Performance Evaluation of the ABX Micros CRP:

ISLH
Laboratory Hematology 7:69-74
© 2001 Carden Jennings Publishing Co., Ltd.
Official Publication
Performance Evaluation of the ABX Micros CRP:
The First Instrument Reporting a Complete Blood
Count, 3-Part Leukocyte Differential, and
C-Reactive Protein Quantitation
BRUCE H. DAVIS, NANCY C. BIGELOW
Maine Medical Center Research Institute, Scarborough, Maine
significant carryover. Correlation of CRP with the Roche
COBAS Integra was excellent with a slope of approximately 0.98, no significant bias, and an r 2 value of >0.99. No
interference was seen in the analysis of any of the samples
from patients with clinical conditions associated with interfering substances. Samples held at either 25°C or 4°C were
stable for 48 hours. The CRP sensitivity was greater than
the manufacturer’s claim of 0.2 mg/dL. The ABX Micros
CRP demonstrated performance suitable for clinical practice. The ease of use, small sample volume (18µL), and
rapid turnaround time (<5 minutes for CBC and CRP) are
features that will make this instrument useful in point-ofcare testing areas including both adult and pediatric intensive care units and in office laboratories, particularly those
of pediatricians, rheumatologists, and infectious disease
specialists. Lab. Hematol. 2001;7:69-74.
ABSTRACT
ABX Diagnostics has developed the ABX Micros CRP
analyzer, the first instrument to provide both a complete
blood count (CBC) with a 3-part differential and a quantitative C-reactive protein (CRP) determination. The CRP
measurement is based on a whole-blood immunoturbidimetric method using latex beads coated with antiCRP–specific–monoclonal antibody. The evaluation
reported here of the linearity, sensitivity, precision, carryover, and clinical performance of the ABX Micros CRP was
performed prior to release of the instrument in the United
States. Precision of CRP analysis was assessed by replicate
determinations of K3EDTA-anticoagulated whole-blood
samples with CRP values in the range of <0.4 mg/dL to
10 mg/dL. Linearity was evaluated by assaying pooled normal samples spiked with CRP control material. The stability of CRP was studied in blood samples held at both room
temperature and refrigerated conditions over a 72-hour
period. Correlation studies were performed by analyzing
100 blood samples on the ABX Micros CRP and comparing the results with those obtained using the Roche
COBAS Integra CRP assay, which is based on latex nephelometry. Within-run and between-run precision of both
CBC and CRP values had a coefficient of variation (CV) of
<10%. Excellent linearity was observed and there was no
KEY WORDS:
Inflammation · Hematology
instrumentation · Point-of-care
testing · Laboratory automation ·
Acute phase reactant
INTRODUCTION
One of the recent trends in laboratory medicine is the
merging of testing in the traditionally separate disciplines of
hematology and chemistry. This merger has created the
unfortunate term, chematology, and has been driven by the
desire to save money. One expensive, still experimental,
approach to achieving that goal has been to put hematology
and chemistry instruments on the robotic track, forcing
Correspondence and reprint requests: Bruce H. Davis, MD, Maine
Medical Center Research Institute, 81 Research Drive Scarborough,
ME 04074 (e-mail: [email protected]).
Received December 1, 2000; accepted December 4, 2000
69
70
B.H. Davis and N.C. Bigelow
data generated with this system correlates with that of the
FDA-approved Roche COBAS Integra CRP assay, which is
based on latex nephelometry.
MATERIALS AND METHODS
Stability
Five samples from healthy patients were divided into 2 aliquots each, 1 of which was placed at room temperature and
the other at 4°C. At baseline (immediately after blood collection) and at 24, 48, and 72 hours post-collection, each aliquot
was carefully mixed and tested in duplicate using the ABX
Micros CRP. Results are shown in Table 2 and are expressed as
percentage deviation from the baseline measurement.
