Clin Chem Lab Med 2014; 52(11): 1657–1665

Magdalena Krintus*, Marek Kozinski, Pascal Boudry, Nuria Estañ Capell, Ursula Köller, Karl Lackner, Guillaume Lefèvre, Lieselotte Lennartz, Johannes Lotz, Antonio Mora Herranz, Mads Nybo, Mario Plebani, Maria B. Sandberg, Wolfgang Schratzberger, Jessie Shih, Øyvind Skadberg, Ahmed Taoufik Chargui, Martina Zaninotto and Grazyna Sypniewska

European multicenter analytical evaluation of the Abbott ARCHITECT STAT high sensitive troponin I immunoassay

Abstract

Background: International recommendations highlight the superior value of cardiac troponins (cTns) for early diagnosis of myocardial infarction along with analytical requirements of improved precision and detectability. In this multicenter study, we investigated the analytical per-formance of a new high sensitive cardiac troponin I (hs-cTnI) assay and its 99th percentile upper reference limit (URL).

Methods: Laboratories from nine European countries evaluated the ARCHITECT STAT high sensitive troponin I (hs-TnI) immunoassay on the ARCHITECT i2000SR/i1000SR immunoanalyzers. Imprecision, limit of blank (LoB), limit of detection (LoD), limit of quantitation (LoQ) linearity of dilution, interferences, sample type, method comparisons, and 99th percentile URLs were evaluated in this study.Results: Total imprecision of 3.3%–8.9%, 2.0%–3.5% and 1.5%–5.2% was determined for the low, medium and high controls, respectively. The lowest cTnI concentration cor-responding to a total CV of 10% was 5.6 ng/L. Common interferences, sample dilution and carryover did not affect the hs-cTnI results. Slight, but statistically significant, differences with sample type were found. Concordance between the investigated hs-cTnI assay and contempo-rary cTnI assay at 99th percentile cut-off was found to be 95%. TnI was detectable in 75% and 57% of the apparently healthy population using the lower (1.1 ng/L) and upper (1.9 ng/L) limit of the LoD range provided by the ARCHI-TECT STAT hs-TnI package insert, respectively. The 99th percentile values were gender dependent.Conclusions: The new ARCHITECT STAT hs-TnI assay with improved analytical features meets the criteria of high sen-sitive Tn test and will be a valuable diagnostic tool.

Keywords: analytical evaluation; analytical performance; high sensitive troponin; 99th percentile upper reference limit.

DOI 10.1515/cclm-2014-0107Received January 29, 2014; accepted May 4, 2014; previously pub-lished online June 4, 2014

IntroductionThe Third Universal Definition of Myocardial Infarction highlights the importance of cardiac troponin (cTn) as the preferred biomarker for the detection of myocardial

*Corresponding author: Magdalena Krintus, PhD, Department of Laboratory Medicine, Nicolaus Copernicus University, Collegium Medicum, 9 Sklodowskiej-Curie Street, 85-094 Bydgoszcz, Poland, Phone: +48 52 5853602, Fax: +48 52 5853603, E-mail: krintus@wp.plMarek Kozinski: Department of Principles of Clinical Medicine, Nicolaus Copernicus University, Collegium Medicum in Bydgoszcz, Bydgoszcz, PolandPascal Boudry: Department of Clinical Biology, CHR Mons-Hainaut, Mons, BelgiumNuria Estañ Capell and Antonio Mora Herranz: Department of Clinical Biochemistry, Dr. Peset University Hospital, Valencia, SpainUrsula Köller and Wolfgang Schratzberger: Institute for Laboratory Diagnostics, Vienna Hospital Association, Hospital Hietzing, Vienna, AustriaKarl Lackner and Johannes Lotz: Laboratory Medicine, University Medical Center, Johannes Gutenberg University Mainz, Mainz, GermanyGuillaume Lefèvre and Ahmed Taoufik Chargui: Department of Biochemistry and Hormonology, AP-HP, Hôpital Tenon, Paris, FranceLieselotte Lennartz: Abbott Laboratories, Wiesbaden, GermanyMads Nybo and Maria B. Sandberg: Department of Clinical Biochemistry and Pharmacology, Odense University Hospital, Odense, DenmarkMario Plebani and Martina Zaninotto: Department of Laboratory Medicine, University Hospital, Padua, ItalyJessie Shih: Abbott Laboratories, Abbott Park, IL, USAØyvind Skadberg: Laboratory of Clinical Biochemistry, Stavanger University Hospital, Stavanger, NorwayGrazyna Sypniewska: Department of Laboratory Medicine, Nicolaus Copernicus University, Collegium Medicum in Bydgoszcz, Bydgoszcz, Poland

Brought to you by | University of MichiganAuthenticated

Download Date | 10/9/14 5:38 PM

1658      Krintus et al.: hs-cTnI evaluation

infarction (MI) with a cTn concentration above the 99th percentile upper reference limit (URL) as the cut-off for diagnosis of MI with a total CV   ≤  10% at this concentra-tion [1, 2]. Evolving recommendations of scientific organi-zations have led to improvements in cTn assays to meet these requirements [1–4]. Only a limited number of con-temporary sensitive cTn assays meet the 10% CV require-ment and as a result, newer high sensitivity cardiac troponin (hs-cTn) assays have been developed to meet these criteria.

