A performance evaluation of sthemO 301 coagulation analyzer and associated reagents

Abstract Aim The study objective was to evaluate the performance of sthemO 301 system and to compare it with the analyzer used in our university hospital laboratory (STA R Max® 2), for a selection of hemostasis parameters. Methods Method comparison (according to CLSI EP09‐A3), carryover (according to CLSI H57‐A), APTT sensitivity to heparin (according to CLSI H47‐A2), HIL level assessment, and productivity were performed using leftover samples from our laboratory (n > 1000). Commercial quality control materials were used to evaluate precision (according to CLSI EP15‐A3) and accuracy. The assays tested on sthemO 301 were: PT, APTT (silica and kaolin activators), fibrinogen (Fib), thrombin time (TT), chromogenic and clotting protein C (PC) activity, and von Willebrand factor antigen (VWF:Ag) levels. Results All intra‐assay and inter‐assay precision CVs were below the maximal precision limit proposed by the French Group for Hemostasis and Thrombosis (GFHT). Accuracy was verified with bias below GFHT criteria and most Z‐scores were between −2 and +2. No clinically relevant carryover was detected. Silica APTT reagent sensitivity to unfractionated heparin was moderate, as expected. Productivity results were consistent over the 10 repeats performed. The overall agreement between the two systems was excellent for all assays, with Spearman rank correlation coefficient all above 0.9 and slopes of Passing–Bablok correlation near 1 and intercepts close to 0. Conclusion For the methods tested, sthemO 301 system met all the criteria to implement a novel coagulation analyzer in the laboratory and result comparability with STA R Max® 2 was good.

ity to unfractionated heparin was moderate, as expected. Productivity results were consistent over the 10 repeats performed. The overall agreement between the two systems was excellent for all assays, with Spearman rank correlation coefficient all above 0.9 and slopes of Passing-Bablok correlation near 1 and intercepts close to 0.
Conclusion: For the methods tested, sthemO 301 system met all the criteria to implement a novel coagulation analyzer in the laboratory and result comparability with STA R Max® 2 was good.

K E Y W O R D S
analytical performance, analyzer, coagulation, laboratory automation

| INTRODUC TI ON
Coagulation laboratories provide a comprehensive set of clinical tests, with a wide variety of routine and specialized coagulation assays, and expert interpretations for the diagnosis and monitoring of bleeding and thrombotic disorders in hospitalized patients and outpatients. The assays range from routine panels with prothrombin time (PT), activated partial prothrombin time (APTT), and fibrinogen (Fib) to more specific protein activity or antigen measurements.
In order to provide reliable results in an efficient and timely manner, laboratories are often equipped with one or several automated analyzers to perform the assays. With evolving technologies, new instruments are regularly placed on the market, with the aim to improve assay performance and overall efficiency of the coagulation workup in the clinical laboratory.
The sthemO 301 analyzer (Diagnostica Stago, France) belongs to this new generation of fully automated hemostasis analyzers.
It is designed for the qualitative and/or quantitative in vitro determination of coagulation parameters by clotting (chronometric), chromogenic, or immunological methods in citrated plasma samples. The dedicated reagents and test procedures developed by the manufacturer for use on this new analyzer enable, among others, improved traceability (with unique reagent vial ID) and optimized performance through adapted pipetting and mixing sequences.
In order to increase the system throughput while decreasing the technician's hands-on time, the overall instrument workflow management for samples, disposables, reagents loading and unloading and for the measurement pace was improved, with an unique measuring cell and higher reagent vial capacity. The aim of the present study was to evaluate and publish for the first time, the performance of sthemO 301 system and to compare it with the analyzer used in our university hospital laboratory (STA R Max® 2), for a selection of hemostasis parameters.

| Study design
This study was performed in the hematology laboratory of the University Hospital of Rennes, France. All citrated plasma samples tested (from blood collected into 0.109 mol/L sodium citrate tubes), anonymized for the study, were leftovers from patients' samples received in the laboratory between January and March 2022 after completion of all testing, as waived by the Ethics Committee of our institution. When appropriate, samples were selected according to their results obtained in routine as described hereafter, and after selection all samples were anonymized, with no option to get back to patients' identities. Samples were analyzed on the STA R Max®  for less than a week. Frozen plasmas were thawed at 37°C for 5 min just before testing.

