Comp Clin Pathol DOI 10.1007/s00580-012-1608-1

ORIGINAL ARTICLE

Evaluation of the IDEXX ProCyte Dx analyzer for dogs and cats compared to the Siemens ADVIA 2120 and manual differential Fee Goldmann & Natali Bauer & Andreas Moritz

Received: 4 May 2012 / Accepted: 28 August 2012 # Springer-Verlag London Limited 2012

Abstract The ProCyte Dx™ was introduced as an in-house hematology analyzer based on focused flow impedance and flow cytometry. It provides a complete hemogram including a five-part leukocyte differential and reticulocyte count. The aim of the study was to evaluate the ProCyte Dx for dogs and cats. EDTA-anticoagulated blood samples from healthy or diseased dogs (n0270) and cats (n0176) were analyzed within 3 to 6 h of sampling. Routine hemogram variables including reticulocytes were compared with reference methods, i.e., the ADVIA 2120, a 200-cell manual differential leukocyte count, and manual reticulocyte counts. Data were analyzed twice (prior to and after dot plot analysis, with the exclusion of samples with invalid separations of cellular populations). Coefficients of variation were <3 % for complete blood cell count and <7 % for differential count, except for eosinophils (cat, 17 %), lymphocytes (cat, 30 %), platelet counts (PLTs; dog, 14 %), and reticulocytes (dog and cat, 16 and 22 %, respectively). Spearman’s rank correlation coefficients (rs) revealed a good to excellent (rs 00.99–0.80) correlation between both analyzers, except for the mean corpuscular hemoglobin concentration (MCHC; rs 00.56–0.44), cat reticulocytes (rs 00.77), and differential count prior to dot plot analysis. Biases were generally close to 0; however, large biases were seen for hemoglobin (HGB), mean corpuscular hemoglobin, MCHC, mean corpuscular volume, PLTs, and differential count prior to dot plot analysis. The majority of variables correlated favorably with the ADVIA 2120. The large biases of HGB and HGB-derived variables were due to the methodology of the ADVIA. Dot plot analysis is an additional tool for quality F. Goldmann : N. Bauer (*) : A. Moritz Department of Veterinary Clinical Sciences, Clinical Pathophysiology and Clinical Pathology, Justus Liebig University Giessen, Frankfurterstr. 126, 35392 Giessen, Germany e-mail: [email protected]

assurance, and a manual differential count is recommended in case of invalid separation of cellular populations. Keywords CBC . In-house hematology analyzer . Point of care . Focused flow impedance . Flow cytometry . Optical fluorescence . Method comparison . Dog . Cat Abbreviations CBC Complete blood cell count FSC Forward scatter EOS Eosinophils HGB Hemoglobin HCT Hematocrit value MCH Mean corpuscular hemoglobin MCV Mean corpuscular volume LYMPH Lymphocytes MCHC Mean corpuscular hemoglobin concentration MONO Monocytes NEUT Neutrophils PCV Packed cellular volume PLT-I Platelets measured with impedance method PLT-O Platelets measured with optical method RBC Red blood cells rs Spearman’s rho SSC Side scatter SFL Side fluorescence WBC White blood cells

Introduction The ProCyte Dx™1 has recently been introduced as an inhouse hematology analyzer providing a complete hemogram within 2 min. It utilizes combined focused flow impedance 1

IDEXX Laboratories, Westbrook, ME, USA.

Comp Clin Pathol

technology and laser flow cytometry with optical fluorescence capabilities. In contrast to most already validated automated in-house hematology systems, the ProCyte Dx performs a five-part leukocyte differential count and a reticulocyte count. Results are displayed in several scattergrams—so-called dot plots—representing the differentiated cell populations. A software was developed for the analysis of blood from dogs, cats, horses, cattle, and ferrets (Anonymous 2008). The ProCyte DX is based on the same technology as the laser-based benchtop analyzer Sysmex XT-2000iV. A recent evaluation of the Sysmex XT-2000iV demonstrated a favorable performance compared with other benchtop analyzers operating with flow cytometry such as the CellDyn35002 (Neuerer 2005), the ADVIA 120 (Mathers et al. 2008), and the ADVIA 21203 (Bauer et al. 2011a, b). The ADVIA 120/2120 analyzers were originally developed for use in human medicine and are currently available for veterinary medicine with software for 21 different animal species. The ADVIA hematology analyzers have been previously validated for use in veterinary laboratories for several animal species such as dog, cat, horse, cattle, pig, goat, and sheep. The studies demonstrated a good correlation between ADVIA 120 measurements and results determined by manual methods (Moritz et al. 2004, 2005; Moritz 2001; Furlanello et al. 2006). Because of the system’s size, price, and quantity of samples run per hour, the ADVIA 2120 is predominantly used in large clinical laboratories. As the ProCyte Dx works similar to the Sysmex XT-2000iV, the hypothesis of this study was that its performance is similar to that of a large laser-based hematology analyzer. Thus, the purpose of this study was to evaluate the performance (precision, linearity, and accuracy) of the IDEXX ProCyte Dx (thereafter ProCyte) and compare the results obtained with this analyzer to the ADVIA 2120 (thereafter ADVIA) and manual counts.

Materials and methods Samples The study was approved by the Ethics Committee for Animal Welfare. Fresh EDTA-anticoagulated blood samples submitted to the laboratory for routine blood analysis were obtained from clinically healthy and diseased dogs (n0270) and cats (n0176) presented due to internal and surgical diseases or blood donation. All samples were analyzed with the Procyte and the ADVIA within 6 h after blood collection. For all specimens, a 200-cell manual differential count, a manual reticulocyte count, and the packed cellular volume (PCV) were determined 2 3

Abbott Laboratories, Abbott Park, IL, USA. Siemens Medical Solution Diagnostics, Eschborn, Germany.

