Evaluation of the Abbott CELL-DYN 4000 Hematology Analyzer

Transcription

1 Hematopathology / EVALUATION OF THE ABBOTT CELL-DYN 4 HEMATOLOGY ANALYZER Evaluation of the Abbott CELL-DYN 4 Hematology Analyzer Ernesto Grimaldi, MD, and Francesco Scopacasa, PhD Key Words: Abbott CD 4; Blood cell counter; Automated hematology analyzer; Accuracy; Precision; Interference Abstract A new generation hematology analyzer, Abbott CELL-DYN 4 (CD 4), capable of providing 6 parameters, including fully automated reticulocyte, nucleated RBC, blast, band, and immature granulocyte, and variant lymphocyte counts, was evaluated by using the National Committee for Clinical Laboratory Standards H-A protocol and compared with the Bayer-Technicon H- analyzer, which is used routinely in our laboratory. A lipid interference experiment and a sample aging study also were performed. Linearity, carryover, and precision were within the limits established by the manufacturer, and satisfactory agreement was found with the H- analyzer. The evaluation of leukocyte differential counts indicated an excellent correlation with the manual reference method for neutrophils and lymphocytes, a good correlation for monocytes and eosinophils, and a poor correlation for basophils in samples with low counts; for basophil counts of % or higher, we found an improvement of the correlation coefficient. In the lipid interference experiment, only hemoglobin determination was influenced significantly on the CD 4, but by using a new Abbott hemoglobin reagent, the interference was eliminated. The CBC and differential counts were stable and reportable up to at least 4 hours. Intrasample viability information on leukocytes provided a quality check on each individual specimen. The Abbott CELL-DYN 4 (CD 4) (Abbott Diagnostics, Abbott Park, IL) is a new generation fully automated hematology analyzer that uses 4-angle argon laser light scatter and -color fluorescence flow cytometry. 1 The analyzer provides the laboratory with up to 6 blood count parameters, including reticulocytes, immature reticulocyte fraction, nucleated RBCs, blasts, bands, immature granulocytes, and variant lymphocytes; in addition, fluorescence technology also provides a new intrasample quality control feature, the WBC viable fraction. 3 A confidence fraction for blasts, bands, immature granulocytes, and variant lymphocytes also was provided. 1 During a 3-month period, we evaluated, on the basis of the H-A protocol of the National Committee for Clinical Laboratory Standards, 4 the CBC and differential WBC count performance of the CD 4 in comparison with a reference fully automated hematology analyzer in use in our laboratory, the Bayer-Technicon H- (Bayer-Technicon, Tarrytown, NY) and with the reference manual method (differential WBC count only). We also evaluated the lipid interference on both instruments and the sample aging effects on CD 4 measurements. Materials and Methods System Description All analyses were performed on the Abbott CELL-DYN 4 using software version R4-1F without individualized laboratory settings for the laser scatter channels (Abbott setup). The system incorporates 4 different measurement technologies, 1 including fluorescence flow cytometry, and uses an argon laser to undertake analyses of reticulocytes and nucleated RBCs in particular. The analyzer fluorometer section Am J Clin Pathol ;113: on 13 February 18

