| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Technical Briefs |
1 ARUP Laboratories, Salt Lake City, UT 84108,
2
Departments of Medicine and Pathology, University of Utah Health Sciences Center, Salt Lake City, UT 84132
aaddress correspondence to this author at: Division of Hematology, University of Utah Health Sciences Center, 30 North 1900 East, Salt Lake City, UT 84132-2408; fax 801-585-5469, e-mail george.rodgers{at}hsc.utah.edu
Recently, increasing coagulation test volume and tight personnel budgets have increased interest in automated coagulation analyzers (1). The first coagulation instrument, the Fibrometer (Becton Dickinson), used a moving electrode to detect the clot. This brought some degree of standardization to the reading of the clotting endpoint, but this method was labor-intensive. The next phase in instrumentation was semiautomation. These instruments had storage and delivery capabilities for reagents, but manual pipetting was required for the samples. Examples of semiautomated instruments include the Coag-A-Mate X-2 (General Diagnostics) (2) and early models of the MLA (Medical Laboratory Automation). The current generation of coagulation instrumentation is fully automated. Their capabilities include primary tube sampling, automatic rerun and dilution capabilities, and clotting, chromogenic, and immunologic methodologies (3). Examples of this type of instrumentation include the MDA (bioMerieux), STA-R (Diagnostica Stago), AMAX (Sigma Diagnostics), BCS (Dade Behring), and the Sysmex CA-6000 (Dade International) (4).
We summarize our technical evaluation of the STA-R. The evaluation addressed several issues, including ease of operation, methodologies available, reagent and patient sample on-board capabilities, ability to perform automatic dilutions, and validation of performance.
The STA-R has robotic capabilities and uses clotting, chromogenic, and immunologic assays. It is an open system with room for 220 patient tubes and 75 positions for reagents. Approximately 360 prothrombin time (PT) and partial thromboplastin time (PTT) tests can be run per hour. The analyzer makes automatic dilutions, reruns samples, and processes STAT samples without interrupting current testing. The system uses Windows NT software and is interfaced.
Maintenance includes a daily assessment of the condensation trap and wiping of the touch screen. The instrument is left on 24 h a day, so there is no extra start-up time involved. Weekly maintenance requires up to 30 min.
The touch-screen monitor allows easy access to all applications needed for performing the various steps in an assay. These steps include running of daily controls, calibration, and reporting of patient results. Although daily calibration is not required, we did so.
Each assay described below was performed on the STA-R instrument, and results were compared with those of the MDA (bioMerieux) and the BFA (Dade Behring) already set up in the laboratory. The final step in the validation process was to perform correlations among three STA-R instruments.
Reagents for the STA-R and Fibrintimer A were from Stago and bioMerieux, respectively. MDA reagents for factors II, V, VII, and X were from Dade Behring; reagents for protein C, protein S, and antithrombin were from BioPool International, Stago, and bioMerieux, respectively. Table 1 in the Data Supplement (available online at the Clinical Chemistry Online web site at http://www.clinchem.org/content/vol48/issue9/) lists all factor-deficient plasmas and reagents used with each instrument and their sources.
Intrinsic and extrinsic coagulation factors were measured according to manufacturer specifications and standard laboratory methods (5). Clotting times for intrinsic assays were initiated by the instrument making a 1:10 (5 µL in 45 µL) dilution of the plasma with Owrens–Kohler buffer (0.028 mol/L sodium barbiturate, 0.126 mol/L sodium chloride). For that assay, 50 µL of the specific factor-deficient plasma and 50 µL of the aPTT reagent were then added to 50 µL of the plasma dilution, which was then incubated for 240 s. Final measurements were taken after the addition of 50 µL of 0.025 mol/L calcium chloride.
The assays for the extrinsic factors were performed similarly. For that particular assay, 50 µL of factor-deficient plasma was added to 50 µL of the plasma dilution. After an incubation period of 240 s, final measurements were taken after the addition of 100 µL of the PT reagent.
The protein C and S assays were initiated with a 1:10 (5 µL in 45 µL) dilution of the sample plasma with Owrens buffer. For the protein C assay, 50 µL of reagent 1 (protein C-deficient plasma) and 50 µL of reagent 2 (protein C activator) were added to 50 µL of the plasma dilution and incubated for 240 s. After the incubation period, 100 µL of 0.025 mol/L calcium chloride was added, and the final measurement was taken (6). After the initial dilution for protein S, 50 µL each of reagent 1 (protein S-deficient plasma), reagent 2 (protein C activator), and reagent 3 (bovine factor V) (7) were added to 50 µL of the patient dilution. After the incubation period of 240 s, final measurements were taken after the addition of 50 µL of 0.025 mol/L calcium chloride.
Clotting times for the intrinsic factors, extrinsic factors, protein C, and protein S were plotted on a reference curve that was generated by the instrument. The results were then calculated, and the concentration of analyte was determined.