| Accuracy
Twenty-three external quality assessment samples from our regular EQA program (Qualiris QC Premium S1-2021, Qualiris QC Premium S2-2021, Diagnostica Stago) were tested on sthemO 301 and compared with the mean values of our regular peer group using bias and the Z-score metrics. Acceptable bias was set according to the French Group on Hemostasis and Thrombosis (GFHT) recommendations. 1

| Intra-assay and inter-assay precision
Intra-assay precision was assessed according to CLSI EP15 protocol. 3 In short, two different quality control plasma samples were tested five times per run on five different runs (5 non-consecutive days) to obtain 25 replicates (n = 25) on sthemO 301.
Inter-assay precision was tested through the repeated measurement of internal QC (IQC) samples, being measured twice a day on 21 different days.
Precision results were compared to the acceptable CV determined by GFHT. 1

| Carryover
Inter-sample carryover was assessed according to CLSI H57-A pro- To assess carryover from normal samples to pathological samples, the test sequence was A1-A2-A3-B1-B2-B3, and for carryover from pathological samples to the normal samples, the test sequence was B1-B2-B3-A1-A2-A3. All test sequences were repeated 10 times.
Inter-reagent carryover was tested using five aliquots of a normal pool of plasma. Each aliquot was measured with sthemO PTT A, sthemO Fib, and sthemO PTT A again, and this test series on the five samples was replicated five times.

| Method comparison studies
Method comparison studies were performed with samples covering, as much as possible, the whole measuring range of each assay with the aim of testing around 100 samples for each parameter as recommended by CLSI EP09-A3. 5 Samples were selected for this study according to the initial result of their coagulation tests, and were then tested again in parallel on sthemO 301 and STA R Max® 2 to avoid any bias in results due to delay in testing. Method comparison was run on the following assays: PT, APTT (both silica and kaolin activators), TT, Fib, PC (chromogenic and clotting activities), and VWF:Ag.

| Sensitivity of APTT assays to unfractionated heparin
To estimate the sensitivity to heparin of the two sthemO APTT reagents available, 20 samples from patients receiving unfraction- Finally, as per CLSI H47-A2 6 recommendations, results were plotted on a graph APTT = f(anti-Xa), to assess the sthemO APTT reagents overall sensitivity to heparin.

| HIL
Hemolysis, icterus, and lipemia/turbidity (HIL) indices are often measured on serum and plasma to validate sample quality, especially since sample's color may influence coagulation test results, though for clotting assays, mechanical clot detection is known to be more robust than optical detection. 7

| Productivity
We assessed laboratory productivity with sthemO 301 instrument using 50 plasma samples from our routine activity tested for PT, APTT, and Fib. We measured the duration of sample loading, reagent loading, the time needed by the instrument to give the first and the whole set of results, and the time to unload the samples. This experiment was repeated on 10 independent occasions and the average throughput was then determined. According to CLSI H57-A, inter-sample carryover was assessed using a two-tailed Student's t test to compare the mean value of sample B1 with that of sample B3 (for normal to pathological carryover), and the mean value of sample A1 with that of sample A3 (for pathological to normal carryover). Similarly, inter-reagent carryover was tested comparing the mean APTT values obtained in the first rank of testing, to the mean APTT values obtained after fibrinogen measurement, using a two-tailed Student's t test.

| Statistical analysis
In both cases, a p-value below 0.05 was considered statistically significant.
As proposed by CLSI EP09-A3, to test method agreement and consistency, we used Passing-Bablok regression and a Bland-Altman plots. The Spearman rank correlation coefficient was also used to further evaluate the strength and direction of the correlation. Values of the mean bias on the Bland-Altman graphs were compared to the limits proposed by GFHT for the corresponding parameters. 1

| Accuracy
We were able to evaluate accuracy on a total of 23 EQA samples from previous campaigns. We compared the results to our peergroup reports. Table 1 summarizes the data from these experiments.
For each EQA sample tested and each test methodology, the observed bias to peers was below the acceptable bias proposed by GFHT. 1 Of note, for one sample the Z-score indicated the need for a corrective action for several tests 2 : APTT (in seconds and ratio), Fib, and VWF:Ag. However, as other samples were tested during the same run with the same reagents and with Z-score values between −2 and +2, we concluded that the observed discrepancy result was likely due to reconstitution error.

| Intra-assay and inter-assay precision
Intra-assay and inter-assay precisions results are shown in Table 2.
Inter-assay CVs ranged from 1.5% to 5.5%, which is always lower than the GFHT acceptable limits. 1

| Carryover
No statistically significant inter-sample carryover was observed (all p-values > 0.05) except: • From normal to abnormal sample on PT assay in seconds only (pvalue = 0.02), the abnormal PT moving from 34.5 to 34.8 s.
• From abnormal to normal sample on APTT assay with sthemO CK Prest (p-value = 0.01), the normal APTT moving from 27.5 to 27.8 s.
When considering the actual differences, the impact is a difference of ±0.3 seconds for both cases.
The statistical relevance of such a low difference is consistent with assay precision results as the observed average difference is higher than the standard deviation of the assay.
No statistically significant inter-reagent carryover was observed either.