in addition and served as reference method for the respective variables. For all samples, the analysis with the ADVIA was done first, followed by a measurement with the ProCyte. The time difference between sample analysis on the ADVIA and the ProCyte was <3 h. Before analysis, each blood sample was inverted 10 times to ensure proper mixing. Evaluated variables included a complete blood cell count (CBC), i.e., the white blood cell count (WBC), hemoglobin (HGB), red blood cell count (RBC), mean corpuscular volume (MCV), mean corpuscular hemoglobin (MCH), mean corpuscular hemoglobin concentration (MCHC), and platelet count (PLT) as well as the leukocyte differential count. IDEXX ProCyte The ProCyte was run with the software version 00-25_18. It utilizes impedance technology to provide an RBC and PLT count (PLT-I: impedance-based) based on hydrodynamic focusing and direct current detection. For feline PLT counts, optical fluorescence is used (PLT-O: optical fluorescence). Particles that fit within predefined species-specific RBC and PLT size ranges are plotted both on the RBC and PLT histograms and additional dot plots. The hematocrit (HCT) is simultaneously obtained by applying the RBC pulse–height detection method as published previously for the Sysmex XT-1000i (Anonymous 2008). A cyanide-free, spectrophotometric method is applied for the determination of HGB concentration. Erythrocyte indices are calculated as follows: MCV ðHCT
10=RBC Þ, MCH ðHGB
10=RBC Þ, and MCHC
ðHGB
100=HCT Þ. For a second RBC-O (O: optical, i.e., laser-based) and PLT-O count as well as the reticulocyte count, the cells are stained with a fluorescent polymethine dye4 binding nucleic acids. A laser irradiates each cell and side fluorescent light (SFL) and forward scatter (FSC) are captured to differentiate mature erythrocytes, reticulocytes, and platelets based on the degree of size and cellular fluorescence. Visual data reports such as the PLT-O dot plot for platelets and the RETIC dot plot for reticulocytes (Fig. 1) are provided automatically by plotting the cellular fluorescence intensity on the x-axis (SFL) and the cellular volume on the y-axis (FSC). In cats, the PLT-O count was compared with the ADVIA platelet count, whereas in dogs, the PLT-I count was used for comparison as this is the default setting of the platelet counts reported routinely by the analyzer. For the total WBC and the five-part WBC differential, a surfactant permeabilizes the leukocytes, enabling a fluorescent polymethine dye to enter the cell and bind nucleic acids 4

IDEXX ProCyte Dx Stain Pack.

Comp Clin Pathol Fig. 1 Reticulocyte dot plots of both instruments. a ProCyte dot plots, left dot plot representing the RETIC dot plot, SFL plotted on the x-axis and FSC plotted on the y-axis. b RETIC dot plots of the ADVIA plotting oxazine staining intensity on the x-axis and cellular volume on the y-axis. Reticulocytes are the light blue cell population extending from the mature erythrocytes (red)

and cytoplasmic organelles. The 90° lateral laser light scatter and optical fluorescence are used to differentiate leukocytes based on their complexity and their RNA content. Results are presented in the DIFF dot plot by plotting the cells’ complexity on the x-axis (side scatter [SSC]) and their RNA contents on the y-axis (SFL) (Fig. 2). Flags were not available for the first software version evaluated here, so the flagging options were not investigated. Maintenance procedures were performed according to the manufacturer’s recommendation and included a monthly rinse. Quality controls were available in three different levels (high, medium, and low) and were measured daily. Reagent replacement was easy, and adaption of each reagent to necessary amounts for equal numbers of measurements took place during the study. ADVIA 2120 The ADVIA was operated with the software version 5.3.1.MS. The technical details of the ADVIA hematology analyzer have been published previously (Bauer et al. 2011a, b). Briefly, the ADVIA is a laser-based analyzer using laser light scatter, cytochemical peroxidase staining, and WBC lysis to generate CBC. Total hemoglobin is assessed in the ADVIA RBC/PLT channel using a cyanide-free colorimetric method which has been evaluated previously (Bauer and Moritz 2008). Manual methods For all specimens, a 200-cell manual differential count was performed on May Gruenwald Giemsa-stained blood films and manual reticulocyte counts were performed on brilliant cresyl blue-stained blood smears. The reticulocyte count was expressed as percentage of 1,000 non-nucleated RBCs. In cats, only aggregated reticulocytes containing >15 dots of ribosomal RNA were counted as recommended (Tvedten and Moritz 2010). Absolute values were calculated by multiplying the microscopically determined fraction of reticulocytes with the RBC count from the ProCyte. Manual PCVs

were determined by centrifugation using a standard microhematocrit procedure. Manual platelet counts were not performed in this study. Precision Intra-assay precision studies were performed by measuring one EDTA blood sample from each species for 20 consecutive times within 1 h. Coefficients of variation (CVs, in percent) were calculated as standard deviation (SD)/mean×100. Linearity For linearity studies, a 3-mlK3-EDTA-anticoagulated blood sample each from three different dogs and cats was collected and pooled. The pooled blood was centrifuged at room temperature for 10 min at 140 g and plasma was removed and transferred into a tube which was considered as level 1 (0 % pool). For WBC determination, buffy coats were separated into another tube and considered as level 5 (100 % pool). Dilutions of level 5 with level 1 in ratios of 1:3 (level 2), 1:1 (level 3), and 3:1 (level 4) were prepared and each level was measured twice with the ProCyte followed by one analysis of level 5. For cats, a second linearity experiment for RBCs and HGB was performed. After centrifugation, the erythrocyte-rich phase was separated from the blood plasma and considered as level 5. Analysis was performed as described before. Carryover Carryover was evaluated for all cell counts and HGB concentration. For carryover studies, level 5 was analyzed twice with the ProCyte followed by three measurements of phosphate-buffered saline (PBS) which was repeated three times. For each PBS analysis, a separate tube was used. Carryover was calculated as follows: Carryover % ¼ ðx1  x2 Þ=x3
100

Comp Clin Pathol Fig. 2 WBC and WBC differential dot plots of the ProCyte analyzer (a, b) and the ADVIA (c, d). a ProCyte canine DIFF dot plot SSC plotted on the x-axis and SFL plotted on the y-axis. b Feline DIFF dot plot. c Canine dot plot shows the results for WBC counts of the PEROX channel. d Dot plot represents the feline results from the PEROX WBC counts. Cellular populations: pink neutrophils, dark blue lymphocytes, green monocytes, light yellow eosinophils, light blue LUCs, white ghosts. e Canine dot plot of the BASO channel. f Feline dot plot of the BASO channel. Polymorphonuclear cells depicted in pink, mononuclear cell population depicted in blue. Lysis-resistant cells, i.e., human basophils and blasts are displayed in the BASO region (white dots)

where x1 is the arithmetic mean of the 0 % pool (level 1) first measurement, x2 is the arithmetic mean of the 0 % pool (level 1) second measurement, and x3 is the arithmetic mean of the 100 % (level 5) pool. Statistical analysis Results were statistically analyzed using the Analyze-it software5 and GraphPad Prism version 4.0.6 Linearity was assessed with a linear regression analysis between the 5 6

Analyze-it Software Ltd., Leeds, UK. GraphPad Software Inc., La Jolla, CA, USA.

calculated, i.e., anticipated values after dilution values, and measured results. The data of the accuracy study were analyzed twice: First, all data, except results with technical flags indicated by the message “ERROR,” were analyzed. Thereafter, statistical analysis was performed a second time after reviewing (validating) all dot plots of both analyzers by one of the authors (F.G.) and excluding results from either the ProCyte or ADVIA with abnormal dot plots indicative of invalid automated leukocyte differentiation (Fig. 3). The remainder data after the validation process were used for the second statistical analysis. Criteria indicating invalid ProCyte results included dot plots with indistinct separation of the different leukocyte

