Clinical Chemistry 46: 1631-1637, 2000;
(Clinical Chemistry. 2000;46:1631-1637.)
© 2000 American Association for Clinical Chemistry, Inc.
Multicenter Evaluation of Five Assays for Myoglobin Determination
Martina Zaninotto1,
Franca Pagani2,
Sara Altinier1,
Paolo Amboni3,
Roberto Bonora2,
Alberto Dolci4,
Patrizia Pergolini5,
Arialdo Vernocchi3,
Mario Plebani1,
Mauro Panteghini2,a and
for the Italian Society of Clinical Biochemistry and Clinical Molecular Biology–Italian Society of Laboratory Medicine Working Group on Markers of Myocardial Damage
1
Dipartimento di Medicina di Laboratorio, Università-Ospedale, 35100 Padova, Italy.
2
Laboratorio Analisi Chimico Cliniche 1, Azienda
Ospedaliera ‘Spedali Civili’, 25125 Brescia, Italy.
3
Laboratorio Analisi Chimico Cliniche, Ospedali Riuniti,
24100 Bergamo, Italy.
4
Laboratorio Analisi Chimico Cliniche, Casa di Cura S.
Maria, 21053 Castellanza VA, Italy.
5
Laboratorio Analisi Chimico Cliniche, Azienda
Opedaliera ‘Maggiore della Carità’, 28100 Novara, Italy.
a Author for correspondence. Fax 39-030-3995369; e-mail panteghi{at}osp.unibs.it
 |
Abstract
|
|---|
Background: Lacking assay standardization, different myoglobin
methods may produce results that differ significantly.
Methods: A multicenter study was carried out to compare the
analytical performance of five commercially available assays for
myoglobin measurement. Linearity, imprecision, interferences, and
method comparison were studied according to NCCLS guidelines, whereas
reference values were determined following IFCC recommendations.
Results: The BNA and Opus showed relatively high imprecision (all
but one total CV >7.4%). Other assays showed lower CVs, but they
varied among laboratories, particularly at a normal myoglobin
concentration (Access, 6.0–11%; Hitachi, 3.8–5.8%; Stratus,
3.4–6.5%). Results were lower in anticoagulated samples on the
Access, in heparin and citrate samples on the Stratus, and in citrate
samples on the BNA and Opus, and increased in heparin and EDTA samples
on the Hitachi. Use of separator gel produced results significantly
lower (P <0.001) on the Hitachi and higher
(P = 0.016) on the Opus. Bilirubin, turbidity, and
hemoglobin had no effect on evaluated methods, but rheumatoid factor
affected the Access. In method comparisons, high correlation
coefficients (
0.98) were obtained. The Stratus gave higher results;
however, the Access and BNA gave the lowest. The following upper
reference limits (µg/L) for men and women, respectively, were
obtained: Access, 70 and 52; BNA, 51 and 49; Hitachi, 67 and 58; Opus,
80 and 50; and Stratus, 86 and 63.
Conclusion: The possibility of high imprecision and marked
disagreement among commercial myoglobin assays should be
carefully considered in clinical practice.
|
Introduction
|
|---|
Myoglobin, a low-molecular weight protein present in the cytosol
of cardiac and skeletal muscle, is an old biochemical marker with new
perspectives (1). The first report on the clinical
usefulness of serum myoglobin in the diagnosis of acute myocardial
infarction, using a complement fixation technique for the protein
measurement, was published in 1975 (2). Since then,
much technical progress have been made, and methods requiring several
hours have been replaced by assays with rapid turnaround times and
better analytical performance (1). The availability of
simple and rapid methods for the measurement of myoglobin has greatly
enhanced its clinical utilization in emergency situations for the early
diagnosis and management of myocardial infarction. As underlined
recently (3), myoglobin currently is the marker that most
effectively fits the role of an early marker of myocardial necrosis,
but because assay standardization and reference materials are
lacking, the diffusion of different analytical procedures and the rush
by commercial manufacturers to develop and market new methods for
myoglobin determination may produce data on patient samples that differ
significantly when measured by different assays. In this report, we
present the results of a multicenter study carried out to evaluate and
compare the analytical performance of five commercially available
assays to measure myoglobin in serum.