FIGURE 1. MICROS CRP instrument from ABX DIAGNOSTICS (Montpellier, France) provides a CBC with a 3part leukocyte differential and C-reactive protein values on
whole-blood samples.
technical specialists to become cross-trained generalists and
on-site computer and mechanical repair experts. An alternative approach is to focus on the point-of-care arena and create
specialized instruments that are focused on specific clinical
needs. One such arena is the outpatient or specialized inpatient setting in which patients with suspected or documented
inflammatory diseases such as infections and rheumatoid diseases are evaluated. ABX Diagnostics (Montpellier, France)
has developed an instrument that provides both a complete
blood count (CBC) with a 3-part leukocyte differential and a
quantitative C-reactive protein (CRP) measurement, a combination of testing that is tailored to this setting.
The ABX Micros CRP (Figure 1) is a fully automated
hematology analyzer identical to the ADVIA 60 (Bayer
Diagnostics, Tarrytown, NY), an FDA-approved hematology
analyzer that measures various CBC parameters, including 3part differential leukocyte counts, hematocrit, hemoglobin,
mean corpuscular volume, mean corpuscular hemoglobin,
mean corpuscular hemoglobin concentration, red blood cell
counts and distribution width, platelet count, and a mean
platelet volume and distribution width. ABX has modified
the Micros to create the ABX Micros CRP by integrating
another assay using turbidimetric measurements with monoclonal antibody–coated latex beads to measure CRP in whole
blood. CRP is an acute-phase reactant that provides diagnostic information similar to the classic erythrocyte sedimentation rate (ESR) and is an indicator of tissue injury or disease
activity in inflammatory processes [1-7]. The purpose of this
study was to evaluate the performance characteristics of the
ABX Micros CRP and to demonstrate the extent to which
Precision
A blood sample from a healthy subject was used to estimate the within-run precision for the normal range using
replicate testing (>10 times). Serial analyses of low-control
(LC) and high-control (HC) specimens (ABX Diagnostics)
were performed to determine between-run and between-day
precision. All normal and abnormal blood specimens were
analyzed in duplicate using paired-precision analysis to assess
day-to-day intra-assay precision.
Linearity
ABO- and Rh-matched blood samples with high and low
CRP levels were mixed in the following proportions and
assayed in duplicate: 100%:0%, 90%:10%, 80%:20%,
65%:35%, 50%:50%, 35%:65%, 20%:80%, 5%:95%,
0%:100%. In addition, known amounts of CRP control
material were added to 0.5 mL aliquots of a normal blood
sample in increments of 0.2 mg and run in duplicate.
Carryover
The ABX Micros CRP carryover was evaluated by assaying the LC specimens in triplicate (LC1-3), followed by triplicate testing of the HC specimens (HC1-3), then rerunning
the LC in triplicate again (LC4-6). It is assumed that carryover
TABLE 1. Carryover for the ABX Micros CRP Assay*
Sample
Number 1
Number 2
Number 3
Mean ± SD
UL/CI
Carryover (%)†
Low Control
High Control
Low Control
0.57
0.60
0.59
0.59 ± 0.01
0.603955
0.13
4.46
4.68
4.65
4.60 ± 0.10
0.61
0.57
0.59
0.59 ± 0.02
*SD indicates standard deviation; UL, upper limit of 95% confidence
interval; CI, confidence interval.
†(0.61 – 0.603955) ÷ 4.60 100.
Performance Evaluation of the ABX Micros CRP
71
by medical chart review using the presence of infection or
other inflammatory disease as the clinical indicator.
Investigation of Clinical Conditions Associated
With Interfering Substances
Samples were obtained from patients with conditions
known to be associated with hyperbilirubinemia, hyperhemoglobinemia, hypergammaglobulinemia, and hyperlipidemia, and evaluated with the ABX Micros CRP; the results
were evaluated using paired precision and clinical correlation.
This evaluation is not meant to be an exhaustive study, but
does provide information about the testing of patients with
clinical conditions associated with substances known to
interfere with other CRP methods.
Statistical Analysis
Linearity and procedural correlation data were analyzed
using linear regression analysis, and comparative data on
patient and control specimens were evaluated using analysis
of variance methods when appropriate. All statistical procedures were performed using Statistica for Windows (StatSoft,
Tulsa, OK) and Excel (Microsoft, Redmond, WA).