Comp Clin Pathol

cell populations (especially between neutrophils and lymphocytes characterized by an absent indentation between the cellular populations), an abnormally high “monocyte” count suggestive for the presence of immature cells, or when there were discrepancies between the numerical data and visual data in the dot plot. Criteria indicating ADVIA results included dot plots of the PEROX channel with indistinct gating borders, moderate amounts of large unstained cells (LUCs) indicative of the presence of activated lymphocytes or lymphatic blasts, moderate amounts of lysis-resistant cells in the BASO cytogram indicative of blasts, as well as discrepant results of neutrophils in the peroxidase channel and the number of polymorphonuclear cells in the BASO channel indicative of left shift. Accuracy was determined as a result of the comparison method of the ProCyte and the reference methods (ADVIA, manual differential count, and manual reticulocyte count). RBC count, HCT, PLT, differential count, and reticulocyte count were compared with the results of the ADVIA and with the respective manual methods. Spearman’s rank correlation, Bland–Altman plots depicting the mean bias±1.96 SD, and results of a Passing–Bablok regression analysis were reported. The methods were chosen as they do not require a normal distribution of data and equal variances. Correlations were classified as excellent: Spearman’s rho (rs)00.93–0.99; good: rs 00.80–0.92; fair: rs 00.59–0.79; poor: rs <0.59 (Papasouliotis et al. 2008). Accuracy was considered to be acceptable when the correlation was good to excellent and only a minimal bias was present. Due to the extremely low number of samples with high basophil counts, statistical evaluation of basophil results was not possible. However, the number of false-negative and false-positive automated basophil counts was assessed to detect basophilia, i.e., a basophil count ≥0.1×109/L.

Results Technical errors did not occur throughout the study and the ProCyte analyzer was easy to use. The results of a differential count were always reported by the analyzer even if an indistinct separation of cellular populations was present. Precision The results of the replication of the replicate measurements at the ProCyte are shown in Table 1. CVs for the CBC were generally <3 %, except for PLTs. In contrast, a much higher CV was seen for canine reticulocytes with a value of 22.4 %.The CVs of the leukocyte differential counts ranged from 0.8 % (percentage of canine neutrophils) to 29.8 % (percentage of feline eosinophils).

Linearity Linearity of the ProCyte was excellent (R2 >0.97) for all variables, except for eosinophil counts. Despite the preparation of a buffy coat, only low absolute numbers were achieved for eosinophils. The slope of the regression line was close to 1.0 and the intercept was close to 0, except for platelets (cats > dogs), WBCs, and differential count (dogs) (Table 2). WBC counts over 1,000×109/L exceeded the system’s precision limit, leading to flagged results without numerical data report. Carryover Carryover of the ProCyte was 0 % for feline and canine RBCs, HGB, HCT, PLTs, and reticulocytes. For canine WBCs, carryover was 0 %, whereas in cat WBCs, a carryover of 0.0003 % was seen. Comparison between analyzers CBC For all variables evaluated here, the results of the first and second analysis were similar. Based on the results obtained from the ADVIA, WBC ranged from 0.68 to 123.5×109/L (median, 13.13×109/L) in canine samples and from 2.83 to 63.82 × 109/L (median, 11.34 × 109/L) in feline samples. PCV results ranged from 0.12 to 0.65 L/L (median, 0.40 L/L) in canine samples and from 0.09 to 0.53 L/L (median, 0.34 L/L) in feline samples. Platelet counts ranged from 5 to 1,289×109/L (median, 298×109/L) in dogs and 12 to 1,207×109/L (median, 257×109/L) in cats. Statistical analysis revealed an excellent correlation of the WBC and RBC counts between both analyzers in dogs and cats (Table 3; Figs. 4 and 5) with a minimal bias. The HGB measurements of both analyzers correlated excellently for dogs and cats; however, a marked proportional bias was observed. There was an excellent correlation between the HCT and the spun PCV (rs 00.97) for both dogs and cats. The equation of the Passing–Bablock analysis was y01.02x+0.001 for dogs and y01.08x+0.009 for cats. There was a mean positive bias for HCT compared with the spun PCV (dog, 0.001 L/L; cat, 0.004 L/L). The same was seen for the HCT measurement of the ProCyte in comparison to the HCT result obtained from the ADVIA for samples of both species. For the MCH, a good (dog) to excellent (cat) correlation between the results obtained with both analyzers was seen. However, similar to the HGB measurements, a marked proportional bias was present. In both species, the results of the MCHC correlated poorly and there was a marked negative bias. A good correlation between both analyzers

Comp Clin Pathol

was seen for the MCV; however, there was a strong negative bias for canine specimens. There was an excellent (dogs) to good correlation (cats) between the PLT counts obtained with the ADVIA and the ProCyte. In both species, however, a strong negative bias due to a combined proportional and constant error was observed. Canine reticulocyte counts correlated excellently with the ADVIA results with a minimal bias. In contrast, manual counts correlated fairly (Table 4) with the automated reticulocyte count of the ProCyte. There was a fair correlation of feline reticulocyte counts obtained with both analyzers, but a poor correlation between manual reticulocyte counts and the ProCyte results was seen (Table 4). Small negative constant biases for both the manual and the automated ProCyte reticulocyte counts were observed. WBC differential The validation process, i.e., the exclusion of samples with dot plots indicative of invalid separation of the leukocyte populations, led to a marked increase of the coefficients of correlation and a marked decrease of the biases between the methods in both species (Tables 4 and 5). Canine and feline automated neutrophil counts correlated excellently between both analyzers with a small negative bias. There was a good correlation between the automated ProCyte neutrophil counts and manual results for dogs and cats with a negative bias of approximately −5 %. Automated canine and feline lymphocyte counts correlated excellently (dog) to good (cat) between both analyzers with a small bias. There was also an excellent (dog) to good correlation between the ProCyte lymphocyte counts and manual results with a positive bias ranging between 4 % (dog) and 5 % (cat). A fair correlation with a small bias was present between automated monocyte counts obtained with the ADVIA and the ProCyte analyzers. The automated ProCyte monocyte counts correlated poorly with the manual results for canine and feline specimens and there was a small positive bias. A good correlation with a small bias was observed between the eosinophil count obtained with the ADVIA and the ProCyte as well as between the ProCyte and manual counts in both species. The manual differential revealed basophils of 0 % in all canine specimens. In 2 out of 155 cats, manual basophil counts of >0.1×109/L, i.e., 0.14×109 and 0.29×109/L, respectively, were detected, corresponding to an absolute basophil count of 0.13×109 and 0.06×109/L of the ProCyte. In 37 out of 155 (24 %) cats, basophilia (median basophil count, 0.15×109/L; range, 0.10–0.84× 109/L) was detected with the ProCyte, which was confirmed with the manual differential count in 1 out of 155 (0.6 %) cases. In 9 out of 244 (4 %) dogs, basophilia was seen with the ProCyte (median basophil count, 0.13×109/L; range,