|
Materials and Methods
|
|---|
The five participating laboratories were located in a narrow
geographic area in northern Italy because it was necessary to
distribute and transport in real time the frozen samples from one
center to another to carry out every phase of the evaluation using the
same samples in different centers. Five assays/platforms from three
different manufacturers were evaluated, each one in two different
laboratories: (a) Access (Beckman Coulter, Brea, CA;
laboratories 2 and 3); (b) BNA II (Dade Behring, Deerfield,
IL; laboratories 1 and 5); (c) Hitachi 911 (Roche
Diagnostics, Mannheim, Germany; laboratories 1 and 4); (d)
Opus (Dade Behring; laboratories 2 and 5); and (e) Stratus
II (Dade Behring; laboratories 3 and 4). The study was carried out
using the manufacturers’ recommendations in the preparation and
storage of reagents, calibrators, and controls and in the frequency of
system calibration. Unless otherwise stated, serum was used as sample.
After centrifugation and separation, serum specimens were stored at
-20 °C until being assayed.
detection limit
The zero calibrator of each assay was tested 10 times on the
corresponding instrument in the same analytical run. The limit of
detection was defined as the concentration of myoglobin corresponding
to a signal 3 SD above the obtained mean value. This experiment was not
carried out on the BNA because the lowest calibration curve
concentration (
26 µg/L) in this instrument is obtained
automatically by serial dilutions of the calibration material and the
analyzer is set to report results lower than this point as <26 µg/L.
linearity
According to the NCCLS EP6-P protocol (4), human serum
pools with myoglobin concentrations 30% higher than the upper limit
of the calibration curve of each assay were used to prepare four
dilutions (3:4, 1:2, 1:4, 1:5), using the manufacturer’s recommended
diluent. Each dilution was tested four times, and all results were
obtained in the same analytical run. The criteria of NCCLS document
EP6-P were used to evaluate assay linearity (4).
precision
Three human serum pools with normal (50 µg/L), borderline (100
µg/L), and high (300 µg/L) myoglobin concentrations were prepared,
aliquoted, frozen, and sent to all participating laboratories. Assay
imprecision was estimated using the analysis of variance method
described in NCCLS EP5-A (5), with two replicates per
specimen per analytical run and one analytical run per day for 20 days.
During the experimental period, a manufacturer’s control sample with a
myoglobin concentration of 100 µg/L was assayed to validate each
run, and a single lot of reagents was used.
interference studies
Sample type.
Fifty-one volunteers were selected, and informed
consent was obtained. A blood sample was drawn from each volunteer,
using vacuum collection tubes (Becton Dickinson) to obtain serum in
tubes with separator gel, lithium heparinate plasma, sodium citrate
plasma, and EDTA-K3 plasma. The results for these
samples were compared with the paired serum values, and the
significance of the differences was evaluated (Wilcoxon rank-sum test).
Endogenous interferents.
According to the NCCLS EP7-P
guideline (6), albumin (final concentration, 100 g/L),
bilirubin (final concentration, 300 mg/L), hemoglobin (final
concentration, 10 g/L), and Intralipid 20% solution were added (one
part of interferent and 19 parts of pool) to two human serum pools with
normal (60 µg/L) and high (110 µg/L) myoglobin concentrations. The
results for these "test" solutions were compared with those of
"control" solutions obtained from the same two pools diluted in the
same way with the manufacturers’ recommended diluents instead of the
interfering substances. Each sample (test and control) was
assayed 10 times in random order to minimize any bias
attributable to analytical drift. Taking into account the analytical
imprecision of various methods, we considered as clinically significant
a difference >10% between test and control mean results
(6). To test interference by rheumatoid factor (RF), samples
(n = 10) with high RF concentrations [48–3020 kIU/L, when
determined with a rate-nephelometric latex method (BNA); previously
established upper reference limit, 20 kIU/L] were evaluated in
duplicate, and the mean results from different instruments were
compared.
method comparison
Forty samples (20 with myoglobin concentrations of 30–50 µg/L,
16 with 50–200 µg/L, and 4 with 200–400 µg/L) were tested on each
system in duplicate, 8 per day on 5 different days, using the
manufacturers’ control samples to validate the analytical runs
(7). To minimize the effects of carryover and analytical
drift on the averages of the duplicates within the run, the two
aliquots of each sample were analyzed in reverse order in the run
(7). Because none of the evaluated assays was considered the
"reference" procedure, we used the Deming regression to calculate
analytical correlation. Difference plots were also obtained
(8).
