Professional Documents
Culture Documents
Bai 2008
Bai 2008
Hemostasis instrumentation has rapidly advanced and assays). Blood Coagul Fibrinolysis 19:569–576 ß 2008
laboratories are demanding fully automated coagulation Wolters Kluwer Health | Lippincott Williams & Wilkins.
systems. Two distinct technological families exist based on
optical and mechanical clot detection methodologies. Until
now, there have been no comprehensive studies to Blood Coagulation and Fibrinolysis 2008, 19:569–576
determine whether one methodology is superior to the
other. In order to answer this question, we conducted a large Keywords: clot detection, coagulation, mechanical detection, optical
detection
clinical study performing standard coagulation testing on
a
more than 2000 clinical samples randomly chosen from a Department of Pathology, Duke University Medical Center, Durham, North
Carolina and bDepartment of Clinical & Scientific Affairs, Dade Behring Inc.,
high-volume laboratory in a tertiary care hospital. Results Newark, Delaware, USA
demonstrated that photo-optical clot detection and electro-
Correspondence to Jogin R. Wu, PhD, Duke University Medical Center, DUMC
mechanical detection systems were highly correlated (r- 3514, 129C CARL Research Drive, Durham, NC 27710, USA
squared values > – 0.96 for all assays) Correlation between E-mail: jogin.wu@duke.edu
the two clot detection systems was maintained even when Received 31 October 2007 Revised 20 April 2008
measuring turbid samples (r-squared values > – 0.98 for all Accepted 30 April 2008
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
570 Blood Coagulation and Fibrinolysis 2008, Vol 19 No 6
will be possible to resolve discrepant results between the cleaning solution, Desorb solution, etc. were purchased
optical and mechanical detection methods better. from Diagnostica Stago.
Patient samples All three routine coagulation tests (PT, APTT and FIB)
All plasma samples were randomly selected from consecu- were performed with both the test and comparative
tive patient samples obtained from the normal course of method and, if necessary, the tilt tube method (described
the clinical laboratory’s daily routine for standard coagu- below), within 4 h of receiving the sample. Residual
lation testing. The laboratory is part of a high-volume samples were aliquoted and frozen for future use. A total
clinical setting at Duke University Medical Center, a of 2057 patient samples were included in the study during
tertiary care hospital that requires rapid turn-around-time a period of 45 days. The degree of visually observable
in diagnostic testing for emergency room, operating room, interference in study samples was characterized accord-
intensive care (ICU) and pediatric patients. All patient ing to a ‘Sample Integrity Chart’ (Ortho Clinical
samples were prospectively collected and tested for PT, Diagnostics, Johnson & Johnson Company, Rochester,
APTT and FIB following the laboratory’s routine proto- New York, USA), which provides semiquantitative
cols for these assays. Patient samples were collected (hemoglobin up to 50 mg/dl, lipid up to 150 mg/dl and
by clean venipuncture using siliconized glass tubes adult bilirubin up to 5 mg/dl) and qualitative (slight,
(Vacutainer Becton Dickinson, Franklin Lakes, New medium or gross) grading based on color/cloudiness.
Jersey, USA). Blood collection was performed with an Whenever either gross turbidity was observed or no clot
initial pilot tube that was discarded, followed by a second resulted or both with either the test or the comparative
tube (Becton Dickinson light blue tube) containing 3.2% method, the tilt tube method was performed on the
sodium citrate. For adult and pediatric patients, 4.5 ml and remaining sample, if available.
