Performance evaluation of leukocyte differential on the hematology analyzer Celltac G compared with two hematology analyzers, reference flow cytometry method, and two manual methods

Abstract Background The automated hematology analyzer Celltac G (Nihon Kohden) was designed to improve leukocyte differential performance. Comparison with analyzers using different leukocyte detection principles and differential leukocyte count on wedge film (Wedge‐Diff) shows its clinical utility, and comparison with immunophenotypic leukocyte differential reference method (FCM‐Ref) shows its accuracy performance. Methods For method comparison, 598 clinical samples and 46 healthy volunteer samples were selected. The two comparative hematology analyzers (CAAs) used were XN‐9000 (Sysmex) and CELL‐DYN Sapphire (Abbott). The FCM‐Ref provided by the Japanese Society for Laboratory Hematology was selected, and a flow cytometer Navios (Beckman‐Coulter) was used. In manual differential, two kinds of automated slide makers were used: SP‐10 (Sysmex) for wedge technique and SPINNER‐2000 (Lion‐Power) for spinner technique. The spinner technique avoids the issue of Wedge‐Diff smudge cells by removing the risk of breaking cells and non‐uniformity of blood cell distribution on films (Spinner‐Diff). Results The Celltac G showed sufficient comparability (r = 0.67–1.00) with the CAAs for each leukocyte differential counting value at 0.00–40.87(109/L), and sufficient comparability (r = 0.73–0.97) with FCM‐Ref for each leukocyte differential percentage at 0.4–78.5. The identification ratio of the FCM‐Ref in CD45‐positive cells was 99.7% (99.4% to 99.8%). Differences were found between FCM‐Ref/Celltac G/XN‐9000/Spinner‐Diff and Wedge‐Diff for monocytes and neutrophils. The appearance ratio of smudge cells on wedge and spinner film was 12.5% and 0.5%. Conclusion The Celltac G hematology analyzer's leukocyte differential showed adequate accuracy compared with the CAAs, FCM‐Ref, and two manual methods and was considered suitable for clinical use.


| INTRODUC TI ON
Different hematology analyzer models use various principles to measure the complete blood count (CBC) and leukocyte differential for routine tests in clinical laboratories. The model-to-model measurement dispersion is a known issue in external quality control surveys using fresh blood samples. 1 Consequently, the accuracy performance of a hematology analyzer is evaluated using the manual differential leukocyte (Manual-Diff) on blood wedge film (Wedge-Diff) as the traditional reference method. 2 However, this method suffers from several disadvantages, including statistical error, slide distribution error, and morphological interpretation error. 3 The Wedge-Diff is influenced by non-uniform distribution, especially of large nucleated cells, on the blood film. 2 Therefore, these errors should be minimized when evaluating accuracy performance. Elevated numbers of smudge cells tend to be present in the wedge film, especially in case such as chronic lymphocytic leukemia. 4 The addition of bovine serum albumin (BSA) to blood samples effectively reduces the risk of erroneously generating smudge cells, and it keeps the chromatin structure on wedge film. 5 An even more effective method to reduce the number of smudge cells on the film is the spinner film, and few smudge cells are found when performing the Manual-Diff on spinner film (Spinner-Diff). Hence, the Spinner-Diff has the potential to improve both the slide distribution error and the morphological interpretation error. 6 To improve the statistical error, current guidelines 2,7 recommend using an immunophenotypic leukocyte

| Blood samples
All samples were collected in tubes containing K2-EDTA. 9 The blood collection tubes, 10 blood collection procedure, 11 and mixing procedure 12 were according to the methods described by ICSH and CLSI. to select healthy volunteer donors. 13 The following occurrences were excluded from sample selection: Failure to adhere to the studyspecific procedure; Instrument, operator-related, or sample-related failure; and a data-invalidating flag as described in the operating instructions for each instrument. 2

| Hematology analyzers
The Celltac G equipped with software version 01-12 was used as the test automated analyzer (TAA). The Celltac G measures leukocyte differential using novel swirling sheath flow control technology, DynaHelix flow technology TM , and the sample leukocytes largely maintain their morphological characteristics with its novel process for lysing. The DynaScatter laser technology TM classifies by threedimensional scattergram using optimized scatter light collection angles, which has shown improvements in the measured cell volume accuracy and cell identification. 15 The XN-9000 (Sysmex Corporation) equipped with software version 18.0 was used as a comparative automated analyzer (CAA). The CELL-DYN Sapphire (Abbott Diagnostics) equipped with software version 4.1 was also used as a CAA.

