Validation of the Sysmex XN‐V hematology analyzer for canine specimens

Abstract Background The Sysmex XN‐V is derived from the new Sysmex XN series of human hematology analyzers. The main changes from the previously validated XT‐2000iV analyzer include an optic‐fluorescent analysis for platelets and nucleated RBC count. Objective We aimed to validate the Sysmex XN‐V for canine blood according to American College for Veterinary Clinical Pathology and International Council for Standardization in Hematology recommendations. Materials and Methods Canine EDTA blood specimens and quality control material were analyzed on the Sysmex XN‐V to evaluate imprecision, bias, linearity, a comparison with the XT‐2000iV analyzer, interference effects, carry‐over, and stability. We also verified previously established Sysmex XT‐2000iV reference intervals (RIs). Results Imprecision and bias were low (<5%) for most variables. Observed total error was lower than allowable total error for most measured variables except lymphocytes and monocytes. Visually determined linearity was excellent for all variables, except for lymphocytes. The correlation between the XN‐V and XT‐2000iV analyzers was high (>0.93) for all variables except MCHC and reticulocyte indices. Correlations between the Sysmex XN‐V and manual differential counts were good for neutrophils and eosinophils, acceptable for lymphocytes, and fair for monocytes. Hemolysis, lipemia, and to a lesser extent icterus, had significant effects on measured hemoglobin concentration and associated variables. Carry‐over was not visually observed for any variable. Changes in the Sysmex XN‐V measurements after storage at 4℃ and 24℃ were similar to those described for the Sysmex XT‐2000iV analyzer. The previously established Sysmex XT‐2000iV RIs can be used to interpret results from the Sysmex XN‐V analyzer for most variables except red blood cell distribution width and mean platelet volume. Conclusions The performance of the Sysmex XN‐V analyzer was excellent and compared favorably with the Sysmex XT‐2000iV analyzer.


| INTRODUC TI ON
Large, automated hematology analyzers with multispecies software are of great importance in veterinary laboratories and industry. The Sysmex XN-V is a new analyzer, derived from the Sysmex XN series for analyzing human blood. The main changes from the XT-2000iV 1,2 include an optic-fluorescent analysis for platelets (PLT-F channel using an oxazine-based fluorescent dye), a nucleated RBC (NRBC) count, and new software for the data analysis of 11 animal species.
Validation of a new instrument is recommended before its implementation. According to the American Society of Veterinary Clinical Pathology (ASVCP), validation includes an evaluation of linearity, repeatability, reproducibility, method comparison, hemolysis, lipemia, and bilirubin interference studies, a recovery study, reference interval (RI) determinations, and a detection limit study. [3][4][5] According to the International Council for Standardization in Hematology (ICSH), evaluation should also include carry-over and specimen stability, but not recovery and detection limit studies. 6 The purpose of this study was to validate the Sysmex XN-V hematology analyzer for canine blood analyses according to ASVCP and ICSH recommendations. We also compared the Sysmex XN-V RIs with those of the XT-2000iV hematology analyzer, which had been previously validated in canine blood samples. 1,2,7,8

| MATERIAL S AND ME THODS
This study was approved by the "Science et Santé animale" Ethics Committee (N° SSA_2019_002).
Fresh blood specimens were received for the hematologic evaluation at the Laboratoire Central de Biologie Médicale (Central Laboratory of Clinical Pathology) to perform patient diagnostic workups or health checkups at the National Veterinary School of Toulouse. As the patients were presented to various disciplines (internal medicine, surgery, emergency, intensive care, oncology department, and routine ambulatory), specimens were obtained from dogs with different diseases or conditions, thus covering a wide range of results and a wide variety of blood disorders. One labeled air-dried blood smear was prepared within 1 hour of blood sampling and stained with a May-Grünwald/Giemsa automatic stainer (Aerospray Hematology Slide Stainer Cytocentrifuge 7150, Wescor), and was then stored before microscopic evaluation. Prior to hematologic analysis, blood specimens were kept at room temperature (24℃), placed on an agitator (Specie mix, Drew Scientific Inc) for 20 minutes, and then gently inverted to ensure homogenization. Specimens with visible or microscopic clots were excluded.

