Evaluation of the Abbott Cell-DYN 3500 Hematology Analyzer in a University Hospital

Transcription

1 HEMATOPATHOLOGY Evaluation of the Abbott Cell-DYN 3500 Hematology Analyzer in a University Hospital JOAN-LLUIS VIVES-CORRONS, MD, ISABEL BESSON, PHD, JOSEP MARIA JOU, MD, AND GABRIELA GUTIERREZ, MB The Abbott" Cell-Dyn 3500 (Abbott CD 3500, Abbott Diagnostics Division, Mountain View, CA) is a fully automated hematology analyzer capable of providing a complete blood count (CBC) profile, including a five-part differential leukocyte count (DLC) and flagging system in this study. The CBC profile and DLC flagging system of the Abbott CD 3500 were evaluated according to the HA-20 protocol of the National Committee for Clinical and Laboratory Standards (NCCLS) and compared to the Technicon H*2 blood analyzer currently used in the authors' laboratory. Linearity, carryover, precision, and stability were all within the limits established by the manufacturer. No significant breakdowns were found during the evaluation period. Evaluation of DLC indicated an excellent correlation with the manual reference method for neutrophils, lymphocytes, and eosinophils (r = 0.916, 0.936, 0.967, respectively), a good correlation for monocytes (r = 0.11) and a poor correlation for basophils (r = 0.22). Overall flagging for morphologic abnormalities displayed higher sensitivity (5%) than specificity (67%), with a false-positive ratio of 33%. In general, these results are in accordance with those obtained by other authors in the same period of time. (Key words: Blood analyzer; Automated YVBC differential count; Abbott CD 3500) Am J Clin Pathol 1996; 105: Advances in hematology automation have resulted in reliable devices that perform multicell differentials, allowing significant gains in speed and precision compared with the traditional 100-cell manual differential. The CELL-DYN 3500 (Abbott Diagnostics Division, Mountain View, CA) is a fully automated hematology analyzer that provides a comprehensive profile including complete blood count (CBC) and differential leukocyte count (DLC). To separate the five leukocyte subpopulations, the instrument uses the multiple angle polarized light scatter separation (MAPSS) process. 1 ' 2 Moreover, when abnormal leukocytes are present, the system generates different "suspect"flagsand can estimate the percentage of blasts, variant lymphocytes, immature granulocytes, and bands. Over a period of 3 months, we evaluated CBC and DLC performance of the Abbott CD 3500 in a University Hospital Hematology Laboratory using blood From the Hematology Laboratory Department and Postgraduate School of Hematology, Hospital Clinic i Provincial. University of Barcelona. Barcelona, Spain. Supported in part by a grant of the Spanish Ministery of Health (FIS Ref. 93/ ). Manuscript received July ; revision accepted December 6, Address reprint requests to Dr. Vives-Corrons: Haematology Laboratory Department, Hospital Clinic i Provincial, Villarroel, Barcelona, Spain. samples collected from adult patients into K3 EDTA tubes, following the guidelines prepared by the International Council for Standardization in Hematology 3 and using the HA-20 protocol of the National Committee for Clinical and Laboratory Standards. System Description MATERIALS AND METHODS The Abbott CD 3500 uses both impedance and laser optical methodology, and is designed to provide CBC and a five-part DLC. Leukocytes are counted and classified using a combination of hydrodynamic focusing and laser light scatter derived from the MAPSS technology. 12 The light source is a helium-neon laser and the angle of scatter is a function of cell size, refractive index, nuclearcytoplasmic ratio, nuclear shape, and granularity. A 10 narrow angle depolarized laser scatter is added to the traditional 0 and 90 angles used by other devices. Hydrodynamic focusing and the use of laminarflowtechnology place cells single file as they stream through a flow cell for precise analysis. A unique feature is the addition to the single-aperture impedance RBC/PLT transducer of a von Behrens plate device that prevents the cells from reentering the sensing zone after passing through the orifice and therefore, protects against an overestimate of 553

