundee, Scotland) can be applied to fully automated platforms. We aimed to establish a reference interval (RI) of HBP using the Axis-Shield HBP assay, and to evaluate the analytical performance of this assay. An RI was established in 212 sodium citrated plasma samples using the non-parametric method (2.5th and 97.5th percentiles). Precision, linearity, and carry-over were evaluated according to the Clinical and Laboratory Standards Institute guidelines. The RI of HBP was between 5.3 ng/mL and 171.0 ng/mL, which could be applied regardless of gender and age. Percentage coefficients of variations (%CVs) of repeatability and within-laboratory precision were 4.9% and 6.3%, respectively, for low-concentration control and 1.6% and 3.0%, respectively, for high-concentration control. The linearity was excellent (coefficient of determination (R2) = 0.99), and the carry-over rate was negligible (0.05%). This is the first study to establish an RI of HBP using the newly developed and fully automated Axis-Shield HBP assay. The Axis-Shield HBP assay showed an acceptable level of analytical performance and could be used to measure HBP concentrations effectively in routine clinical practice. Further studies are awaited to evaluate the clinical utility of HBP using this automated assay.


Introduction
Sepsis is a major health problem that increases mortality and critical illness, and it is important to identify patients with sepsis as early as possible [1,2]. According to the Third International Consensus Definition for Sepsis and Septic Shock (Sepsis-3), the sequential (sepsis-related) organ failure assessment (SOFA) scoring system is used to identify lifethreatening organ dysfunction in sepsis [1]. However, the SOFA scoring system requires clinical judgement, such as the requirement for adrenergic support and Glasgow coma scale scores, which can be subjective and different across institutions [1]. To identify patients with sepsis early, it is necessary to supplement or replace the SOFA scoring system using objective biomarkers [2]. Numerous biomarkers for sepsis have been evaluated, but none are routinely used in clinical practice [3].
Recently, an assay for measuring HBP concentration, the Axis-Shield clinical chemistry HBP assay (Axis-Shield HBP assay; Axis-Shield Diagnostics Ltd., Dundee, Scotland), was developed. It is the first assay that can be applied to fully automated platforms and is based on latex immunoturbidimetry, not an ELISA [19]. To the best of our knowledge, no studies have been conducted using the fully automated Axis-Shield HBP assay. The RI of HBP has not been established using the Axis-Shield HBP assay, and the analytical performance of this assay has also not been evaluated. Considering that the principle of the Axis-Shield HBP assay is different from that of the ELISA, it is important to establish an RI of HBP using the Axis-Shield HBP assay and to compare the HBP concentrations obtained when using the Axis-Shield HBP assay versus the ELISA. In this study, we aimed to establish an RI of HBP using the Axis-Shield HBP assay. We also evaluated its analytical performance according to the Clinical and Laboratory Standards Institute (CLSI) guidelines for the first time.

Study Population
A total of 212 blood samples were collected from 212 apparently healthy Korean adults who visited the Konkuk University Medical Center (KUMC), Seoul, Korea, for a general medical check-up in December 2017. This study protocol was approved by the Institution Review Board of KUMC (KUH1200033 and 10 December 2013) before recruiting the first sample. Informed consent from the subjects was not required because residual samples were collected after performing the requested test. This study required neither study-intended blood sampling nor other interventions. The healthy Korean adults were included on the basis of their physical and laboratory findings determined by a medical chart review according to the CLSI guidelines [20]. They were considered healthy without any evidence of medical problems, especially inflammation, infection, or sepsis. The individuals who showed abnormal laboratory findings as follows were excluded prior to reaching 212 individuals: white blood cells, <4.0 or >10.0 × 10 9 /L; C-reactive protein, >0.3 mg/dL. Sodium citrated plasma samples were prepared and stored at -80 • C until use. The characteristics of the study population are presented in Table 1.

Axis-Shield Clinical Chemistry HBP Assay
The Axis-Shield HBP assay measured HBP concentrations based on a turbidimetric reaction between HBP and the avian HBP polyclonals. The agglutination between HBP and the HBP antibody bound to polystyrene particles is detected as an absorbance change on an automated clinical chemistry analyzer. The magnitude of the absorbance change is proportional to the HBP concentration in the sample. Only sodium citrated plasma samples are suitable for this assay, which can be applied to many chemistry analyzers [19]. In this study, the Axis-Shield HBP assay was performed using a Toshiba 200FR NEO (Toshiba Medical System Co., Tochigi-ken, Japan), according to the manufacturer's instructions.
For quality control, the Axis-Shield HBP assay uses the Axis-Shield clinical chemistry HBP controls (Axis-Shield Diagnostics Ltd.), which are recommended to be run daily in duplicate. The 95th percentile reference limit suggested by the manufacturer was 21.44 ng/mL (90% confidence interval (CI), 19.3-23.5 ng/mL), which was established in 53 individuals (19 men, 28 women, and 6 normal samples with unknown demographic details). According to the manufacturer's instructions, the limit of detection was determined to be 8.0 ng/mL. Percentage coefficients of variations (%CVs) of repeatability and within-laboratory precision were 7% or less for all measured HBP concentrations. The analytical measurement range and linearity was demonstrated from 8.4 to 337.0 ng/mL, and the carry-over was not observed. This assay showed good correlation with a commercially available ELISA in the comparison study [19].