| Variables analyzed
The following variables were analyzed with both analyzers: imped-

| Method validation and comparison study
Repeatability was evaluated from 20 consecutive repeats of a randomly selected canine specimen analyzed within 2 hours of sampling.
Reproducibility was evaluated by duplicate morning and afternoon measurements of the manufacturer's quality control material (QC1 & QC2) for five consecutive days.
Linearity was only tested on analytes expressed as concentrations or counts and not for those expressed as indices or percentages. Linearity was evaluated by analyzing four repeats of a 5-point (0%, 25%, 50%, 75%, 100%) serial dilution of a fresh canine EDTA blood specimen with the Sysmex diluent solution, DCL (Cellpack DCL, Sysmex Europe GmbH). Results of diluted specimens below the lower limit of measurement were considered as 0.
The measurements by the two analyzers were compared by paired analyses of fresh EDTA blood specimens less than 2 hours after blood collection, with a delay of less than 30 minutes between measurements. Blood specimens with platelet clumps visualized on the blood smear or flagged by the analyzer were excluded. Basophils were not included in the comparison because the basophil count has been reported to be unreliable. 9 Each result was classified according to canine RIs previously established for the Sysmex XT-2000iV, 7 and the number of results within, above, or under the reference limits by the two analyzers was counted.
Differences between the XN-V and XT-2000iV scattergrams are presented in Figure 1. The flags and scattergrams were checked and investigated when a clinically relevant discrepancy between the manual and XN-V leukocyte differential count was observed.
The effects of hemolysis, hyperbilirubinemia, and hyperlipemia were assessed using previously published protocols. 10  Carry-over was evaluated using a series of five triplicate analyses with high-level and then low-level quality control materials. 6 Stability was determined using ten randomly selected fresh EDTA specimens stored at room temperature and 4℃. Duplicate analyses were performed immediately and 2, 4, 8, 24, 48, and 72 hours later.
The specimens stored at 4℃ were analyzed after rewarming at room temperature for 20 minutes and were thus analyzed after the room temperature specimens.
The canine hematologic RIs previously established for the Sysmex 7 were verified according to ASVCP and IFCC recommandations 4,11-13 in 20 EDTA blood specimens from healthy dogs selected on the basis of history and physical examination.

| Statistical analysis
Repeatability and reproducibility were evaluated as the coeffi- for WBC differential count, HCT, and NRBC count. 5 Effects of interferences were evaluated by visual inspection of results and repeated measures ANOVA, and when significant, followed by Dunnett's test. Effects of carry-over were examined visually and tested by within-and between-series ANOVA. The effects of storage time and temperature were assessed by comparing results at all durations of storage to initial measurements by ANOVA, and when results were significant, followed by Tukey's honestly significant difference test. Sysmex XT-2000iV RIs were verified according to the Clinical and Laboratory Standards Institute (CLSI) and ASVCP guidelines. [11][12][13] Statistical analyses were performed with Excel (Microsoft), Analyse-It (Analyse-It), and Systat (Systat Software). The level of significance was set at P < 0.05.

| RE SULTS
Given the numerous variables, we have chosen to report the full results for RBC, WBC, PLT, and RET in tables and figures within the manuscript, while detailed results for the other variables are given in the supplementary material.

| Imprecision -Bias -TE obs
Results of repeatability with canine EDTA blood and within-and between-run of quality control material with the Sysmex XN-V analyzer are shown in Tables 1 and S1. Imprecision and bias were <5% for most variables. Imprecision was notable for lymphocyte, monocyte, and eosinophil counts, reticulocyte and PLT indices, and PLT-I and PLT-O counts, and bias was >5% for monocyte counts, PLT counts, and reticulocyte indices. TE obs was <20% for most measured variables, excluding lymphocyte, monocyte, and eosinophil counts, and higher for the PLT and RET indices. TE obs was lower than TE a for all variables except lymphocyte and monocyte counts. The latter QGI were <0.8 and >1.2, respectively (Table S1).