Fig. 3 Examples for dot plot changes (left ProCyte, middle ADVIA PEROX channel, right ADVIA BASO channel) judged as “results to be confirmed by manual differential.” a Dog with severe toxic left shift: For ProCyte, note the indistinct border between neutrophils and lymphocytes as well as the population of cells with high fluorescence activity extending from the neutrophil cluster indicative of immature neutrophils. For ADVIA, an indistinct separation between neutrophils and monocytes is seen in the PEROX channel as well as between mononuclear and polymorphonuclear cells in the BASO channel indicative of a left shift. b Dog with acute lymphatic leukemia and neutrophilia with left shift: For ProCyte, note the indistinct separation between lymphocytes and monocytes and a large cellular population characterized by a higher fluorescence activity than monocytes extending from the monocyte cluster. For ADVIA, a large population of peroxidase-negative cells is extending from the lymphocyte cluster in the LUC area. There is an indistinct separation between neutrophils and monocytes. In the BASO channel, mononuclear and polymorphonuclear cells are not clearly separated indicative of a left shift. Lysisresistant cells are extending from both the mononuclear region and the polymorphonuclear region in the BASO region. c Dog with lymphoma stage V: For ProCyte, note the population of cells with high fluorescence activity extending from the lymphocyte and monocyte cluster. For ADVIA, a population of peroxidase-negative cells is extending from the lymphocytes in the LUC region. In the BASO channel, a population of lysis-resistant cells extends from the region of mononuclear cells in the BASO region and a small “nose-like” cell population (red circle) is also visible indicative of lymphatic blasts. d Cat with toxic left shift: For ProCyte, note the indistinct border between neutrophils and lymphocytes. For ADVIA, there is an indistinct separation between neutrophils and monocytes. Despite the dominance of neutrophils in the PEROX channel, dominatingly mononuclear cells are seen in the BASO channel indicative of hyposegmented peroxidasepositive cells, i.e., a severe left shift. There is a population of lysisresistant cells extending from the region of mononuclear and polymorphonuclear cells indicative of immature myelomonocytic cells. e Cat with eosinophilia: For ProCyte, note the indistinct border between eosinophils and basophils/neutrophils. For ADVIA, note that feline eosinophils are not displayed separately in the PEROX channel due to low peroxidase activity. In the BASO channel, an indistinct separation between mononuclear and polymorphonuclear cells is evident. Based on the measurements of the ProCyte, 2.64 % basophils are detected, corresponding to a basophil count of 0 % of the manual differential count. f Cat with leukemoid reaction: For ProCyte, absence of clearly defined cell populations. For ADVIA, there is an indistinct separation between all cellular populations. The neutrophils are extending in the LUC region indicative of myeloperoxidase deficiency. In the BASO channel, mononuclear and polymorphonuclear cells are not clearly separated. Cellular populations ProCyte: lilac neutrophils, blue lymphocytes, red monocytes, green eosinophils, light blue basophils, brown ghosts. Cellular populations ADVIA PEROX channel: pink neutrophils, dark blue lymphocytes, green monocytes, light yellow eosinophils, light blue LUCs indicative of reactive lymphocytes or lymphatic blasts, white ghosts. Cellular populations ADVIA BASO channel: pink polymorphonuclear cells, blue mononuclear cell population

0.10–0.26×109/L), which, however, was not confirmed with the manual differential count.

Discussion Technically, the ProCyte works similar to the Sysmex XT2000iV, so the results of the current study are discussed with

b

Comp Clin Pathol

Comp Clin Pathol Table 1 The 20-run intra-assay repeatability of the ProCyte in feline and canine specimens

Variable

Unit

Canine sample (CV, %)

Feline sample (mean ± SD)

Feline sample (CV, %)

13.1±0.31

2.4

14.46±0.27

1.9

0.9 2.4 0.8 0.3 2.2 2.3 22.0 22.0 13.8 0.8 2.4 4.9 5.5 4.9 5.8 2.8 3.7

5.24±0.07 7.62±0.07 0.21±0.003 40.08±0.3 1.45±0.01 36.25±0.38 130.85±20.38 1±0.19 123±7 71.21±3.67 10.29±0.48 20.53±3.45 2.97±0.51 7.99±0.47 1.15±0.07 2.8±0.83 0.04±0.01

1.4 0.9 1.4 0.8 0.8 1.0 15.6 16.0 5.5 5.15 4.67 16.81 17.49 5.91 6.82 29.77 28.17

109/L

WBC

HGB hemoglobin, HCT hematocrit value, MCH mean corpuscular hemoglobin, MCV mean corpuscular volume, MCHC mean corpuscular hemoglobin concentration, PLT platelets, RBC red blood cells, WBC white blood cells

Canine sample (mean ± SD)

RBC HGB HCT MCV MCH MCHC RET RET PLT Neutrophils Neutrophils Lymphocytes Lymphocytes Monocytes Monocytes Eosinophils Eosinophils

1012/L g/dL L/L fL fmol mmol/L 109/L % 109/L % 109/L % 109/L % 109/L % 109/L

7.5±0.07 18.62±0.05 0.53±0.04 70.2±0.24 2.5±0.06 35.5±0.80 409.95±90.28102/μL 54.8±12.06 812±113 74.8±0.6 9.9±0.2 58.2±2.8 1.5±0.08 11.2±0.5 0.8±0.04 80.2±2.3 1.1±0.04

previous studies of the ProCyte and the Sysmex XT-2000iV. Generally, as seen in the precision experiment and the linearity measurements, the ProCyte meets the analytical requirements for hematology analyzers (ICHS 1993, 1994; Koepke et al. 1992; Westgard 2003). CVs for reticulocyte counts were higher for canine and feline samples than CVs found for both the Sysmex XT-2000iV (11.8 %; Bauer et al. 2011a) and the ADVIA (15.2 %; Nakagawa 2011). Surprisingly, a very high CV (28 %) for feline eosinophils was seen, which was higher than reported previously for the Sysmex XT-2000iV (4.5 %, Bauer et al. 2011a; 11–24 %, Lilliehook and Tvedten 2009b) and the ADVIA (22 %; Nakagawa 2011). The discrepancies between the studies can be explained with the sometimes low eosinophil counts as in the sample of the current investigation which is well Table 2 Linearity and measuring range of the ProCyte for feline and canine specimens (linear regression analysis)

Species/ variable

Dogs Measuring range

RBC HGB

HGB hemoglobin, RBC red blood cells, WBC white blood cells

Unit

PLT WBC NEUT LYMPH MONO EOS

known to be associated with a high CV (Reed et al. 2002). Compared to CVs observed for the ADVIA (Nakagawa 2011), lower values were obtained with the ProCyte for all cell populations. A limitation of the actual investigation—and many previous studies—is the fact that the CV is ideally calculated also for abnormal specimens preferably close to the medical decision limit of the respective variables. A limitation of the linearity experiment was the fact that no RBC and PLT enrichment was performed for dogs due to the difficulty to obtain larger sample volumes in canine patients. Due to the small sample volumes, the deviation in the WBC differential was considered to be a pipetting imprecision. Carryover values in this study were reported to be <0.01 % and, therefore, within previous recommendations (Bollinger et al. 1987).