2 55 HEMATOPATHOLOGY CD 3500 TABLE 1. CRITERIA OF MORPHOLOGICAL DLC FLAGGING (TRUE POSITIVE) True-Positive Flags BD Bands > 6% Ig Myelocytes and/or metamyelocytes and/or promyelocytes > 2% BD/Ig Bands > 6% and/or myelocytes and/or metamyelocytes and/or promyelocytes > 2% VL Atypical or immature or abnormal lymphocytes > 5% Blasts Blasts > 1% NRBCs Nucleated red blood cells > 2% BD = bands; VL = variant lymphocytes; NRBCs = nucleated red blood cells. the counts. Results are presented in a report that includes 12 CBC parameters, percentages, and absolute numbers of five-part DLC, 6 scattergrams and 2 histograms. The software enables storage of all information including scattergrams in a 10,000 sample memory, and the complete analysis requires 355 nh of blood in the primary automatic sampling mode, but only 130 nl in the secondary mode. The instrument is equipped with samplers that automatically transport and mix the blood tube. In our experiment, the Abbott CD 3500 analyzer was installed and operated in accordance with manufacturer's instructions. Following the manufacturer's guidelines, calibration was done by matching nondifferential parameters to the reference fully automated hematology analyzer, already in use in our laboratory (Technicon H*2 Bayer Diagnostics). Manufacturer-provided materials were used for calibration and daily quality control. Before starting the evaluation, service engineers of the Abbott Diagnostic Division examined the instrument to ensure optimum installation and general performance. Facility The Clinic Hospital of the University of Barcelona is a 1,000-bed tertiary care public institution and a referral center for Hematology Laboratory.The team responsible for the evaluation process included a team leader (JLV), key operators (MK, MP, RG, and ME), and specialists in quality assurance and informatics (JMJ, GG, and IB). Blood Samples The evaluation was performed with whole blood specimens obtained from the routine samples received in the laboratory. All samples were collected as venous blood in 5 ml evacuated tubes containing K 3 EDTA (1.5 mg/ ml), and first run through the reference instrument, then tested within 1 hour, on the Abbott CD-3500, except for the sample stability studies. All blood films were TABLE 2. COEFFICIENTS OF VARIATION OF WITHIN AND BETWEEN BATCH PRECISION OBSERVED WITH THE ABBOTT CD 3500 WBCs RBCs Hemoglobin Packed cell volume MCV MCH MCHC RDW Platelets MPV PDW PCT Neutrophils Lymphocytes Monocytes Eosinophils Basophils Within (%) Between (%) WBCs = white blood cells; RBCs = red blood cells; MCV = mean corpuscular volume; MCH = mean corpuscular hemoglobin; MCHC = mean corpuscular hemoglobin concentration; RDW = red cell distribution width; MPV = mean platelet volume; PDW = platelet distribution width; PCT = plaquetocrit. made within hours of collection. Assessments were conducted in accordance with the recommendations of the ICSH. 3 Samples were processed in parallel on the laboratory's routine Technicon H*2 analyzer. Linearity Linearity was evaluated by analyzing serial dilutions (10% to 100%, in 10% increments) of five specimens in platelet-free autologous plasma selected to cover the ranges for the dilution-sensitive hematologic parameters (HgB, HCT, RBC, WBC, PLT). Results were converted to graph form for visual examination and further tested in accordance with the recommendations of the ICSH. 3 Imprecision Between-batch imprecision was assessed by analyzing 20 samples in triplicate and in three consecutive batches. Within-batch (replicate) imprecision was evaluated by TABLE 3. TOTAL LEUKOCYTE (WBC), RED BLOOD CELL COUNT (RBC), HEMOGLOBIN (HgB), AND PLATELET COUNT (PLT) PERCENT CARRYOVER IN THE ABBOTT CD 3500 WBCs RBCs Hgb Values are given as the mean of five replicate experiments. Pit 0.3 A.J.C.P.-May 1996