| Comparability with the hematology analyzers
For method comparison between the TAA and the two CAAs, test data were measured using 388 samples. Single measurements were used as the test values for routine tests with the CAAs, and the means of the duplicate measurements were used for confirming the reproducibility by the TAA.

| Comparability with Wedge-Diff in negative samples
For method comparison between Manual-Diff using wedge blood smear and the three analyzers (TAA and CAAs), test data were measured using 210 samples, and 14 samples with positive findings 19 on film were excluded.

| Accuracy performance in leukocyte normal samples
To clarify the accurate bias differences in normal samples within 1%, 46 normal samples from healthy volunteers were used. Two

| Comparability
The results of the comparison between the TAA and the CAAs are shown in Table 1. The results compared with Wedge-Diff in the TAA and the CAAs are shown in Table 2.

| Accuracy performance in normal samples
The identification ratio of all identified five-part leukocyte differential in CD45-positive cells was 99.7% (99.4% to 99.8%). Table 3 presents pared with the JSLH-Diff were less than 1%, and the accuracy performance was validated in the TAA for leukocyte differential.
In contrast, the bias from the JSLH-Diff calculated by the mean residual of all samples, which exceeded 1%, was demonstrated in three cases: +2.5% for %NE and −2.0% for %MO in Wedge-Diff, and −1.1% for %LY in XN (CAA). The Celltac G also includes research parameters, including immature granulocytes, bands, and segment cells, in the differential count. However, this was beyond TA B L E 2 Comparability with manual leukocyte differential method on wedge film in negative samples and the three analyzers (TAA and CAAs)  The effect of non-uniformity in cell distribution in the blood film in Wedge-Diff blood film is thought to explain the results obtained in this study for this method (+2.5% for %NE and −2.0% for %MO). The CLSI standard also reported that %MO was 10-20% lower than with the FCM method, including hematology analyzers due to the issue of non-uniformity. 2 A tendency was also observed in this study.
Additionally, the wide bias observed for %NE was attributed to the small bias for %MO causing wide bias for other cell percentages.
The appearance rate of identified smudge cells of neutrophils and lymphocytes was 4.1% and 3.4% in Wedge-Diff. These traumatic injuries can puzzle morphological evaluation; in addition, unskilled operators can be misled. 23 The percentage in Manual-Diff is calculated from identified cells without counting smudge cells, resulting in a leukocyte differential of 100%. These issues should be considered if affected by greater than 1% bias and error. 1 The leukocyte differential in the hematology analyzers (Celltac G and XN-9000) and Spinner-Diff showed consistency compared with JSLH-Diff. In contrast, inconsistency was observed in Wedge-Diff for %MO and %NE. The reason is presumed that the Spinner-Diff was not

| CON CLUS ION
The Celltac G hematology analyzer's leukocyte differential showed adequate accuracy compared with two comparative hematology analyzers, reference flow cytometry method, and two manual methods and was considered suitable for clinical use.

ACK N OWLED G EM ENTS
We would like to thank all patients and healthy volunteers who par- Kohden corporation for support of the overview and standard operation procedure of published Guidelines.

CO N FLI C T S O F I NTE R E S T
This work was supported by a grant from Nihon Kohden Corporation.
Yutaka Nagai is a current employee of Nihon Kohden Corporation.

TA B L E 3 Accuracy performance in leukocyte normal samples
Note: Two hematology analyzers (TAA: Celltac G and CAA: XN-9000) and two manual leukocyte differential methods on wedge film and spinner film compared with the reference flow cytometry method provided by the Japanese Society for Laboratory Hematology as an immunophenotypic leukocyte differential reference method (Reference). The identification ratio was calculated using the following formula: identification ratio =number of identified normal five-part leukocyte differential events (5diff) / number of CD45-positive cell events (CD45+).

AUTH O R CO NTR I B UTI O N S
KY, TY, and TK designed the study, analyzed the data, and performed this study. KY wrote the paper. YN researched for all review and writing the paper. All authors declare that they had full access to the data to revise and approve the manuscript.

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