2 Grimaldi and Scopacasa / EVALUATION OF THE ABBOTT CELL-DYN 4 HEMATOLOGY ANALYZER has fluorescence emission laser optics (4 angles of light scatter). Blasts, immature granulocytes, bands, and variant lymphocytes are detected by a multiparameter, multiweighted, discriminant function; this function generates a flag and reports a confidence fraction from.5 to.99 (ie, the probability that these cells were classified correctly). After detection by the discriminant function, cell counts are estimated by using light scatter. Determination of the hemoglobin concentration is based on spectrophotometry. The default system measurements of the RBC and platelet counts are by impedance and optical methods, respectively; in addition, a simultaneous determination of optical RBC and impedance platelet counts are performed. Discrepancies between the measurements provide an alert suggesting the presence of sample interferences. Fluorescence technology also provides a new intrasample quality control feature, the WBC viable fraction, determined from the ratio of unstained to total WBCs. By combining the WBC viable fraction results with the fluorescence information, the CD 4 can flag specimens (nonviable WBCs) for potential sample deterioration. 3 The CD 4 and H- analyzers were calibrated following the manufacturer s guidelines and using their own calibration samples ; the instruments were controlled by routine quality control methods. The low, normal, and high control samples were supplied by the manufacturers. Blood Samples The evaluation of the instrument was performed by analyzing blood specimens from routine samples after obtaining patient s informed consent. Specimens were selected to span the full range of concentrations expected in clinical practice, and at least half the samples were from patients with blood disorders giving results in abnormal (low and high) ranges. In particular, for differential WBC count abnormalities, 1% of these samples had blasts, % immature granulocytes, 1% variant lymphocytes, 5% nucleated RBCs, and 5% progranulocytes; 5% had no abnormalities; bands ranged from % to %. All samples were collected in evacuated 3.5-mL tubes containing EDTA K3 and were analyzed within hours after obtaining the blood sample. For the aging study, the samples were stored at room temperature or refrigerated (4 C) and analyzed within 4, 48, and 7 hours. Carryover Assessment Carryover assessment was performed according to method of Broughton 5 : a sample in the high range was tested 3 consecutive times (H1, H, H3), and a sample in the low range was tested 3 consecutive times (L1, L, L3). The percentage of carryover for each parameter was calculated as follows: [(L1 L3)/(H3 L3)] 1. Linearity Linearity was evaluated by analyzing serial dilutions of 5 specimens in platelet-free autologous plasma. 6 The results were evaluated in accordance with the International Council for Standardization in Haematology recommendation. 5 Imprecision For the between- and within-batch imprecision study, samples in triplicate and in batches on the same day were analyzed. The instrument was switched off and recalibrated between the batches. The results were expressed as mean, SD, and coefficient of variation percentage. Comparability We selected 1 blood samples on the basis of the recommended International Council for Standardization in Haematology range of values 7 and analyzed them side-by-side with the CD 4 and H- instruments. Calibration was done following the manufacturer s guidelines. Measured WBC, RBC, hemoglobin, mean corpuscular volume, platelet, and differential WBC count parameters were compared by linear regression analysis, and correlation coefficients (r) were calculated. Since bands and immature granulocytes are counted together with neutrophils by the H- analyzer, these cells, well defined by the CD 4, were added to neutrophil counts in the statistical evaluation. We also compared 11 normal blood samples with the manual differential WBC count; all films were examined independently by 4 experienced technologists 4 ; results were compared by linear regression, and correlation coefficients were calculated. Interference Study Lipid interference was evaluated in 1 blood samples added with increasing amounts of Intralipid (soya lipids) (Pharmacia AB, Uppsala, Sweden); they were analyzed sideby-side by using the instruments, and each test was performed in triplicate. The results were expressed as the mean value of 3 determinations. The lipid concentration was measured as triglycerides. At the end of the evaluation period, Abbott supplied a new hemoglobin reagent, Tri-Methyl-Tetra-Decyl-ammonium chloride free (TTAB free), and a new software version (R8-1) that are able to eliminate any interference on CD 4 hemoglobin determination. The lipid interference experiment also was performed with these new components. Time (Aging) Study We analyzed 5 normal samples in duplicate within, 4, 48, and 7 hours after venipuncture and storage at room temperature (5 C) or under refrigeration (4 C). 498 Am J Clin Pathol ;113: on 13 February 18