Currently, there are no definitive requirements for what defines a high sensitivity assay versus a contem-porary sensitive assay. A recommendation was recently published by the Task Force of International Federation of Clinical Chemistry and Laboratory Medicine (IFCC) on Clinical Applications of Cardiac Biomarkers that suggests that highly sensitive assays have the ability to measure at least 50% and ideally over 95% of healthy individuals above the assay’s limit of detection (LoD) along with an imprecision (expressed as CV) of   ≤  10% at the 99th percen-tile URL [2, 5]. Further support for this recommendation comes from the Study Group on Biomarkers in Cardiol-ogy in their critical clinical concepts regarding analytical characteristics of high sensitive cTn assays [6]. In view of the lack of cTn standardization and harmonization, there is a substantial need for assay-dependent determination of the diagnostic threshold of hs-cTn, essential for the clinical decision making process according to current recommendations.

In the context of the analytical requirements and the expectations of the clinicians, the purpose of our study was to validate analytical performance of the ARCHITECT STAT hs-TnI assay and to verify its 99th percentile URL.

Materials and methodsThe ARCHITECT STAT hs-TnI assay performed on the Abbott ARCHI-TECT i2000SR/i1000SR immunoanalyzers (Abbott Laboratories, Wies-baden, Germany) was evaluated in nine clinical laboratories in the following European countries: Austria, Belgium, Denmark, France, Germany, Italy, Norway, Poland and Spain from January through August 2013.

Assay description and procedureThis assay is a two-step, double-monoclonal chemiluminescent microparticle immunoassay (CMIA) for the detection of cTnI in human serum or plasma. According to the manufacturer’s insert, this assay uses a sample volume of 160 μL. In the first step, cTnI present in the sample binds to the specific mouse monoclonal anti-cTnI (capture)

antibody coated microparticles whose epitope binding specificity is directed against amino acids 24–40 on the TnI protein [7, 8]. In the second step, a specific mouse-human chimeric, monoclonal anti-tro-ponin I acridinium-labeled antibody acts as a detection antibody. This chimeric antibody, which has an epitope binding specificity directed against amino acids 41–49 on the cTnI molecule was designed to mini-mize the susceptibility to interferences by heterophilic antibodies.

Analytical performanceA 5-day imprecision study was conducted at eight sites, according to the National Committee for Clinical Laboratory Standards/Clinical and Laboratory Standards Institute (NCCLS/CLSI) guidelines [9]. Two ARCHITECT runs were performed on each day with a minimum of 2 h separating the runs. Three cTnI control levels (low, medium and high across the range 12–21,000 ng/L) were assayed during the 5-day evaluation in replicates of two on each run for a total of 20 replicates for each control. Mean, standard deviation (SD) and CVs for each site were calculated.

For the precision profile lithium heparin panels were prepared and tested at one site (France) using a high-concentration stock of cTnI derived from pooled clinical samples spiked into a negative serum pool. Samples were tested over 21 days in independent assay runs. Results were pooled and a model fit was created to estimate concentrations at 20%, 15% and 10% CV.

Limit of blank (LoB), LoD and limit of quantitation (LoQ) were verified according to the CLSI guidelines at eight sites [10]. LoB was estimated by measuring 20 replicates of zero calibrator and calcu-lating the mean result and the SD. LoD was determined using 20 replicates of a sample with concentration equal to the claimed LoD and estimation of the proportion of results exceeding the LoB was calculated. For LoQ a minimum of 30 replicates of a sample with a concentration close to the claimed LoQ were run. Dilution linearity of the hs-cTnI immunoassay was evaluated at five sites according to the CLSI EP6-A [11]. Low and high TnI pools with undiluted values that ranged between 10–50,000 ng/L were used. The percent recov-ery was calculated.

A mean recovery ( ± SD) of 100% ( ± 20%) was required to demon-strate acceptable linearity of dilution.

Evaluation of potential interference by hemoglobin, bilirubin and lipids was performed based on guidance from the CLSI document EP7-A2 [12]. For all analytes the change in concentration was com-pared and the recovery was calculated. Interference was regarded acceptable if the cTnI recovery was within 100% ± 10%.

CarryoverThe cTnI carryover was assessed by analyzing high cTn concentra-tion heparin plasma pools followed by low cTn concentration hepa-rin plasma pools. The percentage of carryover was calculated.

Sample type comparisonSample type comparison was performed at two sites. At the first site, 31 matched samples obtained from randomly selected patients with cTnI concentrations above the LoD were collected into serum,

Brought to you by | University of MichiganAuthenticated

Download Date | 10/9/14 5:38 PM

Krintus et al.: hs-cTnI evaluation      1659

ethylenediaminetetraacetic acid (EDTA) and lithium heparin plasma tubes. Each specimen type was measured in duplicate. The mean results were compared and the significance of the differences was evaluated. At the second site, 26 frozen paired serum and lithium heparin plasma samples were evaluated in duplicate. Both sites per-formed linear regression and Bland-Altman analyses.