reference values
Two hundred eight apparently healthy subjects (103 women and 105
men; ages 13–87 years) were studied. In addition to the general a
priori exclusion criteria suggested by IFCC documents on the theory of
reference values, creatine kinase activity in serum, a widely accepted
biochemical marker of muscle tissue, was used as an additional
criterion of selection to exclude individuals who exercised or followed
a physical training program (9). In particular, 32 subjects
(from the 240 initially selected) with creatine kinase activities >150
U/L (measured at 37 °C according to the IFCC recommended method)
were excluded to minimize the possible influence of excessive physical
activity on myoglobin concentrations. Samples on each platform were
subdivided into four different analytical runs, carried out in
different days using the same calibration curve. The data obtained were
validated by assaying the manufacturers’ control sera at the
beginning, in the middle, and at the end of each run. Reference limits
were calculated using nonparametric determination of percentiles
(10).
|
Results
|
|---|
determination of the detection limit
All methods/instruments but one (Hitachi in laboratory 4) gave
experimental detection limits lower than values declared by
manufacturers. In particular, the following limits of detection were
obtained: Access, 5.0 and 3.3 µg/L (declared, 8.9 µg/L); Hitachi,
1.6 and 4.6 µg/L (declared, 3.0 µg/L); Opus, 0.2 and 0.4 µg/L
(declared, 1.0 µg/L); and Stratus, 1.2 and 0.4 µg/L (declared, 1.2
µg/L).
linearity
The results of the linearity studies are shown in Table 1
. Because of the wide calibration range, experiments on the
Access were repeated in a lower concentration range. In some cases, the
hypothesis of equality of variances across dilutions was rejected,
indicating relatively high imprecision in the results of each dilution
tested four times (Access in the high concentration range; BNA
and Opus in laboratory 5) or very small variances in the assays, such
as the Hitachi in laboratory 1, which showed good precision. When we
tested the results for lack of fit of the linear model, all assays
except for the Hitachi showed acceptable linearity. In the case of the
Hitachi, a very precise assay, data declared statistically nonlinear
(G values, 8.36 and 23.31) were, however, clinically
acceptable at visual inspection, showing the highest correlation
coefficients as well (Table 1
).
precision studies
The imprecision results for the various assays are shown in Table 2
. The BNA and Opus showed relatively high imprecision. The other
assays had lower CVs, but they varied among laboratories performing the
same assay. Analytical quality specifications analysis was performed to
test whether the precision of each method was satisfactory
(11). According to data on the biological variation of serum
myoglobin (12), the analytical goal (expressed as CV) for
optimum performance is <2.8%, the goal for desirable performance is
<5.6%, and the goal for minimum performance is <8.3%. On the basis
of intralaboratory variations, none of the assays evaluated showed
optimum performance for all three types of samples: two instruments in
one laboratory (Hitachi and Stratus in laboratory 4) showed desirable
performance, and three instruments in one (Access in laboratory 3 and
Opus in laboratory 2) or two laboratories (BNA) exceeded the
requirements for minimum performance in one or more samples. In all,
two methods performed best for analytical precision: the Hitachi and
the Stratus.
interference studies
With the Access, results were significantly lower in all
anticoagulated samples (n = 48 for each type of anticoagulant).
The mean difference (95% confidence interval) for heparin samples was
-9.8% (-16.0% to -3.6%; P = 0.003); the mean
difference for EDTA was -13.8% (-20.1% to -7.5%; P
<0.001); and the mean difference for citrate was -30.2% (-36.7% to
-23.6%; P <0.001). With the Stratus, the mean difference
for heparin was -6.0% (-13.0% to -1.0%; P =
0.002); and the mean difference for citrate was -18.6% (-22.0% to
-15.2%; P <0.001). For citrate samples, the BNA had a
mean difference of -13.1% (-25.8% to -0.4%; P =
0.046; n = 6), and the Opus had a mean difference of -12.1%
(-21.7% to -2.5%; P <0.001; n = 51). However, on
the Hitachi (n = 34), heparin [10.6% (5–15.5%); P
<0.001] and EDTA [7.8% (4–11.4%); P <0.001] results
were increased. The use of separator gel produced results significantly
lower on the Hitachi [-25.8% (-31% to -20.0%);
P <0.001] and higher on the Opus [6.1% (1–10.7%);
P = 0.016].