3 Hematopathology / ORIGINAL ARTICLE Results Carryover Carryover data for the WBC, RBC, hemoglobin, and platelet values are given in Table 1. The results for high to low carryover testing were less than.5%. Linearity Linearity data are given in Table. Excellent correlation was found between expected and obtained values for all parameters sensitive to dilution, also in a very low range. Statistical analysis revealed a correlation coefficient higher than.997 for WBC, RBC, hemoglobin, and hematocrit values and a lower correlation for platelet values (.99). We also found that the unflagged percentage for the WBC differential count results were reproducible and linear down to a level of. 1 9 /L (data not shown). Imprecision As shown in Table 3, there is good within- and between-batch reproducibility for all parameters, including the CBC and WBC differential counts. Only for basophils was the coefficient of variation higher than %. Comparability As shown in Table 4, the CD 4 gave CBC and differential WBC count results that compared very well with data obtained by the H-. The results for basophils showed poor agreement (r =.97) between the instruments in the low range of counts and good agreement (r =.9) for high basophils counts. The results of correlation and linear regression between the CD 4 and manual differential are given in Table 5. Correlation was excellent for neutrophils and lymphocytes, and monocytes and eosinophils showed a good correlation, while basophils results yielded a lower correlation in the low range of values (<%). The correlation for basophils became excellent when only samples with basophil values of % or higher (n = 5) were considered. The comparability data plot for the CD 4 vs the manual differential count for the basophil counts, low and high ranges, are given in Figure 1 and Figure, respectively. The identity line is displayed together with the 95% Table 1 Percentage of Carryover in the CELL-DYN 4 Analyzer * WBC RBC Hemoglobin Platelets * Abbott Diagnostics, Abbott Park, IL. binomial confidence limits for a 4-cell differential WBC count. In Figure 1, several outliers were identified by 95% confidence limits; no outliers were identified in Figure. Interference Study As shown in Figure 3, a significant (P <.1) inaccuracy of the hemoglobin determinations was observed on both instruments starting from a lipid value of 45 mg/dl, resulting a hemoglobin overestimation of 1 g/dl or more. Conversely, no interference was found with either instrument in the WBC and differential WBC counts (data not shown). The RBC and platelet counts in hyperlipidemic samples Figure 4 and Figure 5 were overestimated on the H- system, but only for high lipid concentrations, while no interference was found on the CD 4. No more interference in the hemoglobin determination was observed using the new Abbott hemoglobin reagent (TTAB free) and the new software version Figure 6. Time (Aging) Study In stored samples, no significant changes were observed in WBC, platelet, RBC and hemoglobin determinations by the CD 4 (data not shown). In the WBC count, the WBC viable fraction decreased significantly (P <.1) between 4 and 48 hours, particularly in samples stored at room temperature Figure 7. At 48 hours, a significant (P <.1) increase of samples showing the nonviable WBC flag was observed Figure 8. No significant changes were observed in the differential WBC count in stored samples; only a minor reduction in the proportion of eosinophils was noted in samples stored for 48 hours at room temperature Figure 9. Conversely, a significant (P <.1) change was observed in the segmented neutrophil and band subpopulations Figure 1. The time course shows a progressive decrease in the number of mature neutrophils (segmented) and a side-byside, progressive increase in the number of bands; this feature is more evident in samples stored at 4 C Figure 11. No significant changes were observed in immature granulocytes (Figure 1). In stored samples, the confidence fraction for neutrophil subpopulations was always less than.76, whereas the fraction was always.76 or more in fresh blood samples Figure 1. Discussion In our study, the CELL-DYN 4 reproducibility (between-batch and within-batch imprecision) results were satisfactory with the exception of the basophil count. Furthermore, the CELL-DYN 4 demonstrated minimal carryover and a good linearity. Am J Clin Pathol ;113: on 13 February 18