Statistical Analysis
An RI of HBP was established according to the CLSI EP28-A3c guidelines [20]. The distribution of HBP concentration was examined for normality using the Kolmogorov-Smirnov Z-test, and the outliers were checked and excluded using the Dixon-Reed test [21]. The RI was determined using the non-parametric method (2.5th and 97.5th percentiles). The 90% CI was also calculated non-parametrically for each reference limit [20]. The study population was divided into two groups depending on gender (men, n = 93; women, n = 119) and six groups depending on age (years; 19-29, n = 45; 30-39, n = 40; 40-49, n = 37; 50-59, n = 38; 60-69, n = 34; and 70-79, n = 18). The median value of HBP concentration was compared between genders using the Mann-Whitney U test and between ages using the Kruskal-Wallis test.

Analytical Performance
The analytical measurements for evaluating precision ranged from 16.7 to 21.2 ng/mL for low-concentration control and from 98.3 to 110.9 ng/mL for high-concentration control. The %CVs of repeatability for low-and high-concentration controls were 4.9% and 1.6%, respectively. The %CVs of within-laboratory precision for low-and high-concentration controls were 6.3% and 3.0%, respectively. All %CVs of repeatability and within-laboratory precision were excellent for low-and high-concentration controls. There were no apparent outliers or drift capable of distorting the precision analysis ( Figure 2). The analytical measurements for evaluating linearity ranged from 1.0 to 342.5 ng/mL, and the linearity was excellent with the coefficient of determination (R 2 ) = 0.99 (Figure 3). Percentage recoveries of serially diluted samples were acceptable, ranging from 96.2% to 100.5%. The carry-over rate was negligible with 0.05%.

Discussion
The newly developed Axis-Shield HBP assay is a fully automated assay that has not been evaluated so far. Many previous studies have reported measuring HBP concentration using the ELISA [7][8][9][10][11][12][13][14][15][16][17]. To the best of our knowledge, neither the clinical cut-off value nor the RI of HBP using the Axis-Shield HBP assay have been established. This is the first