Method comparisonAnalytical correlation of the ARCHITECT STAT hs-TnI to the contem-porary ARCHITECT STAT TnI assay was evaluated using 2598 samples that spanned the dynamic range of the hs-TnI as well as the contem-porary TnI assay. Correlation, slope, and y-intercept according to Passing-Bablok and Deming regression analyses were calculated. A concordance analysis was performed in order to assess the degree of agreement between both assays.

99th percentile URLFor estimation of the 99th percentile URL, a cohort of 1769 appar-ently healthy individuals and blood donors recruited in seven coun-tries. 993 women and 776 men, aged between 18 and 91 years, were selected. Individuals were either blood donors (2 countries, n = 409) or recruited for a reference range study (5 countries, n = 1360). Prior to the initiation of participant enrollment, the study was approved by local Ethics Committees/Investigational Review Boards, as required. All study participants completed questionnaires detailing their medi-cal histories and medications in order to confirm their good health before the blood draw. This healthy reference population was defined as individuals who were free of cardiac diseases (no history of car-diac disease, cardiac treatment/medication or cardiac intervention). Additional exclusion criteria included pregnancy, current infection and any chronic inflammatory disease.

Statistical analysisCategorical data were summarized with number and percentages. Data were expressed as mean ± SD and median with 25th–75th

percentiles. Comparison between the groups was performed by using the Mann-Whitney U-test. Dilution linearity was determined by least-squares regression analysis. Statistical comparison between different sample types was assessed with the Wilcoxon rank-sum test for paired samples. Passing-Bablok and Deming regression analyses were performed for both the sample and method comparison.

The 99th percentile values were determined by right-sided non-parametric statistics as described in the CLSI C28-A3c [13]. Ninety percent confidence intervals (CI) were provided for the 99th percen-tile values, if possible. We excluded the outlier observations from the analyzed population based on the method described by Reed et al. [14]. Statistical analysis was performed using Analyse-it 2.2 for Micro-soft Excel (Analyse-it Software, Leeds, UK) and MedCalc 12.7.0 (Med-Calc Software, Ostend, Belgium).

Results

Performance characteristics

Results from the 5-day precision evaluation are summa-rized in Table 1. Total imprecision ranged between 3.3% and 8.9%, 2.0% and 3.5%, and 1.5% and 5.2% for the low, medium and high controls, respectively. Total inter-laboratory CV at all sites was lower than 10% for samples ranging from 18.4 to 15,912.3 ng/L, supporting the man-ufacturer’s claim of “total imprecision of   ≤  10% CV with controls or panels across the range of 10–50,000  pg/mL (ng/L)”.

The lowest cTnI concentration corresponding to a total CV of 10% was 5.6 ng/L (Figure 1); a total CV of 20%, 15% and 5% was at 2.2, 3.6 and 9 ng/L, respectively.

The LoB and LoD results at all sites verified the man-ufacturer’s claim of LoB range of 0.7–1.3 ng/L and LoD range of 1.1–1.9 ng/L. The manufacturer’s LoQ was verified at   ≤  10 ng/L with a 10% CV (Table 1).

Table 1 Evaluation of precision and LoQ in eight sites following the CLSI EP5-A protocol [9].

Site  Country   Instrument   LoQ ng/L   %CV at LoQ    Low control    Medium control   

High control

ng/L   % CV ng/L   % CV ng/L   % CV

1  Belgium   i2000SR   7.7   10.6   19.0   3.3   189.2   2.5   14,032.0   2.62  Denmark  i2000SR -1   4.6   8.4   18.4   3.3   190.2   2.4   14,593.4   1.8

    i2000SR -2       19.7   3.5   193.1   3.5   14,742.2   2.2    i2000SR -3       20.4   8.9   201.2   3.1   14,779.6   2.4

3  France   i2000SR -1   5.8   8.2   18.9   3.4   186.3   2.4   13,994.2   2.5    i2000SR -2       19.9   4.4   186.0   2.0   14,602.1   1.5

4  Germany  i2000SR   4.6   6.8   19.5   3.9   194.4   3.0   14,935.9   2.95  Italy   i2000SR   7.8   7.1   20.6   4.3   193.7   2.5   15,690.2   1.56  Norway   i2000SR   7.2   6.7   20.1   3.9   193.8   2.8   14,610.6   1.67  Poland   i2000SR   4.8   8.0   18.8   5.3   187.5   3.2   15,117.8   5.28  Spain   i1000SR   8.1   7.8   21.5   4.8   198.7   3.0   15,912.3   2.1

CLSI, Clinical and Laboratory Standards Institute; CV, coefficient of variation; LoQ, limit of quantitation.

Brought to you by | University of MichiganAuthenticated

Download Date | 10/9/14 5:38 PM

1660      Krintus et al.: hs-cTnI evaluation

14

12

10

8

6

4

0 5 10 15 20

hs-Tnl, ng/L

CV

, %

25 30 35 40 45

Figure 1 Imprecision profile for a hs-cTnI assay on the Abbott ARCHITECT immunoanalyzer (Abbott Laboratories, Wiesbaden, Germany).hs-cTnI, high sensitivity cardiac troponin I.

Table 2 Passing-Bablok and Bland-Altman analyses for 31 matched serum, EDTA plasma and lithium heparin samples with concentrations across the dynamic range of the assay.