a 1.8 ml tubes, respectively, were filled to the proper line
to guarantee that a sufficient volume of blood had Plasma from 25 normal healthy individuals was assayed to
been drawn. Samples were immediately centrifuged at obtain the geometric means and reference ranges for PT,
3000 rpm (1500 g) for 15 min, and all samples were loaded APTT and FIB for both the CA-1500 and STA instru-
onto the laboratory’s primary coagulation instrument for ments, respectively, using EP Evaluator 6.0 software
routine testing. Following these standard tests (PT, APTT (David G. Rhoads Associates, Inc., Kennett Square,
and FIB), between 50 and 100 samples per day were Pennsylvania, USA). Daily, weekly and monthly quality
randomly selected and immediately delivered to the study control and maintenance were performed following the
instruments (STA and CA-1500). Inclusion criteria were all manufacturers’ recommended procedures. All assays for
routine plasma samples submitted for standard coagulation each patient sample were performed in singleton on each
testing. Test results for a particular method (e.g. FIB) were analyzer and the tilt tube procedure, performed by
excluded if there was insufficient sample volume for one expert technologist, within 4 h of blood collection.
analysis on both the CA-1500 and STA analyzers. Assaying samples on CA-1500 and STA was performed
within 1 h of each other and the tilt tube procedure not
Materials and methods longer than an additional hour. PT, APTT and FIB assays
The Sysmex CA-1500 (CA-1500) (distributed by Dade were measured following the manufacturer’s package
Behring, Deerfield, Illinois, USA), a fully automated insert instructions. For PT, 50 ml of patient sample was
photo-optical coagulation analyzer, was the test method incubated at 378C for 180 s, followed by the addition of
and the STA (Diagnostica Stago, Parsippany, New Jersey, 100 ml of Innovin reagent. For APTT, 50 ml of patient
USA), a fully automated electromechanical coagulation plasma was incubated for 60 s and then mixed with 50 ml
analyzer, served as the comparative method. To reduce of Actin FSL for another 240 s, followed by addition of
variability, only Dade Behring reagents were used on 50 ml of CaCl2 to start the reaction. For FIB, the Dade
the two test systems (CA-1500 and STA) and in tilt tube Behring standard human plasma and Dade thrombin
testing. Innovin was used for PT, Actin FSL for APTT, reagent were used to generate a standard curve and cali-
and Dade thrombin combined with Owrens Veronal buffer bration, from which calibration verification was performed
for FIB. In addition, the Dade Behring standard human according to the manufacturer’s recommendations and the
plasma was used for calibration. The following materials laboratory standard procedure for the FIB test established
were used for the daily quality control: Dade Behring in this laboratory. The abnormal fibrinogen control and
Citrol level 1 (for PT, APTT and FIB method), Citrol Citrol level 1 were used for the FIB quality control.
level 3 (for PT and APTT) and abnormal fibrinogen
control (for FIB method). Consumables used included Statistical methods
Sysmex reaction tubes, sample plates, CA Clean I and Mean, standard deviation (SD) and percentage coefficient
II, deionized water and calcium chloride (CaCl2). For the of variation were calculated for each assay. Contingency
tilt tube method, supplies included a 378C water bath, tables were generated using the Statistical Analysis
stopwatch, siliconized glass tubes and disposable pipettes. System (SAS) System (version 8.02) to calculate the rate
For STA, all consumables such as STA cuvettes, STA of discrepant results of the two systems. Correlation and
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Clot detection for routine coagulation testing Bai et al. 571
Table 1 Intraassay precision of CA-1500 and STA in normal and abnormal control products (n U 20)
Normal control Abnormal controls
PT (s) APTT (s) FIB (mg/dl) PT (s) APTT (s) FIB (mg/dl)
CA-1500 STA CA-1500 STA CA-1500 STA CA-1500 STA CA-1500 STA CA-1500 STA
Mean 10.3 9.1 28.2 25.9 264.1 296.1 45.3 45.8 64.7 62.2 104.2 104.8
SD 0.1 0.1 0.1 0.1 12.6 3.3 0.3 0.8 1.2 0.7 9.0 2.4
CV (%) 0.7 0.9 0.5 0.5 4.8 1.1 0.7 1.8 1.9 1.2 8.6 2.3
APTT, activated partial thromboplastin time; CV, coefficient of variation; FIB, Clauss fibrinogen; PT, prothrombin time; SD, standard deviation.
regression analyses were performed on the assay results excellent interassay precision for all three assays well
using Analyze-it (version 1.67). A correlation coefficient within manufacturer’s specifications.