Discussion
The newly developed Axis-Shield HBP assay is a fully automated assay that has not been evaluated so far. Many previous studies have reported measuring HBP concentration using the ELISA [7][8][9][10][11][12][13][14][15][16][17]. To the best of our knowledge, neither the clinical cut-off value nor the RI of HBP using the Axis-Shield HBP assay have been established. This is the first study worldwide, which established an RI of HBP using the fully automated Axis-Shield HBP assay and evaluated its analytical performance based on the CLSI guidelines.
The established RI for HBP in Korean adults was from 5.3 to 171.0 ng/mL, which could be applied regardless of gender and age. The reference limit as the 95th percentile established in our study population was 131.9 ng/mL, which was much higher than the reference limit suggested by the manufacturer. HBP concentration could increase not only in patients with infection and sepsis but also in patients with non-infectious diseases such as ST-segment elevation myocardial infarction [28]. Although we included apparently healthy individuals, some asymptomatic patients with non-infectious diseases may have been included in this study and may have influenced the establishment of a high reference limit. However, we conducted this study on individuals without critical illness that could affect HBP concentrations through a thorough medical chart review of clinical and laboratory data [2,[7][8][9][10][11][12][13][14][15][16]28]. In addition, the reference limit suggested by the manufacturer was established in a small cohort of 53 samples without information on ethnicity or skewness of the distribution [19]. According to the manufacturer's instructions, the Axis-Shield HBP assay was developed for in vitro diagnostic use, and each laboratory needs to evaluate and establish its own RI of HBP in large cohorts [19]. In this study, there were some skewed populations with high HBP concentrations even in healthy individuals ( Figure 1). This result implies that HBP concentration may be high without sepsis or other known ill conditions. Several biomarkers, including biologically active adrenomedullin and proenkephalin, have been reported that could be objective and useful markers to predict severity, organ failure, and mortality in septic patients [29,30]. Since the concentration of these biomarkers vary among individuals, it is mandatory to understand the biological variation (BV) of a biomarker in healthy individuals [31,32]. However, the BV of HBP has not been assessed [33]. Further prospective studies on the RI and BV of HBP in large cohorts are awaited.
Compared with a clinical cut-off value of 28.1 ng/mL suggested in a previous study using the ELISA (manufactured by Axis-Shield), approximately 40% of our study population had a higher HBP concentration [8]. In the previous study, all individuals in the healthy group had an HBP concentration of <10 ng/mL, but the healthy group consisted of a small number of 56 individuals [8]. The median value of HBP concentration in our study population was 23.5 ng/mL, which was not significantly different from the suggested cut-off value. The clinical cut-off values for sepsis differed between methods with the same ELISA principle [7][8][9]. There is no consensus on the universal and disease-specific cut-off values of HBP, and the cut-off values suggested in previous studies were based on a small number of samples [7][8][9][10][11][12][13][14][15][16][17]. The previous studies were conducted to compare the difference in HBP concentrations between small groups using the ELISA, not to establish the RI of HBP [7][8][9][10][11][12][13][14][15][16][17]. We focused only on establishing the RI of HBP using the Axis-Shield HBP assay. It is necessary to compare the RI established in this study with the HBP concentration measured by the Axis-Shield HBP assay in septic patients. In addition, it should be considered that HBP concentration could increase in non-infectious disease [28]. A clinical cut-off value is appropriate in large-scale studies, but there are no studies yet on the cut-off value of HBP. It is too early to determine the universal cut-off value, as previous studies have used only small cohorts. Even in patients with sepsis or septic shock, the concentration of procalcitonin, a well-known biomarker for sepsis, might be lower than the clinical cut-off value of 0.5 ng/mL [34]. Sepsis cannot be identified with a single biomarker, and a multi-marker approach is emphasized [35,36].
In this study, the Axis-Shield HBP assay showed excellent repeatability and withinlaboratory precision with less than 10% CV. It also showed excellent linearity and an acceptable carry-over rate that had less than 1.0% cut-off limit. These results are similar to the analytical performance provided by the manufacturer [19]. This indicates that the Axis-Shield HBP assay is acceptable for use in measuring HBP concentration in clinical laboratories.
The strength of this study is to provide the fundamental data of the Axis-Shield HBP assay for further studies and clinical use. On the other hand, this study has several limitations. First, this study was conducted on individuals who were considered healthy only by a medical chart review among those who visited the KUMC for a general medical check-up. Individuals with critical illnesses, such as AKI and ARDS that could affect the HBP concentration were excluded. However, factors such as underlying health conditions and drug use could not be considered in establishing the RI of HBP, as undisclosed health conditions of the subjects could not be checked. Second, the number of individuals in each gender and age was less than 120 in this study. The CLSI guidelines have recommended that at least 120 individuals for each gender or other subclass are required to determine whether to partition RIs by calculating the statistic z-value [20]. Inevitably, the median value of HBP concentration for each gender and age was compared in this study. Although the statistic z-value was not calculated, it would be acceptable to apply a single RI regardless of gender and age. Third, we established the RI only for Korean adults. Previous studies have reported that HBP concentrations of urine and sputum in children are associated with urinary tract infections and pulmonary inflammation, respectively; HBP could be a promising biomarker even in children [37][38][39]. Although there were no statistically significant differences in HBP concentration between ages in Korean adults in our study, it is necessary to evaluate the age-specific RI in large cohorts, including children. Finally, we did not compare the HBP concentration between the Axis-Shield HBP assay and the ELISA. Procalcitonin has also been reported to show a modest bias between different reagents and analyzers [40]. Considering that the principle is different between the Axis-Shield HBP assay and the ELISA, the HBP concentration could show a bias between them. It is necessary to further evaluate the correlation between the Axis-Shied HBP assay and the ELISA for HBP. Even though the two methods (immunoturbidimetry and ELISA) show good agreement, there may be a difference in values [41]. If HBP is to be used as a biomarker with a critical threshold for sepsis or other infections, standardization should be achieved between methods with different principles, and even between different reagents and analyzers within the same principle.
In conclusion, we established an RI of HBP using the fully automated Axis-Shield HBP assay, and the RI could be applied regardless of gender and age. The Axis-Shield HBP assay showed an acceptable level of analytical performance according to the CLSI guidelines. It could be useful to measure HBP concentrations in routine clinical practice. Further studies are awaited to evaluate clinical utility of HBP using this automated assay in various critically ill patients. It should be performed after comparing the HBP concentrations of the automated assay and the ELSIA.

Data Availability Statement:
The data presented in this study are available from the corresponding author on reasonable request.