Comparison   

Passing-Bablok  

Bland-Altman

Slope  95% CI of the slope

  Y-intercept, ng/L

  95% CI of the Y-intercept, ng/L

  r Average difference, ng/L

  95% CI of the average difference, ng/L

  Mean bias, %

Serum vs. EDTA   1.04  0.89  1.09  –0.39  –0.7  0.24  0.98  0.6  –4.8  6.0  –5.7 ± 25.5Serum vs. lithium heparin

  1.15  1.05  1.23  –0.24  –0.62  0.28  0.99  –1.2  –6.0  3.6  17.4 ± 39.9

EDTA vs. lithium heparin

  1.15  1.05  1.28  0.30  0.08  0.56  0.97  –1.8  –9.6  6.0  25.9 ± 28.9

CI, confidence interval; EDTA, ethylenediaminetetraacetic acid.

Dilution linearity was evaluated at five sites using 26 plasma samples. Slopes of observed versus expected values ranged from 0.904 to 0.981 and the correlation coefficient did not differ significantly from 1.0. Overall recovery was 90.6%. The overall correlation was 0.988 over the range of 9.60–20,097.10 ng/L with a slope of 0.971 and intercept of 175.29.

No interferences with hs-cTnI results were seen by elevated levels of bilirubin, hemoglobin and triglycerides. A change in hs-cTnI concentration of <10% was observed up to 20 mg/dL bilirubin concentration. No influence of hemolysis, as judged by hemoglobin concentration via hemolysis index (up to 860 mg/dL), was demonstrated and the mean change was less than –3%. There was no influence of lipemia, judged by triglycerides concentra-tion  > 500 mg/dL. The mean recovery of hs-cTnI ranged from 96% to 113%.

The amount of plasma carryover was determined to be 0.006%.

Sample type comparison

The Passing-Bablok and Bland-Altman results for the first experiment carried out on 31 matched serum, EDTA plasma and lithium heparin plasma samples are summa-rized in Table 2. Results for serum and EDTA plasma did not show a significant difference using the Wilcoxon rank-sum test (p = 0.32), but statistically significant differences were found between lithium heparin and EDTA plasma, and between lithium heparin plasma and serum (p < 0.05). In a second experiment comparing matched frozen serum versus frozen lithium heparin plasma samples, linear regression analysis gave a slope of 1.08 and y-intercept of 0.36 ng/L in the cTnI range of 0–4000 ng/L. No signifi-cant bias was detected by Bland-Altman analysis (average bias = 7.8%).

Method comparison

Passing-Bablok analysis of 2598 plasma samples with cTnI results detectable by both assays (range 10.0–238,260 ng/L) using the ARCHITECT STAT hs-TnI and the ARCHITECT STAT TnI assays demonstrated a relatively good correlation between the two methods (y = 0.97 × –10.2; r = 0.97) across the wide range of cTnI concentrations, and the correlation was similar on the 1237 samples from 10.0 to 50,000 ng/L (y = 0.94 × –4.06; r = 0.96, with a systematic bias of –17.7%). When only the results with cTnI values below 500 ng/L were con-sidered, the Passing-Bablok regression analysis demon-strated a slope of 0.94 (95% Cl 0.92–0.96), y-intercept of –3.02 ng/L and r = 0.94, with the average bias of –23.4%. Passing-Bablok analysis of results with cTnI values below 100 ng/L showed a slope of 1.04, y-intercept of –5.87 ng/L and r = 0.94, with an average bias of –25.0%. The method comparison and plots of the differences are presented in Figure 2. Additionally, we compared the two

Brought to you by | University of MichiganAuthenticated

Download Date | 10/9/14 5:38 PM

Krintus et al.: hs-cTnI evaluation      1661

50,000

A B

C D

E F

Scatter plot with Passing-Bablok fitn=1237

Scatter plot with Passing-Bablok fitn=706

Scatter plot with Passing-Bablok fitn=469

45,000

40,000

35,000

30,000

25,000

AR

CH

ITE

CT

hs

trop

onin

I, n

g/m

LA

RC

HIT

EC

T h

s tr

opon

in I,

ng/

mL

Diff

eren

ce (

AR

CH

ITE

CT

hs

Tro

poni

n I,

ng/m

L-A

RC

HIT

EC

T tr

opon

in I,

ng/

L / M

ean

of A

llD

iffer

ence

(A

RC

HIT

EC

T h

s tr

opon

in I,

ng/

mL-

AR

CH

ITE

CT

trop

onin

I, n

g/L

/ Mea

n of

All

Diff

eren

ce (

AR

CH

ITE

CT

hs

trop

onin

I, n

g/m

L-A

RC

HIT

EC

T tr

opon

in I,

ng/

L / M

ean

of A

ll

20,000

15,000

10,000

5000

500

450

400

350

300

250

200

150

100

50

0

00 50 100

ARCHITECT troponin I, ng/L

20

40

60

AR

CH

ITE

CT

hs

trop

onin

I, n

g/L

80

100

120

140

0 100 200 300 400 500 0 100 200

0-200%

-150%

-100%

-50%

0%

50%

100%

150%

50 100

300 400 500

-200%

-150%

-100%

-50%

Identity

Difference plot

Identity

Bias (-17.7%)

95% limits ofagreement(-96.8% to 61.5%)

Identity

Identity

Bias (-23.4%)