RESULTS
FIGURE 2. Linearity of C-reactive protein (CRP) measurements on ABX Micros CRP.
has not occurred if the LC4 value lies within the 95% confidence interval (CI) defined by the LC1-3 replicates. If the LC4
value was greater than the upper limit of the 95% CI (UL),
then the following formula was used to calculate carryover:
(LC4 – UL/CI) ÷ (HC1-3 ÷ 3), where LC4 is the first low
control replicate following the high control testing, UL/CI is
the upper limit of the 95% CI, and HC1-3 ÷ 3 is the average
of the triplicate high control values.
Procedural Correlation
Twenty-five samples from healthy patients and 75 samples
from patients with diseases in which CRP levels are likely to
be elevated were assayed in duplicate with the ABX Micros
CRP. These values were compared with the value obtained
with the Roche COBAS Integra CRP assay according to institution-approved guidelines and Institutional Review Board
(IRB) approval. All samples were stored at room temperature
and tested within 12 hours of collection, except for the stability evaluation phase of the study (described above). When
samples were assayed on multiple instruments, not more than
1 hour elapsed between the beginning of testing with the first
instrument and the beginning of testing with the second.
Sensitivity and specificity for clinical utility was determined
Assay Precision
A blood specimen from a healthy subject was assayed
10 times in succession and the mean level of CRP was found
to be 0.39 ± 0.02 mg/dL, with a coefficient of variation (CV)
of 6.0%. The within-run precision derived from paired duplicate analyses of CRP on all patient samples and healthy donors
was a mean CV of 17.6% ± 37.3% (range 0% to 200%, N =
100). When those samples that had CRP levels below the linearity limit determined by this study (0.05 mg/dL) were
excluded, low imprecision was demonstrated with a mean CV
of 6.5% ± 11.7% (range 0% to 72.7%, n = 86).
The between-run precision estimates were obtained by
assaying the LC and HC specimens 4 times at approximately
2-hour intervals every day for 17 days. CVs ranged from 1.4%
to 11.6% (mean, 4.1% ± 2.8%) for the LC with an overall
mean of 0.56 ± 0.03 mg/dL, and 0.5% to 3.1% (mean, 1.8%
± 0.7%) for the HC with an overall mean of 4.25 ±
0.14 mg/dL. Between-day precision studies for the LC and
HC specimens produced CVs of 5.4% and 3.3%, respectively.
Assay Linearity
The linearity of the ABX Micros CRP was initially assessed
using duplicate assays of dilutions of ABO- and Rh-matched
blood samples with high and low CRP levels mixed in various
proportions and showed a loss of linearity in undiluted specimens with CRP levels between 7 mg/dL and 10 mg/dL (Figure 2). Specimens with CRP values above 10 mg/dL were
diluted prior to analysis according to the product manual and
showed good linearity. Linearity between the normal range
72
B.H. Davis and N.C. Bigelow
for the samples with the lowest CRP levels, with increases of
5% to 10% at 48 hours in samples stored at both ambient
and low temperatures, in contrast to decreases of 1% to 3%
for those with elevated CRP levels (Table 2). At 72 hours
post-collection, deviations from baseline were more apparent, but were still less than 11% for both normal and abnormal samples irrespective of storage temperature.
FIGURE 3. Intermethod correlation between ABX Micros
CRP and Roche Integra CRP (latex) of 100 samples. CRP
indicates C-reactive protein.
and approximately 7 mg/dL were studied further by adding
increasing volumes of 5 µL to 230 µL of a sample with a high
CRP level (21.3 mg/dL) to 0.5 mL aliquots of a normal specimen and showed excellent linearity (Figure 2).
Procedural Correlation
Samples were collected from 25 healthy subjects and from
75 subjects with known disease (44% female), of which
42 were older than 4 years (mean age, 35.3 years) and 58 were
younger than 1 year. Samples were assayed in duplicate using
the ABX Micros CRP and Roche COBAS Integra. The correlation coefficient between the 2 methods was 0.975 for all samples tested (Figure 3), with the ABX Micros CRP method
exhibiting a greater deviation from regression in the 7 to
10 mg/dL range, in which whole-blood samples were not diluted, in accordance with the manufacturer’s original instructions.
When only the 25 normal samples were tested, the 2 methods
were almost perfectly correlated (r 2 = 0.996, Figure 4).