Cats R2

Intercept

Slope

Measuring range

R2

Intercept

Slope

1012/L g/dL

0–5.23 0–11.8

0.99 0.99

−0.08 −1.98

1.00 1.02

0–19.42 0–25.8

0.99 0.99

0.01 −0.72

1.06 1.06

109/L 109/L 109/L 109/L 109/L 109/L

0–229.8 0–584.6 0–401.6 0–90.81 0–86.84 0–0.41

0.99 0.98 0.98 0.97 0.99 0.96

−8.66 −40.20 −26.61 −8.70 −3.93 −0.02

0.99 1.00 1.00 1.00 0.98 0.93

0–949.9 0–47.5 0–39.1 0–5.3 0–2.3 0–0.84

0.98 0.99 0.99 0.99 0.98 0.87

−21.21 0.05 −0.28 0.22 0.02 0.01

1.09 0.99 0.99 0.97 0.97 0.80

0.00 (0.00 to 0.00)

0.03 (−0.19 to 0.30)

−1.64 (−9.51 to 6.23)

Bias (95 % limits of agreement)

−5.56 (−8.72 to −4.00)

−5.56 (−8.72 to −4.00)

(0.91 to 0.96) −0.07 (−0.36 to 0.19) −1.06 (−4.09 to 1.98) (1.06 to 1.09) −0.18 (−0.32 to −0.05) 0.41 (−0.36 to 1.18) (1.06 to 1.09) −0.18 (−0.32 to −0.05) 0.41 (−0.36 to 1.18) (0.84 to 0.87) −0.06 (−0.28 to 0.09) −2.06 (−3.75 to −0.36) (0.84 to 0.87) −0.09 (−0.28 to 0.08) −2.01 (−3.43 to −0.60) (1.06 to 1.14) 0.57 (−0.72 to 1.76) 3.88 (−1.04 to 8.81) (1.06 to 1.14) 0.34 (−0.87 to 1.65) 3.95 (−0.40 to 8.29) (0.76 to 0.83) 0.27 (−0.40 to 0.81) 3.36 (−4.51 to −2.21) (0.75 to 0.83) 0.34 (−0.40 to 0.90) −3.37 (−4.52 to −2.21) (0.99 to 1.45) −17.71 (−28.07 to −9.06) −10.02 (−14.05 to −5.99) (0.99 to 1.45) −17.71 (−28.07 to −9.06) −10.02 (−14.05 to −5.99) (1.10 to 1.32) −6.00 (−11.19 to −1.84) 2.80 (−2.11 to 7.71) (1.11 to 1.33) −6.44 (−11.70 to −2.18) 2.82 (−2.12 to 7.76) (0.96 to 0.77) 0.28 (−10.32 to 12.27) −62.1 (−240.9 to 116.7) (0.96 to 0.76) 0.56 (−10.48 to 9.37) −68.5 (−206.8 to 69.7) (0.84 to 1.03) −0.09 (−0.11 to −0.06) 0.09 (−0.57 to 0.38) (0.83 to 1.0) −0.09 (−0.10 to −0.06) −0.11 (−0.48 to 0.26) (0.81 to 1.06) −5.22 (−7.77 to −3.02) −5.22 (−7.77 to −3.02) 169 0.76 0.96 (0.85 to 1.08)

0.93 1.08 1.08 0.85 0.86 1.10 1.10 0.79 0.79 1.20 1.20 1.20 1.21 0.72 0.72 0.95 0.93 0.94 −1.58 (−55.0 to 51.8)

0.97 0.99 0.99 0.98 0.98 0.96 0.97 0.97 0.97 0.44 0.44 0.83 0.82 0.82 0.90 0.77 0.77 0.74

156 176 176 175 168 176 168 171 168 171 171 171 168 175 170 176 171 171

−1.17 (−4.4 to 2.0) 0.22 (−0.2 to 0.7) 0.22 (−0.2 to 0.6) −2.46 (−4.8 to −0.2) −2.46 (−4.8 to −0.2) 2.22 (−1.7 to 6.2) 2.27 (−1.4 to 6.0) −5.12 (−6.6 to −3.7) −5.11 (−6.4 to −3.9) −8.57 (−11.9 to −5.3) −8.54 (−11.5 to −5.6) 1.71 (−3.3 to 6.7) 1.65 (−3.0 to 6.3) −67.9 (−194.3 to 58.6) −66.7 (−182.5 to 49.1) −0.05 (−1.19 to 1.10) −0.09 (−1.13 to 0.95) −1.65 (−55.1 to 51.8)

Intercept (95 % CI)

1 first analysis with all samples, 2second analysis only with specimens with valid dot plots, HGB hemoglobin, HCT hematocrit value, MCH mean corpuscular hemoglobin, MCV mean corpuscular volume, MCHC mean corpuscular hemoglobin concentration, PLT platelets, RBC red blood cells, RET reticulocytes, WBC white blood cells

143 0.94 0.92 (0.87 to 0.96)

RET 2

0.18 (0.00 to 0.32) −0.16 (−0.23 to −0.10) −0.16 (−0.23 to −0.11) −0.09 (−0.21 to 0.04) −0.12 (−0.26 to 0.01) 1.49 (0.65 to 2.21) 1.43 (0.62 to 2.00) 0.85 (−0.20 to 1.93) 0.63 (−0.20 to 1.89) −5.28 (−10.46 to −0.71) −5.29 (−10.48 to −0.75) −13.31 (−20.43 to −7.08) −12.67 (−19.63 to −6.41) −16.57 (−22.94 to −11.08) −16.88 (−23.25 to −11.43) −0.03 (−0.09 to 0.00) −0.08 (−0.13 to −0.02) −4.68 (−7.29 to −1.30)

0.18 (0.00 to 0.32)

174 0.96 0.92 (0.89 to 0.94)

0.92 (0.91 to 0.94) 1.06 (1.05 to 1.08) 1.06 (1.05 to 1.08) 0.85 (0.85 to 0.86) 0.85 (0.85 to 0.86) 1.01 (1.00 to 1.04) 1.02 (1.00 to 1.04) 0.79 (0.75 to 0.83) 0.80 (0.75 to 0.83) 0.93 (0.82 to 1.05) 0.93 (0.82 to 1.05) 1.22 (1.13 to 1.33) 1.21 (1.21 to 1.32) 0.87 (0.85 to 0.89) 0.87 (0.85 to 0.89) 0.92 (0.86 to 1.00) 0.95 (0.89 to 1.02) 0.99 (0.92 to 1.08)

Slope (95 % CI)

−1.17 (−4.4 to 2.0)