| Linearity
Linearity could not be tested for eosinophils, as the counts were very low. Visually determined linearity was excellent for all variables, except lymphocyte counts (Figure 2; Figure S1), and even when linearity held, a polynomial fit was better. However, maximum differences from linearity were <5.5%, excluding the monocyte count (16.1%).

| Comparison of XN-V results with XT and manual results
This part of the study included 20 healthy dogs and 44 ill dogs.
Hematologic disorders included anemia (15 dogs), polycythemia (7), TA B L E 1 Imprecision, bias, and total error of RBC-I, WBC, PLT-F, and RET measurements with the Sysmex XN-V hematology analyzer using quality control solutions (QC1 and QC2) and a randomly selected canine specimen  (9), toxic change (2), reactive lymphocytes (14), activated monocytes (9), basophilia (6), or mastocytemia (2). To compare the HCT results between analyzers, one specimen was excluded because the manual PCV was unavailable. One specimen was excluded from the comparison between the Sysmex XN-V and the manual differential leukocyte count because the slide was damaged. A comparison between the two analyzers is reported in Table 2, Table   S2, Figure 3 and Figure S2.

| Interferences
Results for the effects of hemolysis, lipemia, and icterus on hematology variables are shown in Figure 6 and Figure S3.

| Carry-over
Carry-over was not visually observed for any variable (Figure 7; Figure S4). However, statistically significant differences between triplicates (P < 0.05) were observed.

| Stability
The changes in Sysmex XN-V measurements after storage at 4℃ and 24℃ are presented in Tables 4 and S4 (Table S3).

TA B L E 2
Comparison of canine blood RBC-I, WBC, PLT-I, and RET measurements (n = 64) using the Sysmex XT-2000iV and Sysmex XN-V hematology analyzers (95% confidence intervals of Spearman's correlation coefficient r and Passing-Bablok equation coefficients between brackets)

Median (min to max)
Spearman's r  F I G U R E 4 WDF (A) and WNR (B) scattergrams from the Sysmex XN-V analyzer for a canine blood specimen with a high percentage of band cells based on the manual differential count. (A) On the WDF scattergram, a well-defined but extended population was observed at the neutrophils position and identified as neutrophils (turquoise dots) and lymphocytes (pink dots), despite a lack of cluster at the normal lymphocytes position. The WBC count was 32.3 × 10 9 cells/L. The neutrophil percentages were 74.9% (XN) and 90.5% (manual; with 68.5% segmented neutrophils and 22% band cells); the lymphocytes percentages were 18.8% (XN) and 4.0% (manual); the monocytes percentages were 6.0% (XN) and 5.5% (manual); and the eosinophils percentages were 0.1% (XN) and 0.0% (manual). A flag was reported for neutrophils and lymphocytes, as well as an error message "WBC Abn Scattergram." In this case, flag and/or error messages were also observed for PLT and RET. (B) WNR scattergram of the same dog. PLT, platelet count; RET, reticulocyte count; WDF, WBC differential; WNR, white cell nucleated [Color figure can be viewed at wileyonlinelibrary.com] F I G U R E 5 WDF (A), WNR (B), and RET-EXT (C) scattergrams from the Sysmex XN-V analyzer for a blood specimen from a dog with acute leukemia. (A) On the WDF scattergram, a large cloud is present at the location of the lymphocytes, neutrophils, and monocytes populations, and an arbitrary separation was performed by the analyzer (white arrow). (B) On the WNR scattergram, the cluster of leukocytes extended to the upper right part of the scattergram. (C) On the RET-EXT scattergram, the reticulocyte clouds were characterized by a gap between mature reticulocytes (pink dots), and a population of high-fluorescence particles (red and turquoise dots) suspected to be leukemic cells and not platelets (turquoise dots) or high-fluorescence RET (red dots). The WBC count was 191.9 × 10 9 cells/L. The neutrophil percentages were 0.0% (XN) and 4.5% (manual); the lymphocytes percentages were 66.1% (XN) and 0.0% (manual); the monocytes percentages were 33.1% (XN) and 2.0% (manual); and the eosinophils percentages were 0.0% (XN and manual). Blastic cells (93.5%) were only revealed by manual counting. A flag was reported for monocytes and lymphocytes results, as well as an error message "WBC Abn Scattergram". In this case, flag and/or error messages were also observed for PLT and RET. PLT, platelet count; RET, reticulocyte count; RET(EXT), reticulocyte extended; WDF, WBC differential; WNR, white cell nucleated [Color figure can be viewed at wileyonlinelibrary.com]