Investigation of Potentially Interfering Substances
During the course of the evaluation, 2 samples were found
to have potentially interfering substances for a CRP assay. One
sample with hyperhemoglobinemia secondary to an active
malaria infestation gave clinically relevant duplicate values of
6.61 mg/dL and 6.92 mg/dL. A second sample was from a
patient with hyperlipidemia (cholesterol, >1000 mg/dL) and a
hypergammaglobulinemia (>3 g/dL), but no clinical evidence
of active inflammation. The ABX Micros CRP gave appropriately normal CRP values for the sample with both hyperlipidemia and hypergammaglobulinemia (0.18 mg/dL and
0.19 mg/dL). Both samples demonstrated a CV with paired
precision comparable to samples without these potentially
compounding substances. These findings show that although
users should be aware of the potential for interfering substances
to cause erroneous results, the presence of hyperlipidemia,
hypergammaglobulinemia, and hyperhemoglobinemia are not
absolute contraindications for using the ABX Micros CRP.
DISCUSSION
The data presented in this study demonstrate that withinrun, between-run, and between-day precision for the ABX
Micros CRP assay is quite acceptable, with the best precision
Sample Carryover
As shown in Table 1, the CRP value for the first LC sample following triplicate testing of the 3 HC samples was
slightly above the upper limit of the 95% CI for the 3 low
level controls initially tested (0.610 versus 0.604) resulting in
a carryover estimate of 0.13%.
Sample Stability
Five normal and 5 abnormal blood specimens were tested
immediately after collection and following storage at both
ambient (approximately 25°C) and low temperatures
(approximately 5°C) at 24, 48, and 72 hours. As expected,
the percentage deviation from baseline was somewhat greater
FIGURE 4. Intermethod correlation between ABX Micros CRP
and Roche Integra CRP (latex) of 25 healthy donor samples
show excellent agreement. CRP indicates C-reactive protein.
Performance Evaluation of the ABX Micros CRP
TABLE 2. Stability of Normal and Abnormal Samples Assayed on ABX Micros
CRP (n = 10)*
Time (h)
Variable
Normal sample RT
Mean
Deviation
Percentage deviation
Abnormal sample RT
Mean
Deviation
Percentage deviation
Normal sample at 5°C
Mean
Deviation
Percentage deviation
Abnormal sample at 5°C
Mean
Deviation
Percentage deviation
0
24
48
72
0.131
0.138
0.007
5.3
0.139
0.008
6.1
0.142
0.011
8.4
4.054
3.981
0.073
–1.8
4.002
0.052
–1.3
3.730
0.324
–8.0
0.131
0.144
0.013
9.9
0.139
0.008
6.1
0.117
0.014
–10.7
4.054
3.935
0.119
–2.9
3.995
0.059
–1.5
3.816
0.238
–5.9
*Mean CRP values expressed as mg/dL. Deviation is the deviation of
the sample mean from baseline. Percentage deviation is the mean deviation
expressed as a percentage. RT indicates room temperature.
obtained for the low control (0.56 mg/dL) tested 4 times per
day for 17 days (CV 1.4 to 11.6%). This CRP assay is linear
for anticoagulated blood without dilution up to a CRP value
of 7 mg/dL. Although the ABX Micros CRP was originally
designed for CRP concentrations up to 10 mg/dL in undiluted blood, this study demonstrated a slight loss of linearity
in the range of 7 to 10 mg/dL. Thus, our findings indicate
that samples with CRP values > 7 mg/dL on initial analysis
with the ABX Micros CRP should be diluted and reanalyzed
to obtain accurate results. When this protocol was followed,
we observed results from the ABX Micros CRP assay to be
highly correlated with those of the Roche COBAS Integra
CRP. Carryover was found to be negligible and samples
stored for up to 48 hours at either room temperature or
under refrigeration were shown to be stable for the ABX
Micros CRP assay. Finally, we observed no interference in
the CRP measurements in a limited sampling of patients
with rheumatoid factor, hyperhemoglobinemia, hyperlipidemia, or hypergammaglobulinemia, conditions associated
with interference problems with other CRP methods [8].