0.99 0.99 0.99 0.98 0.99 0.97 0.98 0.89 0.91 0.56 0.56 0.80 0.81 0.93 0.94 0.89 0.93 0.92

270 0.99 0.92 (0.91 to 0.94)

rs

N

Intercept (95 % CI)

Bias (95 % limits of agreement)

Slope (95 % CI)

N

rs

Cats

Dogs

270 261 260 266 261 266 265 267 259 267 260 266 259 269 266 152 142 153

109/L

Unit

WBC 2 RBC 1 1012/L RBC 2 HGB 1 g/dL HGB 2 HCT 1 L/L HCT 2 MCH 1 fmol/L MCH 2 MCHC 1 mmol/L MCHC 2 MCV 1 fL MCV 2 PLT 1 109/L PLT 2 RET 1 % RET 2 RET 1 109/L

WBC 1

Species/ variable

Table 3 Correlation and agreement of the CBC and reticulocyte count obtained from ADVIA and ProCyte for canine and feline samples (slope and intercept derived from Passing–Bablock regression analysis)

Comp Clin Pathol

Comp Clin Pathol Fig. 4 Correlation and agreement between the CBC obtained with the ADVIA and the ProCyte (n0270 dogs). Left-hand side scatter diagrams with the regression line (blue line) and the identity line (x0y, gray line). Right-hand side Bland–Altman diagrams. Here, the differences are plotted against the averages of the results obtained with both analyzers. The mean difference (bias, solid deep blue line) between both methods ± 1.96 SD (dashed light blue lines) are shown. HGB total colorimetric hemoglobin, HCT hematocrit value, PCV packed cellular volume, PLT platelets, PLT-O platelets measured in the optical channel, Reti reticulocytes

Comp Clin Pathol Fig. 5 Correlation and agreement between the CBC obtained with the ADVIA and the ProCyte (n 0176 cats). For the meaning of the symbols and abbreviations, see Fig. 4

%

%

263 253

263 238 263 229 263 225

N

Dogs

0.87 0.90

0.89 0.98 0.80 0.97 0.52 0.61

rs

to to to to to to

0.99) 1.00) 0.95) 0.97) 1.80) 1.58)

1.00 (1.00 to 1.03) 1.00 (1.00 to 1.03)

0.97 (0.95 0.98 (0.96 0.92 (0.89 0.95 (0.92 1.53 (1.33 1.39 (1.23

Slope (95 % CI)

−0.10 (−0.17 to −0.10) −0.10 (−0.18 to −0.10)

1.25 (−0.02 to 2.75) 0.90 (−0.30 to 2.07) 0.89 (0.24 to 1.45) 0.32 (−0.36 to 0.78) −0.59 (−1.86 to 0.27) −0.19 (−1.03 to 0.55)

Intercept (95 % CI)

0.21 (0.02 to 0.41) 0.11 (−2.7 to 2.9)

170 169

170 151 170 148 170 164

−1.99 −0.51 −0.50 −1.05 3.25 1.72 (−2.78 to 1.20) (−5.3 to 4.2) (−1.43 to 0.43) (−5.6 to 3.5) (2.11 to 4.39) (−2.7 to 6.1)

N

Bias (95 % limits of agreement)

Cats

0.84 0.88

0.64 0.93 0.58 0.90 0.72 0.76

rs

to 1.07) to 1.00) to 1.07) to 1.02) to 1.58) to 1.50)

0.84 (0.77 to 0.92) 0.85 (0.78 to 0.92)

1.00 (0.95 0.96 (0.92 1.01 (0.96 0.98 (0.94 1.45 (1.32 1.35 (1.24

Slope (95 % CI)

−0.07 (−0.18 to 0.15) −0.07 (−0.16 to 0.14)

−4.09 (−8.34 to −0.55) −0.37 (−3.00 to 2.73) 2.07 (0.85 to 3.24) 2.60 (1.62 to 3.40) 0.37 (0.11 to 0.60) 0.48 (0.25 to 0.74)

Intercept (95 % CI)

(−39.52 to 23.95) (−12.71 to 6.25) (−21.32 to 34.01) (−6.38 to 11.37) (−4.26 to 7.99) (−0.86 to 3.60) −0.75 (−5.26 to 3.77) −0.63 (−4.06 to 2.80)

−7.79 −3.23 6.35 2.49 1.87 1.37

Bias (95 % limits of agreement)

270 258

270

%

109/L

RET 1 RET 2

RET 1

%

%

LYMPH 1

LYMPH 2 MONO 1

241 238

0.88 0.89

0.53

0.91 0.52

0.85

0.89 0.92

0.83

0.82

0.63 0.73

rs

1.04 (0.97 to 1.10) 1.03 (0.96 to 1.09)

0.78 (0.68 to 0.90)

0.94 (0.89 to 1.00) 0.83 (0.72 to 0.96)

0.94 (0.89 to 1.00)

0.93 (0.89 to 0.97) 0.92 (0.88 to 0.96)

1.37 (1.25 to 1.50)

1.40 (1.29 to 1.54)

1.46 (1.31 to 1.63) 1.40 (1.25 to 2.54)

Slope (95 % CI)

0.07 (0.04 to 0.12) 0.06 (0.04 to 0.12)

1.35 (0.65 to 1.96)

5.21 (4.47 to 5.81) 1.08 (0.25 to 1.73)

5.25 (4.46 to 5.93)

1.10 (−2.02 to 4.31) 2.01 (−1.02 to 4.99)

0.00 (−0.25 to 0.00)

0.31 (−2.8 to 3.4) 0.28 (−2.4 to 3.0)

−0.39 (−6.3 to 5.5)

4.40 (−4.4 to 13.2) 0.23 (−11.3 to 11.7)

171 167

152

145 174

173

173 146

−5.58 (−19.4 to 8.2) −4.77 (−14.4 to 4.9) 4.61 (−8.2 to 17.4)

165

176

176 173

N

1.44 (−5.4 to 8.3)

1.70 (−7.0 to 10.4)

0.47 (−1.8 to 2.8) 0.36 (−1.2 to 1.9)

−0.02 (−0.08 to 0.06) −0.02 (−0.09 to 0.07) 0.00 (−0.30 to 0.00)

Bias (95 % limits of agreement)

Intercept (95 % CI)

Cats

0.83 0.88

0.48

0.86 0.42

0.61

0.64 0.89

0.54

0.48

0.43 0.51

rs

0.82 (0.77 to 0.88) 0.83 (0.78 to 0.90)

0.78 (0.60 to 0.97)

1.04 (0.97 to 1.14) 0.74 (0.58 to 0.97)

1.18 (1.07 to 1.32)

1.08 (0.99 to 1.19) 0.99 (0.93 to 1.05)

0.80 (0.66 to 0.96)

0.64 (0.50 to 0.80)

0.64 (0.50 to 0.80) 0.94 (0.80 to 1.05)

Slope (95 % CI)

−0.69 (−5.38 to 4.00)