95% limits ofagreement(-113.1% to 66.2%)

Identity

Bias (-25.0%)

95% limits ofagreement(-120.9% to 71.0%)

Passing-Bablok (I)fit(-3.02 + 0.94x)

Identity

Passing-Bablok (I) fit(-5.87 + 1.04x)

Passing-Bablok(I) fit(-4.06 + 0.94x)

50%

100%

150%

0%

-200%

-150%

-100%

-50%

50%

100%

150%

0%

00 10,000 20,000 30,000 40,000 50,000 0 10,000 20,000 30,000

Mean of all, ng/L

Mean of all, ng/L

Mean of all, ng/L

Difference plot

Difference plot

40,000 50,000ARCHITECT troponin l, ng/L

ARCHITECT troponin l, ng/L

Figure 2 Method comparison and bias plots between the ARCHITECT STAT high sensitive troponin I and the ARCHITECT STAT troponin I assay.(A and B) Passing-Bablok regression and Bland-Altman difference plots using 1237 undiluted plasma samples (analytical range 11–42,870 ng/L). (C and D) Passing-Bablok regression and Bland-Altman difference plots using 706 undiluted plasma samples (analytical range 11–495 ng/L). (E and F) Passing-Bablok regression and Bland-Altman difference plots using 469 undiluted plasma samples (analytical range 11–99 ng/L).

methods using fresh serum (n = 135) and plasma samples (n = 581), and frozen plasma (n = 74), all below 50,000 ng/L. The correlation coefficient was 0.99, 0.94 and 0.99 and the average bias of –5.4%, –18.3 and –9.9%, respec-tively. The detailed contributions of the participating

sites in the method comparison are summarized in Sup-plemental Data, which accompanies the article at http://www.degruyter.com/view/j/cclm.2014.52.issue-11/issue-files/cclm.2014.52.issue-11.xml (see Supplemental Data, Tables 1 and 2).

Brought to you by | University of MichiganAuthenticated

Download Date | 10/9/14 5:38 PM

1662      Krintus et al.: hs-cTnI evaluation

Concordance was calculated using the respective package insert 99th percentile cut-offs. Concordance of ARCHITECT STAT hs-TnI (26.2 ng/L) and ARCHITECT STAT TnI assay (28 ng/L) was found to be 95% with κ statistic for agreement of 0.89 (95% CI 0.87–0.92).

Reference population and determination of the 99th percentile URL

The reference population consisted of 1769 individu-als aged 18–91, with a median age (25th–75th percentile) of 49 (18–60) years. The 99th percentile URL values are summarized in Table 3. The overall 99th percentile URL of hs-cTnI concentration was 19.3 ng/L and was higher in men than in women. Men also had a significantly higher median hs-cTnI concentration compared to women (2.7 vs.

Figure 3 Distribution of hs-cTnI results in the overall reference population (n = 1769), in female and male cohort (internal graph) with 99th percentile upper reference limits.hs-cTnI, high sensitivity cardiac troponin I.

Table 3 Values of 99th percentile URL in the overall reference population (n = 1769).

Presumably healthy participants

  Number of subjects

  All participants 99th percentile (90% CI), ng/L

All   1769   19.3 (14–25)Women   993   11.4 (10–15)Men   776   27 (20–67)

CI, confidence interval; URL, upper reference limit.

1.8 ng/L; p < 0.0001). The distribution of hs-cTnI values is shown in Figure 3. TnI was detectable in 75% and 57% of the apparently healthy population using the lower (1.1 ng/L) and upper (1.9 ng/L) limit of the LoD range provided by the ARCHITECT STAT hs-TnI package insert, respectively. Detailed proportions of the study participants with detect-able cTnI concentrations above LoD values for each par-ticipating site along with hs-cTnI concentrations in the overall participating site populations as well as in relation to the participant gender are provided in the Supplemen-tal Data (see Supplemental Data, Tables 3–6).

DiscussionIn this large multicenter European study we evaluated the fundamental validation features of a new hs-cTnI assay, including its analytical performance characteristics. We clearly demonstrated that the ARCHITECT STAT hs-TnI immunoassay has an excellent capability to be classified as “high sensitivity” and “guideline acceptable” accord-ing to current recommendations [2, 6]. The imprecision characteristics obtained from different sites indicated that this assay is robust and independent of the instrument. The concentration of the low control material (20 ng/L) is close to the 99th percentile URL ensuring good preci-sion at this critical cut-off. The recommended criteria for