Thus our findings support the claim that the ABX Micros
CRP provides reproducible, linear measurement of CRP in
the range of 0.05 to 7.0 mg/dL using undiluted, EDTAanticoagulated whole-blood samples and >7.0 mg/dL using
diluted samples. Furthermore, the ease of operation, small
sample volume requirement, and reporting of both CBC and
CRP results in less than 5 minutes are features that make this
a suitable instrument for point-of-care testing.
73
An elevated CRP level is indicative of inflammatory disease activity and is preferred to the ESR as a disease marker.
But the simpler nature of the ESR procedure means it is more
available and that results are reported out more frequently
than the immunologic CRP method. If it is available, the
CRP assay is typically run in batches, and results are reported
once daily (or even less frequently), precluding its use in realtime patient management. Yet it is clear that the CRP has
clinical utility in the management of neonatal sepsis [9-14],
suspected appendicitis in the acute care setting [1,2,15], and
more recently, in cardiac injury and stroke [16-20]. Such clinical conditions typically involve patients in either intensive
care settings or emergency care, where the use of point-of-care
testing has been found to improve patient care and reduce
health care costs. It is clear that instruments such as the ABX
Micros CRP should provide a solution to a real clinical need.
The CRP measurement with the ABX Micros CRP was
shown to have a lower limit of linearity down to 0.05 mg/dL,
which is not sufficient to be defined as a high-sensitivity CRP
method as employed in many of the studies demonstrating
CRP as a risk factor in cardiovascular diseases. However, the
ABX Micros CRP appears to be sufficiently sensitive to identify not only the elevated CRP values that occur in neonatal
sepsis, but also those in patients with subtler abnormalities
that have been found in studies using high-sensitivity CRP
methods which define prognostic risk at threshold values in
the range of 0.2 to 0.4 mg/dL [16-18,20]. Therefore, aside
from the intended use of the ABX Micros CRP in the physician office setting of rheumatologist, pediatrician, or internist,
this instrument may find utility in point-of-care settings such
as neonatal intensive care units, emergency care units, or outpatient cardiac care settings.
The clinical use of the CRP is generally well defined as a
marker in the patient with infection or sepsis, although its
use may be challenged by newer diagnostic tests of responses
to infections, such as procalcitonin, interleukin-6, and neutrophil CD64 measurements [11,21-27]. But at least for
now, the advantage of combining CBC and CRP determinations on a single instrument platform, such as the ABX
Micros CRP, should be a valuable diagnostic tool for reliable
point-of-care testing of individuals with proven or suspected
inflammatory diseases. The combination of CBC parameters
with other laboratory determinations may be a new trend in
laboratory hematology.
ACKNOWLEDGMENTS
This work was supported by ABX Diagnostics, Irvine,
CA, and was performed as part of the clinical evaluation for
the FDA 510(k) submission of the ABX Micros CRP. Dr.
Davis and Ms. Bigelow are paid consultants to ABX Diagnostics, Montpellier, France.
This work was presented in part at the annual ISLH
meeting in Banff, Canada, in April 2000 and the American
74
B.H. Davis and N.C. Bigelow
Society of Hematology meeting in San Francisco, California,
in December 2000.
REFERENCES
1. Clyne B, Olshaker J. The C-reactive protein. J Emerg Med.
1999;17:1019-1025.
2. Gronroos J, Gronroos P. Leucocyte count and C-reactive protein
in the diagnosis of acute appendicitis. Br J Surg. 1999;86:501-504.
3. Yeun J. Clinical utility of C-reactive protein measurements. Semin
Dial. 2000;13:56-57.
4. Szalai A, Agrawal A, Greenhough T, Volanakis J. C-reactive protein: structural biology and host defense function. Clin Chem Lab
Med. 1999;37:265-270.
5. Rodriguez-Sanjuan J, Martin-Parra J, Seco I, Garcia-Castrillo L,
Naranjo A. C-reactive protein and leukocyte count in the diagnosis of acute appendicitis in children. Dis Colon Rectum. 1999;42:
1325-1329.
6. Posen R, de LR. C-reactive protein levels in the extremely premature
infant: case studies and literature review. J Perinatol. 18:138-141.
7. Povoa P, Almeida E, Moreira P, et al. C-reactive protein as an indicator of sepsis. Intensive Care Med. 1998;24:1052-1056.