0.14 (0.05 to 0.22) 0.10 (0.00 to 0.21)

0.78 (0.16 to 1.30)

4.26 (2.83 to 5.12) 0.83 (0.11 to 1.37)

3.51 (1.30 to 5.00)

−0.63 (−5.48 to 4.22) −0.51 (−3.97 to 2.94)

0.02 (−3.90 to 3.93)

5.83 (−6.17 to 17.83) 0.11 (−7.36 to 7.58)

10.52 (−18.62 to 39.65)

−10.75 (−44.17 to 22.68) −5.26 (−17.64 to 7.12)

−6.79 (−6.79 to 4.77)

0.06 (0.04 to −0.08) −0.15 (−0.50 to 0.09) −11.67 (−20.71 to −5.25) −4.20 (−9.06 to 0.19)

−0.04 (−1.03 to 0.94) −0.09 (−0.64 to 0.47)

Bias (95 % limits of agreement) −0.06 (−0.04 to −0.08) −0.05 (−0.09 to −0.02)

Intercept (95 % CI)

1 first analysis with all samples, 2second analysis only with specimens with valid dot plots, NEUT neutrophils, LYMPH lymphocytes, MONO monocytes, EOS eosinophils, RET reticulocytes, rs Spearman’s rho

230

EOS 1 EOS 2

230 241

MONO 2

%

241 234

%

NEUT 1 NEUT 2

241

266

RET 2

N

Dogs

Unit

Species/variable

Table 5 Correlation and agreement of the manual reticulocyte and manual differential count with the results obtained with the ProCyte for canine and feline samples (slope and intercept derived from Passing–Bablock regression analysis)

1 first analysis with all samples, 2second analysis only with specimens with valid dot plots, NEUT neutrophils, LYMPH lymphocytes, MONO monocytes, EOS eosinophils, rs Spearman’s rho

EOS 1 EOS 2

%

NEUT 1 NEUT 2 LYMPH 1 LYMPH 2 MONO 1 MONO 2

%

Unit

Species/variable

Table 4 Correlation and agreement of the differential count obtained from ADVIA and ProCyte for canine and feline samples (slope and intercept derived from Passing–Bablock regression analysis)

Comp Clin Pathol

Comp Clin Pathol

Though commonly used in veterinary and human method validation studies, it has to be taken into account that a high coefficient of correlation merely reflects a statistical correlation between two methods. As it is not affected by a potential systematic error, a high coefficient of correlation does not necessarily mean that two results are identical (Petrie and Watson 1999). Regardless of these drawbacks, the coefficient of correlation is still used because, in case of rs 00.99 or greater along with a high precision of measurements, it is generally accepted that the regression analysis can be used to estimate the errors between two methods (Westgard 2003). The ProCyte has been compared previously with the CellDyn for canine specimens (Boes 2010). Differences between the ProCyte and the ADVIA in HGB and variables derived from the HGB such as MCH and MCHC are due to the methodology of the ADVIA. So is the cyanide-free method of total HGB measurement applied by the ADVIA analyzer known to be associated with a mean proportional bias of approximately −20 % (Bauer and Moritz 2008). High negative biases with the reference analyzer were observed for the PLT count in both species. In canine samples, this was mainly due to a constant error so that software adaption can be recommended in this case. Although measurements with the ProCyte were performed later than analyses with the ADVIA (with a maximum time difference of 3 h), platelet reduction resulting from sample aging is considered unlikely but cannot completely be ruled out. However, correct platelet count is impossible if platelet aggregates are present, which might explain the comparably low correlation between both analyzers in this study. Although the ProCyte is operating with the same technology as the Sysmex XT-2000iV, the bias between automated and manual reticulocyte counts is not similar for dogs and cats. While a large positive bias between absolute reticulocyte counts obtained with the Sysmex XT2000iV and the manual reticulocyte count have been reported previously for dogs and cats (Bauer et al. 2011b), the bias was close to 0 for the ProCyte. The most likely cause for the different biases of both analyzers compared to the manual reticulocyte count is an adaptation of the threshold for reticulocytes as the methodology of the analyzers is the same. The examples depicted in Fig. 3b, c clearly show that dot plots without a distinct separation of cellular populations must be verified with the microscopic examination of a blood smear. Except for monocytes, a good to excellent correlation and minimal bias between the ProCyte results and the reference methods was seen after excluding samples with dot plots indicative of invalid automated differentiation. Thus, the automated differential count appears to be reliable in these cases. A possible reason for the poor correlation between monocyte counts obtained with both analyzers and the manual method might be the fact that the manual

differentiation of cellular populations present in low numbers such as monocytes is known to be associated with a high CV (Rumke 1960). Moreover, the cellular morphology and size of monocytes are highly variable (Rizzi et al. 2010), so they are not always easy to differentiate from other cellular populations such as toxic band neutrophils. The negative mean bias between ProCyte and ADVIA neutrophil counts seen prior to excluding results with dot plots indicative of invalid leukocyte differentiation was most likely due to the different methodologies. Immature neutrophils only affect the automated neutrophil count of the ADVIA when concurrent myeloperoxidase deficiency is present (Moritz and Becker 2010). In contrast, immature neutrophils containing more RNA than mature neutrophils are displayed and counted in the lymphocyte or even the monocyte region of the ProCyte as it has been also reported for the Sysmex XT-2000iV (Lilliehook and Tvedten 2009a).As obvious in Fig. 3, lymphatic blasts or reactive lymphocytes appear to contain an increased amount of fluorescent RNA/DNA which explains the overlap with the population of monocytes resulting in a false high monocyte count as reported previously for the Sysmex XT-2000iV (Bauer et al. 2011b; Lilliehook and Tvedten 2009b). The correlation between the automated eosinophil count obtained with the ProCyte and the manual differential was comparable to the results reported for the Sysmex XT-2000iV (Bauer et al. 2011b; Lilliehook and Tvedten 2009b) and was higher compared to other laser-based hematology systems such as the ADVIA (Bauer et al. 2011b). Further studies are needed evaluating a higher number of dogs and cats with basophilia to assess the capability of the ProCyte to detect basophils. The current results indicate, however, that false-positive basophil counts are possible especially in cats. For the Sysmex XT-2000iV working with the same technology as the ProCyte, it is known that basophils are not detected in canine specimens; however, feline samples have not been evaluated for this analyzer so far (Lilliehook and Tvedten 2011). The current study shows that the ProCyte dot plots represent an additional tool for quality control. In contrast to the Sysmex XT-2000iV, however, results of a differential count are always released by the analyzer even if the separation between cellular populations was indistinct. Flags indicating an invalid leukocyte differential were not available in the first software version evaluated here. They are, however, part of later software versions where a message “WBC abnormal distribution” is given by the analyzer (Anonymous 2008). The diagnostic utility of the flagging options, however, has not been evaluated so far. Generally, the interpretation of dot plots is considered to be a task for experts, so for less experienced persons, verification of all results with a blood smear is recommended. A

Comp Clin Pathol

time-consuming manual differential count, however, is only needed if discrepancies between automated count and screening of a blood smear are detected. Overall, a good to excellent correlation with the reference method with only minimal biases was observed for the majority of variables. The dot plot analysis provides an additional tool for quality assurance. The automated leukocyte differential is reliable when a distinct separation of cellular populations is present. Thus, the blood smear has to be reviewed only for toxic changes of neutrophils and blood parasites without performing a time-consuming manual differential count. In contrast, a manual differential count is recommended in case of invalid separation of cellular populations. Large biases of HGB and HGB-derived variables with the ADVIA have to be taken into account and were due to the methodology of the ADVIA. Especially for the MCV in dogs, a large negative bias between the ADVIA and ProCyte results is present and the MCHC correlates poorly between the analyzers. As for other analyzers, a correct PLT and monocyte count is difficult to achieve.

