Evaluation of the Atellica COAG 360 coagulation analyzer in a specialized coagulation laboratory

Abstract Background Diagnosis of bleeding disorders includes correct analysis of coagulation factors VIII, IX, XI, XII, XIII, II, V, VII, and X and von Willebrand antigen and activity. The aim of this study was to evaluate the analytical performance of the Atellica COAG 360 analyzer in a specialized coagulation laboratory with focus on specific coagulation parameters involved in the diagnosis of bleeding disorders. Methods Verification included assessment of precision, reference interval, and method comparison according to local guidelines. For FVIII (Chromogenix) and FIX (Rossix), extended verifications were performed with additional assessment of linearity, detection limit, and comparability to BCS‐XP. Results The precision was below 5% (normal levels) and below 10% (abnormal levels) and either improved or similar when compared to expected target values from a BCS‐XP. The locally established reference range agreed well (≥80% of measured values within manufacturer's assigned ranges) for most of the methods. The lower limit of quantification was calculated to below 0.01 IU/ml for FVIII chromogenic (Chromogenix) and FIX chromogenic (Rossix), both with acceptable linearity. Bland–Altman analyses revealed generally good agreement between Atellica COAG 360 and BCS‐XP in the determination of coagulation parameters, and differences between the two instruments did not result in any diagnostic change. Conclusions The results of the evaluation show that the Atellica COAG 360 analyzer performs as expected to target values and equivalent to BCS‐XP for the diagnosis of bleeding disorders in a specialized coagulation laboratory providing service to a hemophilia treatment center (HTC).

any bleeding, deficiencies of FII, FV, FVII, FX, and FXIII give an increased bleeding tendency. 1 Correct diagnosis of von Willebrand disease involves many different methods and this study will only focus on the methods that can be performed on an automated analyzer such as von Willebrand factor (VWF) activity, based on the glycoprotein Ib-containing gain-of-function mutation (VWF:GPIbM), and antigen (VWF:Ag). The recommendation for laboratories that perform diagnosis of hemophilia A and B is to have two methods available for factor activity measurements covering both one-stage assay (OSA) and chromogenic substrate assay (CSA). [2][3][4] In addition, the need to distinguish severe from moderate hemophilia requires a method with low detection limit, below 0.01 IU/ml factor activity. 3 The need for CSA and OSA for factor activity measurement of FVIII and FIX is also demonstrated by assay discrepancy seen for extended half-life (EHL) products, which was covered previously. [5][6][7][8] While high-throughput and short turnaround times are essential requirements for coagulation analyzers in routine laboratories, the requirements for specialized laboratories that provide services to a hemophilia treatment center (HTC) are somewhat different. Besides correct phenotype and severity classification for hereditary bleeding disorders, testing also involves diagnosing acquired or treatment-induced bleeding conditions, some of which are considered critical hemostasis tests (i.e., anti-FXa activity for heparin/low molecular weight heparin [LMWH]). 9 The Atellica COAG 360 analyzer from Siemens Healthineers is a fully automated hemostasis analyzer. Besides performing coagulation testing using different techniques such as clotting, immunologic assay, luminescent oxygen channeling (LOCI) assay, and aggregation testing, it features a HIL check (hemolysis, icterus, and lipemia) and storage of reagents in a cooled compartment. Hörber and colleagues evaluated the Atellica COAG 360 analyzer in a central laboratory and concluded that it provided high analytical performance. 10 The performance of OSA and CSA methods has also been tested for hemophilia replacement therapy with different EHL products using Atellica COAG 360. 5,11 The aim of the present study was to evaluate the analytical performance of the Atellica COAG 360 analyzer in a specialized coagulation laboratory with focus on specific coagulation parameters involved in the diagnosis of hereditary and acquired bleeding disorders. The methods provided by Siemens on Atellica COAG 360, included in this study, covers chromogenic assays for FVIII, FXIII, and anti-FXa activity (LMWH); OSA for FVIII, FIX, FXI, FXII, FII, FV, FVII, and FX; immunoturbidimetric assay for VWF:Ag; and latexbased VWF platelet-binding assay (in the absence of ristocetin) for VWF:GPIbM. 12 Methods not provided by Siemens, also adopted on the Atellica COAG 360 analyzer, were chromogenic assays for FVIII and FIX using third-party reagents that required laboratory developed test (LDT) and an extended verification.
Verification of precision on at least two levels of control material and reference range was performed for all methods. Comparability to the previous BCS-XP method, linearity and detection limit were performed on selected methods in accordance with international guidelines and local regulatory requirements. 13 2 | MATERIAL S AND ME THODS

| Sample collection
This was a method comparison in which samples from routine clinical follow-up and spiked normal plasma were analyzed in coagulation assays for comparability studies. Ethical approval was obtained from the local Ethics Board (Dnr2015/886). Normal pooled platelet poor plasma was obtained in house from male (26%) and female donors (74%) (n = 50, aged 22-70) with informed consent. Blood samples (3.2% sodium citrate, 109 mmol/L) were collected and centrifuged for 20 min at 2000 g at room temperature. Plasma supernatants were frozen and stored at −70°C until analysis.