8. Lelong M, Renard M, Giraudeaux V. Solid immunoassay and
immunonephelemetric methods for serum C reactive protein
(CRP) determination: discrepancy in CRP value for a patient presenting a monoclonal immunoglobulin G. Clin Chim Acta.
1999;288:147-152.
9. Benitz W, Han M, Madan A, Ramachandra P. Serial serum
C-reactive protein levels in the diagnosis of neonatal infection
[abstract]. Pediatrics. 1998;102:965.
10. Bomela H, Ballot D, Cory B, Cooper P. Use of C-reactive protein
to guide duration of empiric antibiotic therapy in suspected early
neonatal sepsis. Pediatr Infect Dis J. 2000;19:531-535.
11. Carlet J. Rapid diagnostic methods in the detection of sepsis.
Infect Dis Clin North Am. 1999;13:483-494.
12. Chiu C, Lin T, Bullard M. Identification of febrile neonates
unlikely to have bacterial infections. Pediatr Infect Dis J.
1997;16:59-63.
13. Gerdes J, Polin R. Early diagnosis and treatment of neonatal sepsis. Indian J Pediatr. 1998;65:63-78.
14. Kaftan H, Kinney J. Early onset neonatal bacterial infections.
Semin Perinatol. 1998;22:15-24.
15. Asfar S, Safar H, Khoursheed M, Dashti H, al-Bader A. Would
measurement of C-reactive protein reduce the rate of negative exploration for acute appendicitis? J R Coll Surg Edinb. 2000;45:21-24.
16. Ford E, Giles W. Serum C-reactive protein and self-reported stroke:
findings from the Third National Health and Nutrition Examination Survey. Arterioscler Thromb Vasc Biol. 2000;20:1052-1056.
17. Ridker P, Hennekens C, Buring J, Rifai N. C-reactive protein and
other markers of inflammation in the prediction of cardiovascular
disease in women. N Engl J Med. 2000;342:836-843.
18. Tommasi S, Carluccio E, Bentivoglio M, et al. C-reactive protein
as a marker for cardiac ischemic events in the year after a first,
uncomplicated myocardial infarction. Am J Cardiol. 1999;83:
1595-1599.
19. Gaspardone A, Crea F, Versaci F, et al. Predictive value of C-reactive
protein after successful coronary-artery stenting in patients with
stable angina. Am J Cardiol. 1998;82:515-518.
20. Anderson J, Carlquist J, Muhlestein J, Horne B, Elmer S. Evaluation of C-reactive protein, an inflammatory marker, and infectious
serology as risk factors for coronary artery disease and myocardial
infarction. J Am Coll Cardiol. 1998;32:35-41.
21. Hatherill M, Tibby S, Sykes K, Turner C, Murdoch I. Diagnostic
markers of infection: comparison of procalcitonin with C reactive
protein and leucocyte count. Arch Dis Child. 1999;81:417-421.
22. Hedlund J, Hansson L. Procalcitonin and C-reactive protein levels
in community-acquired pneumonia: correlation with etiology and
prognosis. Infection. 2000;28:68-73.
23. Somech R, Zakuth V, Assia A, Jurgenson U, Spirer Z. Procalcitonin correlates with C-reactive protein as an acute-phase reactant
in pediatric patients. Isr Med Assoc J. 2000;2:147-150.
24. Toikka P, Irjala K, Juven T, et al. Serum procalcitonin, C-reactive
protein and interleukin-6 for distinguishing bacterial and viral
pneumonia in children. Pediatr Infect Dis J. 2000;19:598-602.
25. Karzai W, Reinhart K. Sepsis: definitions and diagnosis. Int J Clin
Pract Suppl. 1998;95:44-48.
26. Davis BH, Bigelow NC, Curnutte JT, Ornvold K. Neutrophil
CD64 expression: potential diagnostic indicator of acute inflammation and therapeutic monitor of interferon-γ therapy. Lab
Hematol. 19951:3-12.
27. Fjaertoft G, Hakansson L, Ewald U, Foucard T, Venge P. Neutrophils from term and preterm newborn infants express the high
affinity Fcgamma-receptor I (CD64) during bacterial infections.
Pediatr Res. 1999;45:871-876.