Conflict of interest The authors state that there is no conflict of interest for any of the authors of this publication.

References Anonymous (2008) Customers information. Software-info XT-IV series software version 00-09, 1st edn. Bauer N, Moritz A (2008) Evaluation of three methods for measurement of hemoglobin and calculated hemoglobin parameters with the ADVIA 2120 and ADVIA 120 in dogs, cats, and horses. Vet Clin Pathol 37:173–179 Bauer N, Nakagawa J, Dunker C, Failing K, Moritz A (2011a) Evaluation of the automated hematology analyzer Sysmex XT2000iV™ compared to the ADVIA® 2120 for its use in dogs, cats, and horses. Part I—precision, linearity, and accuracy of complete blood cell count. J Vet Diagn Investig 23:1168–1180. doi:10.1177/1040638711425572 Bauer N, Nakagawa J, Dunker C, Failing K, Moritz A (2011b) Evaluation of the automated hematology analyzer Sysmex XT2000iV™ compared to the ADVIA 2120® for its use in dogs, cats, and horses. Part II—accuracy of leukocyte differential and reticulocyte count, impact of anticoagulant and sample aging. J Vet Diagn Investig 24:74–89 Boes K (2010) Validation of the ProCyte Dx Hematology Analyzer for Canine Complete Blood Counts. CPB 69700 Research Seminar Bollinger PB, Drewinko B, Brailas CD, Smeeton NA, Trujillo JM (1987) The technicon H*1—an automated hematology analyzer for today and tomorrow. Complete blood count parameters. Am J Clin Pathol 87:71–78 Furlanello T, Tasca S, Caldin M et al (2006) Artifactual changes in canine blood following storage, detected using the ADVIA 120 hematology analyzer. Vet Clin Pathol 35:42–46 ICHS (1993) Recommendations of the International Council for Standardization in Haematology for ethylenediaminetetraacetic acid anticoagulation of blood for blood cell counting and sizing.

International Council for Standardization in Haematology: expert panel on cytometry. Am J Clin Pathol 100:371 ICHS (1994) Guidelines for the evaluation of blood cell analysers including those used for differential leucocyte and reticulocyte counting and cell marker applications. International Council for Standardization in Haematology: prepared by the ICSH Expert Panel on Cytometry. Clin Lab Haematol 16:157–174 Koepke JA, Bentley SA, Pierre RV et al (1992) Reference leukocyte differential count (proportional) and evaluation of instrumental methods. A reference method of the evaluation of automated differential counters, based on the visual differential count. NCCLS Document H-20-A, vol. 27, no. 4 Lilliehook I, Tvedten H (2009a) Validation of the Sysmex XT2000iV hematology system for dogs, cats, and horses. I. Erythrocytes, platelets, and total leukocyte counts. Vet Clin Pathol 38:163–174 Lilliehook I, Tvedten H (2009b) Validation of the Sysmex XT-2000iV hematology system for dogs, cats, and horses. II. Differential leukocyte counts. Vet Clin Pathol 38:175–182 Lilliehook I, Tvedten H (2011) Errors in basophil enumeration with 3 veterinary hematology systems and observations on occurrence of basophils in dogs. Vet Clin Pathol 40:450–458. doi:10.1111/ j.1939-165X.2011.00353.x Mathers RA, Evans GO, Bleby J, Tornow T (2008) Evaluation of the Sysmex XT-2000iV haematology analyser for rat, dog and mouse whole blood samples. Comp Clin Pathol 17:137–144 Moritz A (2001) Der Einsatz lasergestützer Multiparameter Hämatologiesysteme in der Veterinärmedizin (The use of automated laserbased multiparameter hematology systems in veterinary medicine). 1 ed Büchse der Pandora, Moritz A, Becker M (2010) Automated hematology systems. In: Weiss DJ, Wardrop KJ (eds) Schalm's veterinary hematology. WileyBlackwell, Ames Moritz A, Fickenscher Y, Meyer K et al (2004) Canine and feline hematology reference values for the ADVIA 120 hematology system. Vet Clin Pathol 133:32–38 Moritz A, Walcheck BK, Weiss DJ (2005) Evaluation of flow cytometric and automated methods for detection of activated platelets in dogs with inflammatory disease. Am J Vet Res 66:325–329 Nakagawa J (2011) Evaluation der Hämatologiesysteme Sysmex pocH-100iV Diff und XT-2000iV für die Tierart Katze [Evaluation of the hematology analyzer Sysmex pocH-100iV Diff and XT-2000iV for cats]. Thesis work, Thesis Giessen Neuerer FF (2005) Evaluation des vollautomatischen Hämatologiegerätes CELL-DYN 3500 im klinischen Einsatz bei Hund und Katze [Evaluation of the automated hematology analyzer CELL-DYN 3500 for the clinical use in dog and cat]. Thesis work Munich Papasouliotis K, Cue S, Crawford E, Pinches M, Dumont M, Burley K (2008) Comparison of white blood cell differential percentages determined by the in-house LaserCyte hematology analyzer and a manual method. Vet Clin Pathol 35:295–302 Petrie A, Watson P (1999) Statistics for veterinary and animal science. Blackwell, Oxford Reed GF, Lynn F, Meade BD (2002) Use of coefficient of variation in assessing variability of quantitative assays. Clin Diagn Lab Immunol 9:1235–1239 Rizzi TE, Meinkoth JH, Clinkenbeard KD (2010) Normal hematology of the dog. In: Weiss DJ, Wardrop KJ (eds) Schalm's veterinary hematology. Blackwell, Ames, pp 799–810 Rumke CL (1960) Laboratory aids. Variability of results in differential cell counts on blood smears. Triangle 4:154–158 Tvedten H, Moritz A (2010) Reticulocyte and Heinz body staining and enumeration. In: Weiss DJ, Wardrop KJ (eds) Schalm's veterinary hematology. Wiley-Blackwell, Ames Westgard JO (2003) Basic method validation. Westgard Quality Corporation, Milwaukee