All tested endogenous interferents had statistically (P
<0.05) significant positive or negative effects on the BNA and Hitachi
assays, whereas none interfered in the Access. Considering the a priori
established limit of acceptability (± 10% in comparison with control
values), only bilirubin (300 mg/L) on the Opus and Stratus, the two
immunoassays with fluorometric detection, caused markedly negative
interference (mean difference between myoglobin results in test and
control pools, -75% and -76% on the Opus and -24% and -21% on
the Stratus for normal and high myoglobin pools, respectively). Because
purified bilirubin (mixed isomers) from bovine gallstones (Sigma),
which has a high intrinsic fluorescence, was used as interferent in
this experiment, we decided to reevaluate the bilirubin interference
with a different experimental method, by preparing a dilution series
using human serum samples with high (510 mg/L) and low (3 mg/L)
bilirubin concentration. At a bilirubin concentration of 306 mg/L, mean
myoglobin recoveries were 96.1% on the Opus and 100.9% on the
Stratus, showing no influence by human serum bilirubin on these assays.
Among the evaluated assays, only the Access produced results higher
(1.3- to 4.3-fold) than the manufacturer’s recommended cutoff (70
µg/L) in samples with high concentrations of RF (>600 kIU/L),
whereas the other four myoglobin assays gave results for all samples
below the cutoffs recommended by each of the manufacturers. However,
there appears to be little relationship between RF concentrations and
the degree of assay interference.
method comparison
The Deming correlations for different assays are summarized in
Table 3
. Although good correlation coefficients were observed, there
were significant proportional and constant biases in these data sets.
The Stratus assay typically produced the highest results, whereas the
Access and the BNA gave the lowest. All correlations involving the
Hitachi showed a significant positive constant bias (intercepts,
9.9–17.3 µg/L). Fig. 1
shows the differences between methods, expressed as a
percentage of the average, plotted against the method average. It is
apparent from Fig. 1
that there was a systematic bias in the
correlations involving the Stratus and the Hitachi and that for the
latter there was also a clear relationship between the method
difference and the myoglobin concentration.
View larger version (35K):
[in this window]
[in a new window]
|
Figure 1. Plots of the differences among the evaluated methods for
myoglobin determination.
Percentage of difference (y axis) = [(method
A - method B)/average of the two methods] x 100. If the 95%
confidence interval (CI) of the mean difference includes
0, there is no evidence of systematic bias between the compared
assays.
|
|
reference values
Because the preliminary evaluation of findings showed a
significant sex-related difference in myoglobin concentrations in the
reference population, with P values ranging from 0.032 for
the Hitachi to <0.0001 for the Access, Opus, and Stratus, the
calculation for the reference limits for these assays was performed
separately for men and women. In particular, the reference limits
[95% central range of the distribution (and the 99th percentile
value) in men and women, respectively] were estimated to be 5.1–69.5
µg/L (78.5 µg/L) and 3.3–51.5 µg/L (81.2 µg/L) for the Access;
9.6–67.0 µg/L (81.4 µg/L) and 8.5–57.7 µg/L (88.9 µg/L) for
the Hitachi; 13.8–79.8 µg/L (93.4 µg/L) and 11.0–49.9 µg/L
(83.1 µg/L) for the Opus; and 11.1–85.5 µg/L (93.3 µg/L) and
13.4–62.7 µg/L (112.4 µg/L) for the Stratus (although values below
the lower limit have no clinical significance). Results obtained with
the BNA assay showed no statistically significant sex-related
difference (P = 0.151) because >75% of the myoglobin
values from the reference individuals were lower than the detection
limit of the assay. The upper reference limits for the BNA were 51.4
and 49.4 µg/L for men and women, respectively. The corresponding 99th
percentile values were 74.7 and 82.0 µg/L. No age-related trends
(r 
0.26) in myoglobin values were detected for any method.