Brought to you by | University of MichiganAuthenticated

Download Date | 10/9/14 5:38 PM

Krintus et al.: hs-cTnI evaluation      1663

a hs-cTn assay are based on precision and detectability in healthy individuals [6, 15, 16]. To ensure the required low-end assay performance we verified LoB, LoD and LoQ. Our results were in close agreement with the data in the manufacturer’s package insert [17]. The newer hs-cTnI assay is more analytically sensitive than the previous gen-eration of cTnI when comparing analytical sensitivity or LoD [8]. Further analytical comparisons of this hs-cTnI assay with other assays, either on market or in develop-ment, have been discussed in the literature [2, 18]. The high sensitive assays described in these publications have a 10% CV concentration below the 99th percentile URL whereas most of the contemporary sensitive assays did not meet this requirement. In regard to the clinical appli-cation, LoQ reflects the lowest amount of cTn that can be quantitatively determined with stated acceptable impreci-sion and bias [8]. Current data support clinical application of cTn assays with LoQ  ≥  LoD and imprecision at LoQ not exceeding 10% CV [2, 8]. Our results demonstrate that this new ARCHITECT STAT hs-TnI assay fulfills these criteria. Another important issue among analytical factors that should be considered due to the heterogeneity of cTns is antibody specificity [19]. Although antibodies can poten-tially react with different epitopes present on the antigen, the ARCHITECT STAT hs-TnI immunoassay utilizes anti-bodies directed against epitopes which are located in the stable part of the cTnI molecule and therefore should not be affected by complex formation and in vivo modifica-tions [20].

Sample dilution is an important factor that can poten-tially affect the result of cTn measurement. In our study we did not find any significant deviation from linearity, which indicates that samples with cTn concentrations above 50,000 ng/L can be diluted and re-assayed using the manual or automated dilution procedure.

Preanalytical factors are the most common source of errors which can lead to unreliable hs-cTn results, with hemolysis being the most important one [21–23]. Our study clearly confirmed that results obtained with the investigated hs-cTnI assay are not affected by hemolysis, hyperbilirubinemia and lipemia, as previously described by Koerbin et  al. [24]. Consequently, ARCHITECT STAT hs-TnI is more robust than Roche hs-cTnT and some other cTnI assays with regard to hemolysis interference [25]. In the setting of hs-cTn assays even subtle interfer-ences are unacceptable due to profound clinical conse-quences of false-positive or false-negative results [5, 22]. Based on the obtained results, we can conclude that the investigated assay is resistant to interferences, even at low hs-cTnI concentrations close to the critical threshold of MI.

Another preanalytical limitation of hs-cTn assays may be the use of different anticoagulants in plasma samples. The clinical importance of cTn measurements in patients with suspected MI necessitates a rapid and effective turn-around time. Although the use of both serum and plasma is recommended, plasma is usually the matrix of choice in order to ensure timeliness of results and avoid poten-tial problems associated with prolonged clotting time in patients receiving anticoagulant therapy [26]. Therefore, we evaluated the impact of the type of an anticoagulant on cTnI concentration. This study analyzed the most com-monly used sample matrices such as serum, heparinized and EDTA plasma. Previous studies with hs-cTnI assays and the information provided by the manufacturer’s in package insert indicated that both serum and plasma can be used for routine cTnI determination [17]. However, in one study a slight but statistically significant difference was found between lithium heparin and EDTA plasma and between lithium heparin plasma and serum, whereas serum and EDTA plasma were not significantly different. The negative impact of the anticoagulants used in the sample matrix can lead to adverse clinical implications resulting from the falsely low or high cTn concentrations [1, 26, 27]. However, results from the second site did not confirm these differences. Therefore, it remains unclear from our study if the interchangeable use of heparinized plasma, serum or EDTA plasma should be recommended and this aspect requires further studies and validation for practical use [28]. Following recent clinical recommenda-tions, serum or plasma (preferably heparinized) are con-sidered appropriate samples but serial measurements in a given patient must be done from the same sample type to minimize variation [29].

Additionally, we demonstrated that carryover did not affect results of the ARCHITECT STAT hs-TnI immunoas-say, thus eliminating a potentially important source of false positive results. In contrast to our findings, some researchers have previously shown that carryover may affect Tn results obtained by using hs-cTn assays, but it was recently not confirmed by others [30].

The real challenge was to compare the new hs-TnI assay with the currently available TnI assay taking into account a nearly 10-fold reduced detection limit of the new assay, with its improved precision at the critical decision limit. Of major importance, we achieved overall agreement of 95% between both tests using the assay-spe-cific 99th percentile URL. However, we found a systematic error between both assays regardless of the type of mate-rial which could be expected when comparing assays with different performance characteristics. This observation is in line with previous reports indicating that both methods

Brought to you by | University of MichiganAuthenticated

Download Date | 10/9/14 5:38 PM

1664      Krintus et al.: hs-cTnI evaluation

should not be used interchangeably or at least converted carefully [31]. However, the excellent concordance of 95% between the two assays using the assay-specific 99th percentile URL allows the transparent conversion to the hs-TnI without unfavorable clinical implications.

Estimation of the 99th percentile URL value is a key component during the validation process and perhaps the most clinically important. The normal population investigated in this study was of sufficient size to calcu-late robust overall and gender-specific cut-offs. The 99th percentile values calculated in this study, were lower than those reported in the package insert, however the significant difference between genders was also found in our study. Our female 99th percentile value is close to the value found by Koerbin et  al. (11.1 ng/L) but male 99th percentile value is different (14.0 ng/L in Koerbin’s study and 27 ng/L in this paper) [24]. It should be noted that the Koerbin’s study was performed in a smaller number of subjects.

Gender-specific cut-offs are recommended by the Third Universal Definition of Myocardial Infarction [1], but this concept warrants further investigation. Despite the lower 99th percentile values in the female subgroup, the imprecision goal below 10% CV was maintained. These findings are of particular importance in regard to the clini-cal application of hs-cTnI using gender-specific cut-offs. Lowering the diagnostic threshold for hs-cTnI when com-pared with the contemporary cTnI is expected to facilitate the decision making process in different clinical settings [32, 33].