3 VIVES-CORRONS ET AL. 555 Evaluation of Abbott CD 3500 Blood Analyzer TABLE. COMPARABILITY TEST BETWEEN ABBOTT CD 3500 AND TECHNICON H*2 Parameter CD 3500 H*2 P Slope Intercept r P WBC(X10 9 /L) RBC(X10'7L) Hemoglobin (g/l) Packed cell volume (L/L) MCV(fL) MCH (pg) MCHC(g/L) RDW Plt(X10 9 /D MPV(fL) Neutrophils (%) Lymphocytes (%) Monocytes (%) Eosinophils (%) Basophils (%) 10.3 ± ± ± ± ± ± ± ± ± ± ± ±3. 0. ± ± ± ± ± ± ± ± ±2..3 ± ± ± ± ±.6 2. ± ± <10~ <10' <10~ 6 <10-5 <10" <10-3 <10~ <I0" 6 <I0" 6 <10" 2 See Table 2 for definition of acronyms. analyzing 20 samples in triplicate, and three times in the same batch. The results were expressed in terms of coefficient of variation (CV) for each CBC parameter. In both groups the statistical method used was the two-way analysis of variance. Carryover Carryover was evaluated for total WBC, RBC, HB and PLT as described by ICSH. 3 High to low sample carryover testing was performed by testing a high value specimen consecutively in triplicate (a!,a 2,a 3 ), followed immediately by testing a low value specimen consecutively in triplicate (b,b 2,b 3 ).For each magnitude, five experiments were performed. Percent carryover for each parameter was calculated from the formula: (bi-b 3 )/(a 3 -b 3 ) X 100. Comparability One hundred fifty blood samples selected on the basis of the recommended ICSH range of values 3 were analyzed side-by-side with the Abbott CD 3500 and Technicon H*2 instruments. All the CBC and DLC parameters measured were compared by linear regression and paired t-test analysis. Effect of Storage Blood samples were obtained from 10 healthy volunteers and 10 patients with abnormalities affecting the various parameters. Measurements were performed immediately (0 time), then again at 30 minutes, 1 hour, 2 hours, hours, 6 hours, hours, 12 hours, 2 hours, and hours from the time of collection on separate specimens stored for the various times at room temperature (20 C) and at C. Results for all parameters were compared by using initial analyses (time 0 minutes) as 100% values and values obtained at varying times as percentages of the initial time value. Inaccuracy Inaccuracy was evaluated in 50 blood samples selected on the basis of the ICSH recommended range of values 3 by comparing three CBC parameters (HgB, HCT, and MCHC) from the Abbott CD 3500 with the ICSH reference standard. 56 The data were tested by linear regression analysis and the paired t- test g 2 m 0 o> S -" ) Time (h) FIG. 1. Effect of specimen storage at room temperature (1-20 C) on the complete blood count (CBC) parameters of the Abbott CD 3500 blood analyzer. Vol. 105-No. 5

4 556 HEMATOPATHOLOGY Neutrophils * Lymphocytes Monocytes Eosinophils * Basophils 12 2 Time (h) FIG. 2. Effect of specimen storage at room temperature (1-20 C) on the differential leukocyte count (DLC) results of the Abbott CD 3500 blood analyzer. Evaluation of DLC A total of blood specimens classified into "instrumentflagged" (7 samples) and "normal" (103 samples) groups, were coded and examined independently by four experienced technologists who each carried out a blood film evaluation and a 00-cell manual differential (200- cell counts on two different smears). Results from the two 200-cell differential counts were averaged and the results analyzed according to the NCCLS H20-A protocol. The percentage of neutrophils from the Abbott CD 3500 were compared with the smear total percentage of neutrophils (segmented, band, metamyelocytes, myelocytes and promielocytes). The Abbott CD 3500 lymphocytes, eosinophils monocytes, and basophils were compared with the percentage of lymphocytes, eosinophils, monocytes and basophils on the blood film. Clinical Sensitivity Morphologic classification of bloodfilmsas normal or abnormal was based on the visual reference DLC data as described in Table 1. Reference qualitative abnormals included samples exhibiting one or more of the following abnormalities: neutrophil left shift (BAND), presence of immature granulocytes (IG), IG/BAND, blast cells (BLAST), nucleated red blood cells (NRBC) and atypical or variant lymphoid cells (VarL). Abbott CD 3500 qualitative classifications were based on the positivity for certain descriptive flags reported by the instrument. Samples were identified as qualitatively abnormal (flagged) when one or more of the flags that indicated an abnormality of the