| Study design
Verification included assessment of precision, reference interval, and method comparison according to local guidelines from the Swedish national accreditation body, SWEDAC, using Swedac DOC 01:55, 2011-08*10 release 4 for ISO15189 accreditation and when applicable to target values obtained for the previous BCS-XP method. 14 For FVIII and FIX chromogenic assays that required a LDT, an extended verification was performed with the above verification requirements and additional assessment of linearity and detection limit according to guidelines. 3 Comparability was assessed to BCS-XP for the anal-

| Reagents, calibrators, and controls
Reagents and calibrators were used on the Atellica COAG 360 analyzer (Siemens Healthineers) according to the instructions provided by the manufacturer (Table 1). Factor VIII chromogenic activity was measured using two different CSA methods; CSA-1 (see below) and CSA-2 (Siemens). Methods not provided by Siemens and thus re-

| Assessment of linearity and detection limit
A linearity test was performed using manual dilution of the normal pooled plasma in FVIII-deficient plasma for the FVIII CSA-1 method and dilution of standard human plasma (Siemens) in FIX-deficient plasma (George King Bio Medical) for the FIX CSA method. Samples were measured in duplicates and a comparison of the theoretical assigned value and measured value was evaluated using linear regression analysis, with r 2 as the coefficient of correlation. Linearity was assumed as acceptable when r 2 > 0.998. 3 The detection limit for FVIII and FIX CSA assays was validated by measuring a blank sample, containing assay buffer, 20 times. Since most of the measurement would yield a result below the detection limit, the raw value (absorbance/min) was used in order to calculate the lower limit of detection (LLOD) as the mean + 3 SD. The lower limit of quantification (LLOQ) was calculated as three times the value of LLOD. 3 A LLOQ < 0.01 IU/ml was considered acceptable for hemophilia severity classification.

| Assessment of reference interval, precision, and comparability
Reference intervals were verified locally and performed by measur- Reference range was reported as mean ± 2 standard deviations (SD) after normality test using D'Agostino and Pearson and Shapiro-Wilk tests (data not shown). Precision was determined by measuring control samples on a minimum of two levels per method, five times during 1 day over 5 days, yielding at least 25 measured results for each level and reported with a total coefficient of variation (CV%). At least two control levels were used for each calibration curve. A target CV% ≤ 5.0% on normal levels and ≤10.0% on abnormal levels were used. 16 The target CV was chosen based on Marlar et al, i.e., the CV is usually accepted at 3%-6% for clotting, chromogenic, and most immunologic analytes but never more than 10%. 16 Comparability was based on the performance on the previous BCS XP and performed on selected methods. Bland-Altman analysis was used for assessing bias. A bias <10% for 95% of samples was considered acceptable. 13 Regression analysis was also performed, with a target of slope between 0.90 and 1.10 and Pearson r 2 ≥ 0.95 for the correlation assessment. 13 A correlation study of anti-FXa activity (LMWH) assay was conducted using the in-house normal pooled plasma spiked with different amounts (1.0, 0.5, and 0.25 IU/ml) of Fragmin ® (Pfizer).

| Assessment of linearity and detection limit
For assays with reagents not provided by Siemens, i.e., CSA methods for FVIII and FIX, an assessment of linearity and detection limit were completed. Linearity was accepted with r 2 ≥ 0.998 for FVIII CSA-1 and FIX CSA ( Figure 1A,B). The detection limit, i.e., LLOQ, was calculated to <0.01 IU/ml (<1%) for FVIII CSA-1 and FIX CSA, 0.004 and 0.009 IU/ml, respectively.

| Reference interval
The reference interval was verified locally for all methods using 30-50 individual donors,

| Precision
In general, the total coefficient of variation (CV) reached the assigned target values, below 5.0% for normal levels and below 10.0% for abnormal levels except for 10.1% at 0.05 IU/ml FVIII OSA and 5.6% at 0.9 IU/ml FXIII, see Table 3. For most of the methods, CVs  Table 2). Although one result with high discrepancy for VWF:Ag and one for VWF:GPIbM, these results were at values above 2 IU/ml and thus did not change the diagnosis of the patients ( Figure 2E,F). Correlation studies on Quick's PT revealed good correlation (r 2 > 0.95, Table 2) and Bland-Altman plots revealed relative high bias (−3.23; 27%, Table 2). For FII, FV, FVII, and FX OSA, a correlation study was not performed due to lack of patient samples over the measuring range. However, for the anti-FXa activity (LMWH) assay, spiked samples showed acceptable agreement with low bias (−0.02; −1.2%) when compared to the assigned value (Table 2).
Regression analysis revealed optimal comparability (slope 0.90-1.10) for the majority of the assays and suboptimal comparability for FVIII CSA-1 and VWF:Ag (slopes at 0.84 and 0.77, respectively), see Table 2.    The linearity and detection limits of the chromogenic assays (FVIII and FIX LDTs) were all approved. Unfortunately, the chromogenic assay provided by Siemens (here stated as CSA-2) has a reported detection limit of 0.035 IU/ml that was also confirmed in our setting (0.024 IU/ml data not shown) which did not meet our requirement for the detection limit, <0.01 IU/ml. 12 As a consequence, in our laboratory that provides service to a HTC, we need to use a LDT method with third-party reagent (CSA-1) for chromogenic FVIII activity measurements in order to differentiate between the moderate form and severe form. 18

TA B L E 3 (Continued)
with the methods and reagents tested for the diagnosis of bleeding disorders. Most importantly, the small differences between the compared instruments did not result in any diagnostic change for the patients.

ACK N OWLED G EM ENTS
The authors thank Heidi Isacson, Elena Mikheeva, and Fikrije Rakip for excellent technical assistance.

CO N FLI C T O F I NTE R E S T
The authors state that there are no conflicts of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
Available upon request.