(r) value equal to 1.0 indicates a perfect correlation
whereas values less than 0.50 indicate a poor correlation. Sample summary
Statistical significance was defined by a hypothesis test We randomly collected 2057 clinical samples from a rou-
yielding a P value of less than 0.05. The P value defines tine high-volume clinical laboratory. As shown in Table 3,
a probability for how often the observed difference among the 2057 samples, 2052 (99.8%) had PT results,
would occur by chance variation if there really were 2034 (98.9%) had APTT results and 2023 had FIB results.
no difference. Five hundred and forty-six samples had visually observed
interferences due to hemolysis, lipemia, or icterus or
Results all three, as determined by the ‘sample integrity chart’
Reference ranges and precision (see methods and materials).
Plasma samples from 25 healthy normal donors yielded the
following reference ranges: for PT, 9.5–11.3 s and Correlation between photo-optical and
8.3–10.4 s, respectively, for CA-1500 and STA; for APTT, electromechanical clot detection
25.5–31.9 s and 24.5–31.8 s, respectively, for CA-1500 and Correlation was determined using more than 2000 clinical
STA; and for FIB, 181–370 mg/dl and 216–378 mg/dl, samples over a 45-day time period. All three standard
respectively, for CA-1500 and STA. Intraassay precision coagulation assays were tested. Statistical analysis
for quality control products representing normal (within demonstrated an excellent correlation between the
reference range) and abnormal (within pathological range) photo-optical and electro-mechanical analyzers for PT
samples was performed by repeating the same sample in 20 (r ¼ 0.99), APTT (r ¼ 0.99) and FIB (r ¼ 0.98) (Table 4;
individual tubes during the same run. Intraassay precision Figs 1–3). Because of differences in reagent sensitivity
results are shown in Table 1. For the normal controls, to the vitamin K-dependent clotting factors (factors II,
CA-1500 and STA demonstrated a similar precision for VII and X) and in different reagent/instrument combi-
PT (0.7–0.9% coefficient of variation) and APTT (0.5% nations, the PT has been normalized to an international
coefficient of variation), whereas for FIB, precision for normalized ratio (INR) to standardize the reporting PT
CA-1500 and STA were 4.8 and 1.1% coefficient of vari- results for patients on stable oral anticoagulant therapy
ation, respectively. For the abnormal controls (in the (OAT). A local PT result is then converted to an INR
pathological range), both CA-1500 and STA had precision value, in which INR represents the prothrombin ratio
for PT and APTT ranging between 0.7 and 1.9% coeffi- (patient PT/mean normal PT) raised to the power of an
cient of variation and for FIB, between 2.3 and 8.6% international sensitivity index (ISI). The appropriate ISI
coefficient of variation. Interassay precision was obtained for the particular reagent used in the reagent/instrument
by testing both normal and abnormal controls for 30 con- combination is determined by the responsiveness of the
secutive days. Table 2 shows the interassay precision PT reagent as compared with and graded against a
results. Both normal and abnormal controls demonstrated sensitive WHO international reference preparation
Table 2 Interassay precision of CA-1500 and STA in normal and abnormal control products (n U 30)
Normal control Abnormal controls
PT (s) APTT (s) FIB (mg/dl) PT (s) APTT (s) FIB (mg/dl)
CA-1500 STA CA-1500 STA CA-1500 STA CA-1500 STA CA-1500 STA CA-1500 STA
Mean 10.4 9.3 28.2 31.1 275.6 288.7 45.9 49.5 65.5 75.2 104.0 97.4
SD 0.1 0.2 0.4 0.6 9.5 7.3 2.0 1.5 2.5 1.5 9.0 4.1
CV (%) 1.4 1.8 1.5 1.9 3.5 2.5 4.3 3.0 3.8 2.0 8.7 4.3
APTT, activated partial thromboplastin time; CV, coefficient of variation; FIB, Clauss fibrinogen; PT, prothrombin time; SD, standard deviation.