We were able to detect cTnI concentrations above LoD in more than 50% of the healthy cohort [2]. Moreover, con-sidering the excellent precision at the 99th percentile URL, the second criterion of this classification, the assay meets the requirements to be “guideline acceptable” [15, 34].

Two limitations of our study should be noted. First, there have been several reports of outliers (or flyers) with cTn assays from numerous manufacturers [35, 36]. The cause and frequency of these outliers have been difficult to determine, but they typically are seen only with real patient samples. A review of the data from our method comparison and reference range studies did not identify any outliers. A recent study by Sawyer et al. investigated the occurrence of outliers with both the contemporary ARCHITECT STAT TnI assay as well as the ARCHITECT STAT hs-TnI assay [36]. There were no critical outliers detected with the hs-TnI assay.

The second limitation is that this was purely an ana-lytical evaluation of the assay. Therefore, both diagnostic performance and clinical benefits of the ARCHITECT STAT hs-TnI immunoassay warrant further investigation.

In summary, independent observations at nine dif-ferent sites confirmed the analytical characteristics of the ARCHITECT STAT hs-TnI immunoassay. The assay meets guideline recommendations for a high sensitivity assay and demonstrates excellent precision and low-end sen-sitivity. Additionally, based on a large European healthy cohort from seven different countries, we successfully established its 99th URLs, overall URL and gender-spe-cific URLs, which hopefully will facilitate its implementa-tion into clinical practice.

Acknowledgments: Abbott Laboratories provided the pro-tocol, the reagents, calibrators, controls and minor sup-port for this study. We thank the laboratory scientists and technicians who participated in the study, particularly Slawomir Manysiak (Laboratory Medicine, Bydgoszcz), Magali Grastilleur (Biochimie, Paris), Rosemarie Lott (Laboratory Medicine, Mainz) and Rønnaug Grude (Clini-cal Biochemistry, Stavanger).

Conflict of interest statement

Authors’ conflict of interest disclosure: The authors stated that there are no conflicts of interest regarding the publica-tion of this article. The minor support by Abbott Diagnostics played no role in the study design; in the collection, analy-sis, and interpretation of data; in the writing of the report; or in the decision to submit the report for publication.Research funding: None declared.Employment or leadership: L. Lennartz and J. Shih are employees of Abbott Laboratories.Honorarium: G. Lefevre received a speaker honoraria from Abbott.

References1. Thygesen K, Alpert JS, Jaffe AS, Simoons ML, Chaitman BR, White

HD, et al. Third universal definition of myocardial infarction. J Am Coll Cardiol 2012;60:1581–98.

2. Apple FS, Collinson PO. Analytical characteristics of high-sensi-tivity cardiac troponin assays. Clin Chem 2012;58:54–61.

3. Koerbin G, Abhayaratna WP, Potter JM, Apple FS, Jaffe AS, Rav-alico TH, et al. Effect of population selection on 99th percentile values for a high sensitivity cardiac troponin I and T assays. Clin Biochem 2013;46:1636–43.

4. Thygesen K, Mair J, Katus H, Plebani M, Venge P, Collinson P, et al. Recommendations for the use of cardiac troponin measure-ment in acute cardiac care. Eur Heart J 2010;31:2197–204.

5. Apple FS, Saenger AK. The state of cardiac troponin assays: looking bright and moving in the right direction. Clin Chem 2013;59:1014–6.

Brought to you by | University of MichiganAuthenticated

Download Date | 10/9/14 5:38 PM

Krintus et al.: hs-cTnI evaluation      1665

6. Thygesen K, Mair J, Giannitsis E, Mueller C, Lindahl B, Blanken-berg S, et al. How to use high-sensitivity cardiac troponins in acute cardiac care. Eur Heart J 2012;33:2252–7.

7. Tate JR, Bunk DM, Christenson RH, Katrukha A, Noble JE, Porter RA, et al. Standardisation of cardiac troponin I measurement: past and present. Pathology 2010;42:402–8.

8. International Federation of Clinical Chemistry and Labora-tory Medicine. Troponin assay analytical characteristics 2013. Available from: http://www.ifcc.org/media/245272/IFCC%20Troponin%20I%20and%20T%20%28ug_L%20units%29%20_update%20December%202013.pdf. Accessed 1 May, 2014.

9. Clinical Laboratory Standards Institute. Evaluation of precision performance of quantitative measurement methods; approved guidelines, EP5-A2. Wayne, PA: Clinical and Laboratory Stand-ards Institute, 2004.

10. Clinical Laboratory Standards Institute. Evaluation of detec-tion capability for clinical laboratory measurement procedures; approved guidelines, 2nd ed, EP17-A2. Wayne, PA: Clinical and Laboratory Standard Institute, 2012.

11. Clinical Laboratory Standards Institute. Evaluation of the linearity of quantitative measurement procedures: a statistical approach; approved guidelines, EP6-A. Wayne, PA: Clinical and Laboratory Standards Institute, 2003.

12. Clinical Laboratory Standards Institute. Interference testing in clinical chemistry; approved guideline, 2nd ed. EP7-A2. Wayne, PA: Clinical and Laboratory Standards Institute, 2005.