The antithrombin assay uses a chromogenic method and follows similar manufacturer instructions. Briefly, 100 µL of reagent 1 (thrombin) was added to 50 µL of a 1:20 (5 µL in 95 µL) plasma dilution and incubated for 60 s. After the incubation, the final measurement was taken after the addition of 100 µL of reagent 2 (substrate) to the sample (8).
For each assay, we evaluated linearity, precision (intra- and interrun), lowest limit of detection, and reference intervals according to relevant NCCLS guidelines (9)(10)(11)(12)(13)(14).
Clinical specimens that had previously been tested were run concurrently on both systems. We analyzed 35–40 samples for each assay, encompassing the entire reportable range (9). Deming regression analysis was used (15)(16).
For linearity, five samples with increased values for each assay were diluted in Owrens–Kohler buffer. Five dilutions of each sample were assayed, beginning with a 1:2 dilution and continuing through a 1:40 dilution (10). This encompassed all concentrations of samples that would be tested in the remaining studies. We assessed within-run precision (n = 10) with normal and abnormal control materials. To estimate interrun precision, a normal and an abnormal control were run in replicates of five on each of 5 days (11). To evaluate the limit of detection, 10 replicates of a zero concentration analyte (Owrens–Kohler buffer) and a low-value calibrator were assayed (12). The low-value calibrator was prepared by diluting a pooled normal plasma purchased from Precision Biologic.
Blood was obtained by clean venipuncture; if an evacuated tube was used, a pilot tube was drawn first. An exact ratio of 9 volumes of blood to 1 volume of anticoagulant (32 g/L citrate) was maintained. The anticoagulated specimen was then centrifuged at 3000g for 10 min (13). Samples were then aliquoted and kept frozen at -60 to -80 °C. Samples were stored for up to 18 months (17). Forty donor samples were collected in the same manner to estimate reference values (14). The mean age of the donors was 32 years (range, 20–55 years). All healthy donors had no history of bleeding or thrombosis and were taking no medications for at least 2 weeks before specimen collection.
Table 1
summarizes results of the method comparison studies. The correlation (r2) was >0.90 for all assays. All assays were linear on dilution to the lowest concentration tested (data not shown). Linearity was determined by calculating the observed/expected ratio (O/E), as a percentage. The sample dilution was considered linear if the O/E was 90–115%. Estimated reference values are shown in Table 1
.
|
Data Supplement Table 2 (available online at the Clinical Chemistry Online web site at http://www.clinchem.org/content/vol48/issue9/) summarizes the imprecision data. CVs were <10% except for the between-run (total) CVs for the factor XI and XII interrun values for the abnormal control. Protein C, protein S, and antithrombin reagent sets purchased from Diagnostica Stago all exceeded the manufacturer’s precision claims with CVs <4%, except for the protein S intrarun CV on the abnormal control, which was 9%. This was attributable to an outlier in that day’s run.
Lower limits of detection were 1–2% for the intrinsic factors (VIII, IX, XI, and XII), 5–7% for the extrinsic factors (II, V, VII, and X), and 10% for proteins C and S and antithrombin activity.
We assayed 50 samples in duplicate, and the absolute difference was then calculated and evaluated by histogram. All differences were <10%. When 1:20 and 1:40 dilutions of these 50 samples were run in duplicate, we again observed a <10% difference between values. We concluded that duplicate analyses were unnecessary.
The STA-R has allowed our laboratory to convert to primary tube sampling because Diagnostica Stago manufactures a rack sized for our tubes. This saved the technician time required to pour off each sample and avoided purchasing of nonstandardized tubes and labels for each tube. The mechanical clot detection methodology offered an advantage not quantified here: icteric and lipemic samples have no effect on this method (as they sometimes do with photooptical clot detection methods). We now perform 75% of our coagulation menu on the STA-R platform.
We observed marked changes in estimated reference values during evaluation of the STA-R coagulation analyzer. These changes were attributable to different reagents, calibrators, and methodologies. For example, the previously established range for factor IX was 50–150%, but when these samples from healthy donors were run on the STA-R platform, the range was 82–162%. Such differences were also apparent for other assays. Therefore, when using the slope and intercept to estimate correlations, it is necessary to appreciate that on one platform a value of 100% for an analyte may not be equivalent to 100% on another platform. The ranges that we observed for slope and SE values in our studies on the STA-R were similar to those previously reported for another automated coagulation analyzer (4).
The STA-R analyzer has potential disadvantages. Although the instrument is able to run multiple assays at various dilutions simultaneously, as of now it does not have software capabilities for automatic reflex testing. As with other fully automated systems, downtime is also a potential problem.
In summary, for coagulation laboratories performing a large number of esoteric assays, the STA-R coagulation analyzer may be a useful platform to improve laboratory efficiency.
References
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