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
572 Blood Coagulation and Fibrinolysis 2008, Vol 19 No 6
a
Varying sample sizes were due to lack of clot formation in certain assays as well as samples excluded for insufficient sample volume. APTT, activated partial thromboplastin
time; FIB, Clauss fibrinogen; PT, prothrombin time.
(IRP) of thromboplastin [12,13]. After the normalization lation’ or ‘analysis time over’ (CA-1500). The number of
of PT to INR, the two detection methods yielded very samples that showed ‘no clot’ by either method is shown
similar results (with slope of 0.99 and intercept of 0.02 in Table 3. Although a total of 2057 samples were tested,
INR units) and were highly correlated (P < 0.001) some were not included in the data set due to insufficient
(Table 4; Fig. 1). For APTT, CA-1500 yielded results volume for one of the three methods. For PT, among the
similar to STA at values below 40 s but slightly lower 2052 samples tested, one sample yielded a ‘no clot’ error
absolute results when APTT values were above 40 s (due on STA due to the fact that the upper limit of the PT
to a slope of 0.86 and intercept of 4.92 s). Unlike the setting is 100 s on STA, whereas it produced results by
normalized and widely used INR for PT results, there is both CA-1500 (132.4 s) and the tilt tube (120.4 s) method.
no standardized index for APTT, and each hospital has to For APTT, 2034 samples were tested, with five samples
establish its own APTT sensitivity against heparin for its showing ‘no clot’ on STA, whereas they provided results
use in monitoring patients under heparin therapy. For FIB, with CA-1500. Among these five samples, one demon-
the comparison of the two detection methods yielded a strated a regular clot and normal result by tilt tube (24.4 s)
slope of 0.96 and intercept of 16.58 mg/dl. Considering and CA-1500 (24.3 s), whereas for the remaining four
that STA has a slightly higher reference range than samples, the tilt tube method showed longer CTs as
CA-1500, the two methods provided virtually identical compared with CA-1500 (34.1, 213.8, 169.7, 200.4 s for
results over the entire analytical range (40–800 mg/dl). CA-1500 and 59.1, >300, 208.9, 241.4 s for tilt tube,
Further analysis of the subgroup of turbid samples indi- respectively). As shown in Table 3, in no case did
cated that the correlation remained strong between the two STA provide an APTT result whereas CA-1500 yielded
detection methods (Table 5). Although there were several ‘no coagulation’; although in 13 samples, the APTT
samples in each assay that qualified as statistical outliers showed ‘no clot/no coagulation’ on both STA and CA-
(lying outside the 95% prediction interval), they were not 1500, all of which also showed ‘no clot’ by tilt tube. For
deleted from any of the regression analyses because they fibrinogen, there were 2023 samples tested, with 23
were found to have no substantial effect on the results. samples showing ‘no value’ on STA yet yielding results
on CA-1500. Conversely, there were five samples that
Discrepant results between the two methods were defined showed ‘no value’ on CA-1500 but provided results on
as either ‘normal on STA and abnormal on CA-1500’ or STA. Only three samples had ‘no value’ for FIB on both
‘abnormal on STA and normal on CA-1500’. For PT and STA and CA-1500. The tilt tube method is not used for
APTT, ‘abnormal’ was defined as a CT result above the assaying fibrinogen levels.
reference range, and for FIB, ‘abnormal’ was defined as a
value below the reference range. As shown in Table 6, Samples with visually observed interferences
overall, 0.2–2.3% of test results were reported as discre- A total of 546 samples (26.5% overall) with visually
pant between the two methods. observed interferences were identified and tested on
STA, CA-1500 and tilt tube for PT, APTT and FIB.