| Method comparison between STA R® Max 2 and sthemO 301 analyzers
Sample size, minimal and maximal observed values, and Passing-Bablok regression results for all assays are presented in Table 3. All Spearman rank correlation coefficients are above 0.95 (except for TT) and we notice that regression slopes are close to 1 and intercepts close to 0. Bland-Altman plots enable to see that average biases observed in these method comparison studies are all below the maximal acceptable bias proposed by GFHT. 1 The method comparison on TT shows a tendency to shorter clotting times for TT on sthemO 301 analyzer with sthemO Thrombin, in comparison to STA R Max®2 with STA-Thrombin®, and a weaker correlation coefficient (0.919). However, only 44 samples could be used for this analysis, as many samples selected for the study eventually had clotting times longer than the upper limit of the assay (>60 s on STA R Max® 2; >80 s on sthemO 301).
Besides, for INR, there were some differences mostly in the highest INR values, however, those differences were all below 15%.
As graphical examples, the method comparison for PT (in seconds), APTT (in seconds), PC chromogenic, and fibrinogen assays are presented in Figures 1 and 2.

| Sensitivity of APTT assay to heparin
The correlation between sthemO PTT A values and anti-Xa levels measured on STA R Max® was low as shown in the scatter plot As all samples but one had an APTT ratio >1.5 it was not possible to determine the exact sensitivity to UFH of sthemO PTT A.

| HIL
The

| Productivity
The sample sets used for the different runs of throughput, consisted on average in each run of 49% of samples within the normal reference ranges for PT, APTT, and Fib and 51% of samples falling outside the normal reference ranges for the same assays.

| DISCUSS ION
In this study, we aimed at assessing and publishing for the first time Overall, the performance and agreement between the two analyzers were excellent, meeting the validation criteria, and our results enable us to better anticipate the performance that can be achieved with sthemO 301 system in a hemostasis laboratory.

F I G U R E 3 Sensitivity of sthemO PTT
A to UFH: correlation between APTT clotting time ratios and UFH anti-Xa levels.
reagents and parameters, were slightly shorter on sthemO 301 analyzer. The two thrombin time reagents of the two systems contain the same thrombin amount but could be slightly different by other aspects not disclosed by the manufacturer. This constant bias on TT with sthemO 301 could be addressed using appropriately and locally defined reference ranges. There were also few differences in the highest INR values, all below 15% and therefore without any clinical relevance. 9 It is important to note that the reagents used for PT on the two analyzers were different.
Unfortunately, sthemO anti-Xa reagent was still on development during the study, so no anti-Xa reagent could be tested on this analyzer and the relationship between sthemO PTT A values and anti-Xa levels was analyzed only using STA-Liquid anti-Xa® values on STA R Max® 2. This is a limitation to the evaluation of sthemO PTT A sensitivity to heparin, that would deserve further investigation when anti-Xa measurement will be available on sthemO analyzer. However, this approach reflects the reality of a number of laboratories where anti-Xa levels are not accessible or done on another coagulation platform than APTT measurement.
The relationship between sthemO PTT A clotting times and anti-Xa levels was low and this was expected due to the low specificity of APTT, especially in critical ill patients. 10 We were able to calculate the sthemO PTT A clotting time values corresponding to anti-Xa activities of 0.30 and 0.70 IU/mL. Even if this range needs to be validated on a larger number of samples, it shows the sensitivity of this silica-based APTT to UFH, a sensitivity which is known to be highly variable among reagents. 11 Detection of HIL was comparable between analyzers. Evaluation of productivity was limited because, due to organizational constraints in our laboratory, we could not compare the sthemO 301 throughput to that of our STA R Max® 2. Nonetheless, we observed very consistent throughput on the sthemO 301 analyzer over the 10 different runs we performed.
One last limitation of our evaluation is the lack of access to normal donor samples, to establish or at least verify locally reference intervals for the different assays. This would have to be done if the sthemO 301 system was to be implemented for routine use in any coagulation clinical laboratory.
In conclusion, for the methods tested, we found a good comparability between the new sthemO 301 analyzer and the STA R Max® 2, using their respective dedicated reagents. Overall, the results we obtained with sthemO 301 meet the expectations one can have while implementing a novel coagulation analyzer in the laboratory.

CO N FLI C T O F I NTER E S T S TATEM ENT
AC is full-time employee of Diagnostica Stago. The study was financially supported by Diagnostica Stago.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings for this study are available from the corresponding author upon reasonable request.