| D ISCUSS I ON
According to the results obtained for the validation process based on the ASVCP and ICSH recommendations, 4,6 the Sysmex XN-V hematology analyzer can be used for canine blood specimens.
ASVCP guidelines help clinical pathologists assess the quality of analyzers. 3 A first step can be to compare TE obs with the TE a proposed by experts for some variables. 15,16 In this study, all TE obs values were lower than TE a values, except for lymphocytes at low concentrations and monocytes within RIs. This was due to lymphocyte count imprecision and monocyte bias, according to the QGI calculation. 15 The WBC differential count showed a higher TE obs than the other hematology measurands in agreement with the recently proposed TE a . 15,16 Nevertheless, as recommended for all instruments, including the Sysmex XN-V, the automated WBC differential of a patient can only be accepted after reviewing the scattergram and a blood smear to check automated findings. 16 Carry-over and linearity were excellent for all variables except lymphocytes. This was expected since similar results were obtained with the Sysmex XT-2000iV analyzer. 7,8 However, in the previous Our study revealed significant effects of hemoglobinemia and lipemia on HGB and associated variables. Furthermore, lipemia and bilirubinemia also had a significant effect on PLT counts. ASVCP recommendations 3,21 for instrument validation include interference studies, but a search in PubMed for interference studies with hematology analyzers indicated that no publications are available. This is quite surprising, as hemolysis and hyperlipemia are known to artifactually increase HGB measurements using spectrophotometry. 22 Publications on erroneous cell counts or HGB measurements are more numerous than those on hemolysis, and it is interesting to note that interfering substances can cause different types of errors depending on the analyzer. For laser-based hematology analyzers, such as the ADVIA 120 or 2120, which measure the hemoglobin concentration within each red blood cell and generate so-called measured MCHC or CHCM, a difference between the CHCM and the calculated MCHC is observed with an artifactual increase of MCHC but not of CHCM. 22,23 It was interesting to note that, with the Sysmex XN-V, no effects of hemoglobin, lipemia, or bilirubin on RET-He or RBC-He (which were also measured using a laser beam) were observed.
The mean RBC variable values after storage for up to 72 hours were significantly different from T 0 . Such results are well-known and secondary to red blood cell swelling, as previously reported in numerous studies. 8,17,[24][25][26][27][28][29] By contrast, stability is not often reported in validation studies, 17 probably because it is not included in the ASVCP recommendations. However, in the ICSH guidelines for evaluating blood cell analyzers in human medicine, specimen stability testing at room temperature and 4℃ after 4, 8, 12, 24, 48, and 72 hours of storage is recommended. 6 Three studies have reported stability of canine blood with the Sysmex XT-2000iV, 8,17,25 yielding similar results to those of the Sysmex XN-V, with a significant increase in the HCT, MCV, and RET, and a decrease in the MCHC, monocytes, and PLT; the reported changes were more intense at 4℃ than at room temperature, except for PLT. 17,25 TA B L E 4 Percent changes in RBC-I, WBC, PLT-F, and RET measurements on the Sysmex XN-V analyzer for 10 canine EDTA blood specimens according to temperature and duration of storage Note: T 0 , range of measurements at T 0 ; P, ANOVA of the effects of temperature and duration of storage; 2-72, effects of duration of storage: mean of percent of T 0 ; in italics, comparison to T 0 by Tukey's HSD test.