|
Discussion
|
|---|
Over the past decade, there have been significant changes in the
biochemical approach for the evaluation of patients with chest pain and
suspected myocardial injury. Because of the rapid progress in
biochemical knowledge and technological improvements, more sensitive
and specific tests for early diagnosis, triage, and risk stratification
of patients with acute coronary syndrome have been developed
(13). Recommendations have been published for the optimal
use of these new markers, suggesting as best compromise for diagnosis
of myocardial infarction the association of an early marker (which
increases in blood within 6 h after symptom onset) and a
cardiac-specific marker, i.e., cardiac troponin
(3)(14). As recently confirmed in a systematic
review of the literature (15), myoglobin is at present the
commercially available marker that most effectively fits the role of an
early marker. For this reason, practically all of the diagnostic
companies offering a cardiac marker panel on various heterogeneous or
homogeneous systems have implemented assays for this protein. Of
course, it would be in the best interests of the medical community and
of in vitro diagnostic industries that results from different assays
resemble each other. The data obtained in our study show that the
assays correlate well, but they do not agree in standardization.
Furthermore, the differences in performance of the same
method/instruments in different laboratories underline that current
routine analysis of myoglobin may also lack congruity among
laboratories using the same platform. Although some of the evaluated
methods have recently been replaced by new efficient procedures,
the results obtained in a national External Quality Assessment Scheme
for cardiac markers, organized by our Working Group, confirm that
for myoglobin, as well as for troponin I, the results of some
commercial analyzers are significantly different from each other (data
not shown). Biases observed in patients’ samples and the
corresponding differences in the upper reference limits for different
myoglobin assays arise directly from a standardization problem mainly
attributable to the lack of an accepted reference material for
myoglobin and of a high-quality reference method; a dedicated effort by
international scientific bodies, such as the IFCC Committee on
Standardization of Markers of Cardiac Damage, is therefore needed to
achieve harmonization of myoglobin results from different assays
(16).
In this multicenter study, we have also shown that some commercial
assays for myoglobin determination do not meet the desired target for
imprecision, which is derived directly from data on biological
variation (11). Although the overall intralaboratory
variation was <11% in this study, this variation may be clinically
unacceptable, and an improvement in the precision of measurements is
needed if myoglobin is to be offered on a routine clinical basis
(12). Great attention was also given to the possible
influence of different types of anticoagulants on the measured
myoglobin values and to potential interferences by most important
endogenous substances. National and international committees agree that
plasma is the specimen of choice for cardiac marker analysis
(3)(14). However, we have demonstrated that
different myoglobin assays can show significant anticoagulant
interferences, which precludes the use of the anticoagulant to
eliminate the extra time needed for clotting and to reduce the overall
preanalytical time.
Interference from RF, which can mimic the measured protein, has been
reported recently in commercial immunoassays for cardiac markers
(17)(18)(19). We found several false positives in the Access
myoglobin assay attributable to RF. In agreement with previously
published findings (17), there was no concordance between RF
concentrations and the degree of interference. RFs are a group of
heterogeneous immunoglobulins whose common property is their ability to
react with the Fc portion of the IgG molecule. This heterogeneity
probably accounts for the lack of uniform interference in immunoassays
(17). Interestingly, in a recent study, measurement of
cardiac troponin I on the same platform was unaffected by RF, which may
reflect a difference in diluent composition or in sample dilution in
the two Access assays (20).
In conclusion, the results of this study stress the importance of
achieving standardization of myoglobin measurements, also in terms of
maximum allowable imprecision and lack of interference. Until then, the
possibility of high imprecision and significant disagreement among
commercial myoglobin assays should be carefully considered in clinical
practice.
|
Acknowledgments
|
|---|
The manufacturers of the myoglobin assay kits examined here
generously provided us with materials, information, and advice, for
which we are grateful.
|
Footnotes
|
|---|
Presented in part at the 50th AACC National Meeting, August 2–6, 1998,
Chicago, IL (Clin Chem 1998;44:A121).
|
References
|
|---|
-
Plebani M, Zaninotto M. Diagnostic strategies using myoglobin measurement in myocardial infarction. Clin Chim Acta 1998;272:69-77.[Medline]
[Order article via Infotrieve]
-
Kagen L, Scheidt S, Roberts L, Porter A, Paul H. Myoglobinemia following acute myocardial infarction. Am J Med 1975;58:177-182.[Web of Science][Medline]
[Order article via Infotrieve]
-
Panteghini M, Apple FS, Christenson RH, Dati F, Mair J, Wu AHB. Use of biochemical markers in acute coronary syndromes. Clin Chem Lab Med 1999;37:687-693.[Web of Science][Medline]
[Order article via Infotrieve]
-
. National Committee for Clinical Laboratory Standards. Evaluation of the linearity of quantitative analytical methods; proposed guideline EP6-P 1986 NCCLS Wayne, PA. .