‘No clot’ samples All three assays showed a good correlation between the
For all three assays examined, any result that showed ‘no comparative and test methods (Table 5 and Figs 4–6).
clot’ was either printed as ‘error’ (STA), or ‘no coagu- No clot was observed for 13 turbid samples on both STA
Table 4 Correlation analysis for all samples performed on CA-1500 and STA
Median (ranges) Reference ranges
CA-1500 STA Correlation (r) Slope (95% CI) Intercept (95% CI) Comparison (P) CA-1500 STA
PT (s) 10.6 (8.0–104.8) 9.8 (6.6–100.0) 0.99a 0.99a 0.02a <0.001 9.5–11.3 8.3–10.4
APTT (s) 29.4 (20.8–208.2) 28.6 (17.7–199.2) 0.99 0.86 4.92 <0.001 25.5–31.9 24.5–31.8
FIB (mg/dl) 344.8 (48.0–901.7) 369.0 (53.0–968.0) 0.98 0.96 16.58 <0.001 181.0–370.0 216.0–378.0
a
Regression based on INR units. APTT, activated partial thromboplastin time; CI, confidence interval; FIB, Clauss fibrinogen; INR, international normalized ratio; PT,
prothrombin time.
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Clot detection for routine coagulation testing Bai et al. 573
Fig. 1 Fig. 3
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
574 Blood Coagulation and Fibrinolysis 2008, Vol 19 No 6
a
Varying sample sizes were due to lack of clot formation in certain assays. APTT,
activated partial thromboplastin time; FIB, Clauss fibrinogen; PT, prothrombin time.
Fig. 4 Fig. 6
Correlation of prothrombin time performed on turbid samples (n ¼ 546). Correlation of Clauss fibrinogen performed on turbid samples
The center line represents the regression line between the STA and (n ¼ 539). The center line represents the regression line between the
CA-1500 assay results, whereas the outer lines represent the STA and CA-1500 assay results, whereas the outer lines represent
boundaries of the 95% prediction interval. Although points outside the the boundaries of the 95% prediction interval. Although points outside
prediction interval can be considered outliers, these points were the prediction interval can be considered outliers, these points were
included in the calculations as their effect on the final results was included in the calculations as their effect on the final results was
negligible due to their locations and the large number of patient negligible due to their locations and the large number of patient
samples. PT, prothrombin time. samples. FIB, Clauss fibrinogen.
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Clot detection for routine coagulation testing Bai et al. 575
that yielded ‘no clot’ on both systems. Of interest are four photo-optical and mechanical detection methods are
nonturbid samples that yielded ‘no clot’ on the STA but statistically equivalent.
gave valid results on both the CA-1500 and tilt tube. We
noted that all four ‘no clot’ samples on STA were forming An advantage of photo-optical clot detection is the ability
tiny weak white clots by the tilt tube method, three of to combine the review of the entire clot formation with
which presented with ‘early reaction error’ along with an analysis by sophisticated software to aid in determining
APTT value on CA-1500. This indicated that the strength the correct coagulation test parameters for patients with
of forming a mechanical clot on STA was so weak that the unusual in-vitro clot reactions. Conversely, the limita-
formation of a firm clot was beyond the upper limit of the tion with any mechanical detection system is that it only
CT setting (236 s for APTT). The fourth ‘no clot’ sample provides the operator a single data point with no means of
was actually repeated on STA and revealed 35.5 s for further review. In the rare instance when a measurable
APTT, whereas it yielded 34.1 s on CA-1500 and 59.1 s end point is not detected, it is important for laboratories
by the tilt tube method. The issue of sample integrity for with either a photo-optical or mechanical system to
coagulation testing continues to be a growing concern in check the sample with an alternative method of clot
clinical laboratory practice due to preanalytical error detection.
caused by the blood collection personnel with varied
backgrounds and experience. Particularly, hemolysis All data obtained during this study indicate that the
during the blood collection process must be identified patient results obtained by the photo-optical detection
and those samples replaced with nonhemolyzed samples system are as reliable and statistically equivalent as those
when possible [19,20]. obtained using the mechanical detection system.
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
576 Blood Coagulation and Fibrinolysis 2008, Vol 19 No 6
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.