13. Clinical Laboratory Standards Institute. Defining, establish-ing and verifying reference intervals in the clinical laboratory; approved guideline, 3rd ed. EP28-A3c. Wayne, PA: Clinical and Laboratory Standards Institute, 2010.

14. Reed AH, Henry RJ, Mason WB. Influence of statistical method used on the resulting estimate of normal range. Clin Chem 1971;17:275–84.

15. Apple FS. A new season for cardiac troponin assays: it’s time to keep a scorecard. Clin Chem 2009;55:1303–6.

16. Marini MG, Cardillo MT, Caroli A, Sonnino C, Biasucci LM. Increasing specificity of high-sensitivity troponin: new approaches and perspectives in the diagnosis of acute coronary syndromes. J Cardiol 2013;62:205–9.

17. Abbott ARCHITECT STAT high sensitive troponin-I. Package insert. G1-0139/R02. 2013.

18. Apple FS, Ler R, Murakami MM. Determination of 19 cardiac troponin I and T assay 99th percentile values from a common presumably healthy population. Clin Chem 2012;58:1574–81.

19. James S, Flodin M, Johnston N, Lindahl B, Venge P. The antibody configurations of cardiac troponin I assays may determine their clinical performance. Clin Chem 2006;52:832–7.

20. Panteghini M, Gerhardt W, Apple FS, Dati F, Ravkilde J, Wu AH. Quality specifications for cardiac troponin assays. Clin Chem Lab Med 2001;39:175–9.

21. Ji JZ, Meng QH. Evaluation of the interference of hemoglobin, bilirubin, and lipids on Roche Cobas 6000 assays. Clin Chim Acta 2011;412:1550–3.

22. Sztefko K. Troponin measurement by immunoassay – problem of low assay sensitivity and interference from heterophilic anti-bodies. In: Sztefko K, editor. Immunodiagnostics and patient safety. Berlin: De Gruyter, 2013:129–36.

23. Plebani P, Lippi G. Hemolysis index: quality indicator or criterion for sample rejection? Clin Chem Lab Med 2009;47:899–902.

24. Koerbin G, Tate J, Potter JM, Cavanaugh J, Glasgow N, Hickman PE. Characterisation of a highly sensitive troponin I assay and its application to a cardio-healthy population. Clin Chem Lab Med 2012;50:871–8.

25. Bais R. The effect of sample hemolysis on cardiac troponin I and T assays. Clin Chem 2010;56:1357–9.

26. Tate JR, Panteghini M. Measurement of cardiac troponins revis-ited. Biochim Clin 2013;32:535–46.

27. Schiettecatte J, Anckaert, Smitz J. Interferences in immunoas-says. In: Norman HL, Christopoulos TK, editors. Advances in immunoaasay technologies. Rijeka: InTech, 2013:45–62.

28. Pagani F, Stefini F, Chapelle JP, Lefevre G, Graine H, Luthe H, et al. Multicenter evaluation of analytical performance of the Liaison troponin I assay. Clin Biochem 2004;37:750–7.

29. Jardine RM, Dalby AJ, Klug EQ, Vermaak WH, White HD, Baden-horst JC, et al. Consensus statement on the use of high sensitiv-ity cardiac troponins. SA Heart J 2013;9:1–8.

30. Lippi G, Avanzini P, Musa R, Aloe R, Cervellin G. Carryover does not affect results of Beckman Coulter highly-sensitive-AccuTnI assay on Access 2. Clin Chem Lab Med 2013;51:e141–3.

31. Saenger AK, Beyrau R, Braun S, Cooray R, Dolci A, Freidank H, et al. Multicenter analytical evaluation of a high-sensitivity troponin T assay. Clin Chim Acta 2011;412:748–54.

32. Collinson PO, Heung YM, Gaze D, Boa F, Senior R, Christenson R, et al. Influence of population selection on the 99th percen-tile reference value for cardiac troponin assays. Clin Chem 2012;58:219–25.

33. de Lemos JA, Morrow DA, deFilippi CR. Highly sensitive troponin assays and the cardiology community: a love/hate relationship? Clin Chem 2011;57:826–9.

34. Korley FK, Jaffe AS. Preparing the United States for high-sensi-tivity cardiac troponin assays. J Am Coll Cardiol 2013;61:1753–8.

35. Pretorius CJ, Dimeski G, O’Rourke PK, Marquart L, Tyack SA, Wilgen U, et al. Outliers as a cause of false cardiac troponin results: investigating the robustness of 4 contemporary assays. Clin Chem 2011;57:710–8.

36. Sawyer N, Blennerhassett J, Lambert R, Sheehan P, Vasi-karan SD. Outliers affecting cardiac troponin I measurement: comparison of a new high sensitivity assay with a contem-porary assay on the Abbott ARCHITECT analyser. Ann Clin Biochem 2014;51. Epub ahead of print 23 Sep 2013. DOI: 10.1177/000456321349937.

Supplemental Material: The online version of this article (DOI: 10.1515/cclm-2014-0107) offers supplementary material, available to authorized users.

Brought to you by | University of MichiganAuthenticated

Download Date | 10/9/14 5:38 PM