leukocyte population was positive. The flags were assessed collectively. Linearity RESULTS Inspection of plots of CBC parameter values against a wide sample dilution range indicated a satisfactory linearity for all the parameters sensitive to dilution (HgB, RBC, WBC, HCT, and PLT). Statistical analysis revealed a correlation coefficient higher than 0.9 for all the CBC parameters evaluated. Imprecision Coefficients of variation (CV%) for each CBC parameter and for both between-batch and within-batch experiments are shown in Table 2. These results indicate a good precision (within and between batches) for RBC, HgB, MCV, and MCHC, but moderate precision for WBC, HCT, MCH, and RDW. As expected with the PLT count and the DLC, the instrument imprecision showed a wider CV range. Carryover Carryover data for WBC, RBC, HgB and PLT are presented in Table 3. Except for RBC count, the results of high to low carryover testing were less than 1%. TABLE 5. RESULTS OF WHOLE BLOOD STABILITY TESTS GIVEN IN ABSOLUTE VALUES Time WBCs RBCs HgB MCV Pit Hct Neutrophils Lymphocytes Monocytes Eosinophils Basophils Hours) (X 10 9 /L) (X 10,2 IL) (g/q (X 10* IL) (L/L) (<*>) (%) (Ob) (%) (%) w See Table 2 for definition of acronyms. A.J.C.P.'May 1996

5 VIVES-CORRONS ET AL. 557 Evaluation of Abbott CD 3500 Blood Analyzer TABLE 6. RESULTS OF INACCURACY ASSESSMENT OF THE ABBOTT CD 3500 DLC Parameter CD 3500 Manual DLC* Slope Intercept Neutrophils (%) Lymphocytes (%) Monocytes (%) Eosinophils (%) Basophils (%) 6.5 ± ± ± ± ± ± ± ± ± 0. ± 0. <10~ 5 <10~ 5 <10~ 5 <10~ 3 <10" <10-6 <10" 3 <10" 5 00 cell manual differential counts. Comparability Abbott CD 3500 and Technicon H*2 measurements of CBC parameters were compared. The results of linear regression analysis and a paired t-test are shown in Table. There was an excellent correlation between the two instruments as demonstrated by the r values for WBC, HgB, HCT, RBC, MCV, PLT, NEUT, LYM, and EOS, moderate correlation for MCH, RDW, and MONO, and poor correlation for MCHC, MPV, RDP, PCT, and BASO. The paired t-test demonstrated significant differences between instruments for all the parameters with the exception of the WBC, HgB, MCV, LYM, and BASO. Effect of Storage Results of the stability tests are represented in Figures 1 and 2. The results given in absolute values are represented in Table 5. At room temperature (1-20 C) variation from the original values was <10% for all parameters, except for the WBC count that decreased by 12.1% at hours, and the percentage of monocytes and basophils that decreased at 2 hours. The eosinophil counts showed a slight increase at 12 hours. The refrigerated samples ( C) also showed essentially stable results for up to 2 hours for all the parameters (VPM, PCT, MONO, and BASO presented a variation >10% at 2 hours). TABLE 7. ASSESSMENT OF THE ABBOTT CD 3500 FLAGGING SYSTEM ACCORDING TO THE CRITERIA DEFINED IN TABLE 1 Total samples tested: True positive (0) Sensitivity: X 100 = 5% True positive (0) + False negative (7) True negative (96) Specificity: X 100 = 67% True negative (96) + False positive (7) False positive (7) False positive (%): X 100 = 33% False positive (7) + True negative (96) False negative (7) False negative (%): X 100= 15% False negative (7) + True positive (0) Inaccuracy Results of inaccuracy studies revealed an excellent correlation between the Abbott CD 3500 and the reference standard for HgB and HCT (r = in both cases), but a poor correlation for MCHC (r = 0.720). Paired t-test showed a significant difference only for MCHC (301 ±. vs. 295 ± 12., P =.05). Evaluation of DLC Of the initial blood samples, 7 were excluded because of the presence of qualitative morphologic abnormalities after visual observation of the blood film (reference 00-cell manual differential). Therefore for the comparative study of the instrument and the manual method, a total of 13 samples were used. The results of the statistical analysis of the Abbott CD 3500 and the 00-cell manual differential count are shown in Table 6. A close correlation was observed between the instrument and reference method for neutrophils (r = 0.936), lymphocytes (r = 0.916) and eosinophils (r = 0.967). Clinical Sensitivity One hundred-ninety blood samples were chosen to evaluate the significance of the Abbott CD 3500 flags. The results of the instrument's flagging assessment, performed following the method of Galen and Gambino 7 are summarized in Table 7. The Abbott CD 3500 performance was substantially higher for true negative (TN) than false negative (FN) flags, and essentially similar for false positive (FP) and true positive (TP) flags. Considering the presence of any flag, 7 samples were flagged by the system, giving an FP rate of 33% and a specificity of only 67%. The breakdown of specific flags is summarized in Table. DISCUSSION The Abbott CD 3500 is an automated hematology analyzer that provides a CBC with cell sizing and a five-part DLC by using the MAPSS technology. 1,2 With regard to Vol. 105-No. 5

6 55 HEMATOPATHOLOGY TABLE. BREAKDOWN OF SPECIFIC FLAGS Flag True Positive False Positive False Negative True Negative Total FP Rate <%) FN Rate <%) Bands Ig Bands/Ig VL BL NRBCs VL = variant lymphocytes; BL = blast cells: NRBCs = nucleated red blood cells. reproductibility (between and within-batch imprecision), the instrument's performance in our study was satisfactory. Furthermore it demonstrated minimal carryover and a good linearity for HgB, WBC, RBC, HCT, and PLT. Stability studies of blood samples in EDTA (K 3 ) at 1-20 C demonstrated time dependent variations in the monocyte and basophil percentages after 2 hours of storage. However, by hours, there was an increase in the proportion of eosinophils, and a decrease in the total WBC count and in the proportion of monocytes and basophils. Storage effects can be reduced by maintaining samples at C. Although no reduction of the proportion of neutrophils was observed after 2 hours of storage as reported by Theodorsen and colleagues, our results give further support to the ICSH recommendations that automated DLC should be performed within a day of blood sampling. 9 For nearly all parameters measured in our study, there was an excellent correlation between the Abbott CD 3500 and the Technicon H*2, and the Abbott CD 3500 provides excellent quantitative results for the WBC differential without quantitative morphologic abnormalities. Furthermore, agreement between Abbott CD 3500 and the 00-cell manual differential count was good for eosinophils, lymphocytes, and neutrophils, modest for monocytes and poor for basophils. For monocytes, this low performance may have been due to the reduced presence of this population in the samples examined 10 or to the difficulty of monocyte recognition by the operators when using Romanowsky stain." For basophils, the low correlation with the reference method was probably because of the reference method's statistical fallibility, due to the low number and irregular distribution of the cells in the peripheral blood films. Like other automated hematology analyzers, the Abbott CD 3500 can generate leukocyte flags to alert the user to the presence of abnormal cells. The sensitivity and specificity of the flagging system (clinical sensibility) was tested by classifying all samples into TN, TP, FN, FP, according to the instrument's flags and the morphologic abnormalities encountered after blood film examination. For suspect flags, the instrument exhibited a good sensitivity rate (5%) with a specificity rate of only 67%, reflecting the high false-positive ratio (33%). Recent evaluations of the Abbott CD reported similar results with a specificity rate of 72% and a false-positive ratio of 2%. Gulati and colleagues 1 mentioned that morphologic abnormalities not consistently detected by the Abbott CD 3500 include occasional cases of increased band cells and the presence of neutrophil hypersegmentation or atypical lymphocytes. In this study, the Abbott CD 3500 failed to identify four cases with 7% to 15% of band cells, two cases with atypical lymphocytes, and one case with a moderate number of immature granulocytes. The clinical significance of missing such abnormalities when the overall hematology profile is normal has been questioned. 