-
. National Committee for Clinical Laboratory Standards. Evaluation of precision performance of clinical chemistry devices; approved guideline EP5-A 1999 NCCLS Wayne, PA. .
-
. National Committee for Clinical Laboratory Standards. Interference testing in clinical chemistry; proposed guideline EP7-P 1986 NCCLS Wayne, PA. .
-
. National Committee for Clinical Laboratory Standards. Method comparison and bias estimation using patient samples; approved guideline EP9-A 1995 NCCLS Wayne, PA. .
-
Pollock MA, Jefferson SG, Kane JW, Lomax K, MacKinnon G, Winnard CB. Method comparison—a different approach. Ann Clin Biochem 1992;29:556-560.
-
PetitClerc C, Solberg HE. IFCC approved recommendation (1987) on the theory of reference values. Part 2. Selection of individuals for the production of reference values. J Clin Chem Clin Biochem 1987;25:639-644.
-
Solberg HE. IFCC approved recommendation (1987) on the theory of reference values. Part 5. Statistical treatment of collected reference values. Determination of reference limits. J Clin Chem Clin Biochem 1987;25:645-656.[Web of Science]
-
Fraser CG, Hyltoft Petersen P, Libeer JC, Ricos C. Proposals for setting generally applicable quality goals solely based on biology. Ann Clin Biochem 1997;34:8-12.
-
Panteghini M, Pagani F. Biological variation of myoglobin in serum [Letter]. Clin Chem 1997;43:2435.[Free Full Text]
-
Panteghini M. Biochemical assessment of myocardial damage with new diagnostic tools [Review]. Cardiologia 1999;44:419-425.[Medline]
[Order article via Infotrieve]
-
Wu AHB, Apple FS, Gibler WB, Jesse RL, Warshaw MM, Valdes R, Jr. National Academy of Clinical Biochemistry Standards of Laboratory Practice: recommendations for the use of cardiac markers in coronary artery diseases. Clin Chem 1999;45:1104-1121.[Abstract/Free Full Text]
-
Panteghini M, Pagani F, Bonetti G. The sensitivity of cardiac markers: an evidence-based approach. Clin Chem Lab Med 1999;37:1097-1106.[Web of Science][Medline]
[Order article via Infotrieve]
-
Dati F, Panteghini M, Apple FS, Christenson RH, Mair J, Wu AH. Proposal from the IFCC Committee on Standardization of Markers of Cardiac Damage (C-SMCD): strategies and concepts on standardization of cardiac marker assays. Scand J Clin Lab Investig 1999;59(Suppl 230):113-123.
-
Krahn J, Parry DM, Leroux M, Dalton J. High percentage of false positive cardiac troponin I results in patients with rheumatoid factor. Clin Biochem 1999;32:477-480.[Web of Science][Medline]
[Order article via Infotrieve]
-
Dasgupta A, Banerjee SK, Datta P. False-positive troponin I in the MEIA due to the presence of rheumatoid factors in serum. Am J Clin Pathol 1999;112:753-756.[Web of Science][Medline]
[Order article via Infotrieve]
-
Onuska KD, Hill SA. Effect of rheumatoid factor on cardiac troponin I measurement using two commercial measurement systems [Letter]. Clin Chem 2000;46:307-308.[Free Full Text]
-
Pagani F, Stefini F, Bonetti G, Panteghini M. Evaluation
of interference effects from heterophilic antibodies in four commercial
platforms for cardiac marker measurement [Abstract]. Proceedings of
the Meeting on "Basic and Clinical Enzymology 2000", May 21–24,
Naples, Italy, 2000:109..
The following articles in journals at HighWire Press have cited this article:
|
|
|
|
 
D. H. Wilson, G. Williams, R. Herrmann, D. Wiesner, and P. Brookhart
Issues in Immunoassay Standardization: The ARCHITECT Folate Model for Intermethod Harmonization
Clin. Chem.,
April 1, 2005;
51(4):
684 - 687.
[Full Text]
[PDF]
|
|
|
Top
Abstract
Introduction