15 The detection of increased bands adds little to the diagnosis of infection or inflammation in persons with neutrophilia 16 or with normal total WBC counts. 17 Two recent evaluations of four fully automated hematology analyzers 119 reported similar conclusions for all the instruments (Technicon H*2, Coulter STKS, Abbott Cell-Dyn 3000 and Sysmex NE-000). These results underscore the importance of routine blood film examination, even in cases that do not require a visual differential count. In conclusion, the Abbott CD 3500 essentially appears to improve on the Abbott CD in its ability to identify abnormal cell populations by lowering the falsenegative rates without increasing the false positives. This raises our confidence in the qualitative flagging of the instrument and in its usefulness as an adjunct to routine hematologic blood examination. Acknowledgments. The authors thank M. Kinder, M. Perez, R. Garcia, and M. Espasa for their skill and willingness in technical support. REFERENCES Terstappen LW, degrooth BG, Visscher K. Four parameter white blood cell differential counting based on light scattering measurements. Cytometry 19;9:39-. Al-Ismail SA, Bond K, Carter AB, et al. Two-centre evaluation of the Abbott CD 3500 blood counter. Clin Lab Haematol 1995;17: AJ.C.P.-May 1996

7 VIVES-CORRONS ET AL. 559 Evaluation of Abbott 3500 Blood Analyzer 3. International Council for Standardization in Haematology. Guidelines for the evaluation of blood cell analyzers including those used for differential leukocyte and reticulocyte counting and cell marker applications. Prepared by the expert panel on Cytometry. Clin Lab Haematol 199; 16: National Committee for Clinical Laboratory Standards. Leukocyte Differential Counting NCCLS Approved Standard H20-A. Villanova, PA, National Committee for Clinical Laboratory Standards. Reference and Selected Procedures for the Quantitative Determination of Haemoglobin in Blood NCCLS Approved Standard H15-A2. Villanova, PA, International Committee for Standardization in Haematology. Recommended methods for the determination of packed cell volume by centrifugation. WHO/Lab, 199, pp Galen RS, Gambino SR. In: Beyond Normality: The Predictive Value and Efficiency of Medical Diagnosis. New York: John Wiley, 1975, p 19.. Theodorsen I, Holter R, Akselberg K, Lande K. Comparison of five automated haematology analyzers. SysmexJ Intern 199;: International Council for Standardization in Haematology (ICSH). Recommendations for ethylenediaminetetraacetic acid anticoagulation of blood for blood cell counting and sizing: Expert Panel on Cytometry. Am J Clin Pathol 1993; 100: Rumke CL. Imprecision of ratio-derived differential leukocyte counts. Blood Cells 195:11: Verwilghen RL. Standardization and cell counting in haematology: ICSH document. Biochim Clin 1990; 1: Cornbleet PJ, Myrick D, Judkins S, Levy R. Evaluation of the Cell- Dyn 3000 differential. Am J Clin Pathol 1992;9: Aubert-Letrillart B, Binet JL, Baudet S, Azguiz Z, Merle-Beral H. Performance evaluation of the haematological analyzer Cell- Dyn 3000 (Abbott). Nouv R Francaise d'haematologie 1993;35: GulatiGL. Advances of the past decade in automated hematology. Am J Clin Pathol l992;9(suppll):sl I-SI Bridgen ML, Page NE. The lack of clinical utility of white blood cell differential counts and blood morphology in elderly individuals with normal hematology profiles. Arch Pathol Lab Med 1990;11: Morse EE, Nashed A, Spilove L. Automated differential leukocyte counts. Ann Clin Lab Sciences 199; 19: Ardron MJ, Westengard JC, Dutcher T. Band neutrophil counts are unnecessary for diagnosis of infection in patients with normal total leukocyte counts. Am J Clin Pathol 199; 102: Bentley SA, Johnson A, Bishop CA. A parallel evaluation of four automated hematology analyzers. Am J Clin Pathol 1993; 100: Jones RG, Faust AM, Matthews RA. Quality team approach in evaluating three automated hematology analyzers with five-part differential capability. Am J Clin Pathol 1995; 103: Jou JM, Aymerich M, Villamor N, Brugues RM, Vives-Corrons JLL. Resultadosde la evaluation del autoanalizador hematologicocell-dyn 3000 (CD 3000) (Summary). Sa^re 1992;37:103.