4 Grimaldi and Scopacasa / EVALUATION OF THE ABBOTT CELL-DYN 4 HEMATOLOGY ANALYZER Table Linearity of the CELL-DYN 4 Analyzer * Range r Intercept Slope WBC ( 1 9 /L) RBC ( 1 1 /L) Hemoglobin (g/dl) Hematocrit (%) Platelets ( 1 9 /L) 6.-1, * Abbott Diagnostics, Abbott Park, IL. Table 3 Within- and Between-Batch Precision Observed With the Abbott CELL-DYN 4 * in Samples Within Batch Between Batch Mean SD CV (%) Mean SD CV (%) WBC ( 1 9 /L) RBC ( 1 1 /L) Hemoglobin (g/dl) Hematocrit (%) Mean corpuscular volume (fl) Platelets ( 1 9 /L) Neutrophils (%) Lymphocytes (%) Monocytes (%) Eosinophils (%) Basophils (%) * Abbott Diagnostics, Abbott Park, IL. Table 4 Comparability Test Between Abbott CELL-DYN 4 * and Bayer-Technicon H- * for 1 Samples Intercept Slope r WBC ( 1 9 /L) RBC ( 1 1 /L) Hemoglobin (g/dl) Mean corpuscular volume (fl) Platelets ( 1 9 /L) Neutrophils (%) Lymphocytes (%) Monocytes (%) Eosinophils (%) Basophils (%) Range of values, % to <% Range of values, % to 1% * Abbott Diagnostics, Abbott Park, IL, and Bayer-Technicon, Tarrytown, NY. Segmented neutrophils, bands, and immature granulocytes. Normal and variant lymphocytes. Table 5 Results of Inaccuracy Assessment: Comparison Between CELL-DYN 4 * and Manual Differential Leukocyte Count Intercept Slope r Neutrophils (%) Lymphocytes (%) Monocytes Eosinophils Basophils (%) Range of values, % to <% Range of values, % to 1% * Abbott Diagnostics, Abbott Park, IL. 5 Am J Clin Pathol ;113: on 13 February 18

5 Hematopathology / ORIGINAL ARTICLE CD CD Cell Manual Differential Count Cell Manual Differential Count Figure 1 Percentage of basophils with the CELL-DYN 4 (Abbott Diagnostics, Abbott Park, IL) compared with 4- cell manual differential count for basophil counts in the low range. The identity line is displayed with the 95% binomial confidence limits. Few outliers are identified. Figure Percentage of basophils with the CELL-DYN 4 (Abbott Diagnostics, Abbott Park, IL) compared with 4- cell differential count for the basophil counts in the high range. The identity line is displayed with the 95% binomial confidence limits. No outliers are identified. 4 6 Hemoglobin (g/dl) RBCs ( 1 /L) ,1,1 3,1 4,1 1 1,1,1 3,1 4,1 Triglycerides (mg/dl) Triglycerides (mg/dl) Figure 3 Interference of hyperlipidemic samples on the hemoglobin determination. Both the Abbott CELL-DYN 4 (Abbott Diagnostics, Abbott Park, IL) (squares) and Bayer- Technicon H- (Bayer-Technicon, Tarrytown, NY) (diamonds) show significant overestimation (1 g/dl or more) starting from a triglyceride value of 45 mg/dl or more. Figure 4 Interference of hyperlipidemic samples on the RBC count. The RBC count is stable on both instruments (Abbott CELL-DYN 4, Abbott Diagnostics, Abbott Park, IL [squares]; and Bayer-Technicon H- Bayer-Technicon, Tarrytown, NY [diamonds]) up to a triglyceride value of mg/dl. With higher lipid concentrations, only the H- analyzer shows an RBC count overestimation. For the CBC and differential WBC counts, we found a good correlation with H- analyzer. Blasts and variant lymphocytes were both counted as large unstained cells by the H-, while they were well defined by the CD 4; therefore only neutrophil, lymphocyte, monocyte, eosinophil, and basophil data were comparable. High nucleated RBCs would give invalid WBC comparison data (interference with the WBC count on the H- Am J Clin Pathol ;113: on 13 February 18

6 Grimaldi and Scopacasa / EVALUATION OF THE ABBOTT CELL-DYN 4 HEMATOLOGY ANALYZER 3 5 Platelets ( 1 /L) Hemoglobin (g/dl) ,1,1 3,1 4,1 1 1,1,1 3,1 4,1 Triglycerides (mg/dl) Triglycerides (mg/dl) Figure 5 Interference of hyperlipidemic samples on the platelet count. The platelet count performed by the Abbott CELL-DYN 4 (Abbott Diagnostics, Abbott Park, IL) (squares) is stable up to a triglyceride value of 4,5 mg/dl. The platelet count by the Bayer-Technicon H- (Bayer-Technicon, Tarrytown, NY) (diamonds) shows significant overestimation starting from a triglyceride value of 1, mg/dl. Figure 6 Interference of hyperlipidemic samples on the hemoglobin determination performed by CELL-DYN 4 (Abbott Diagnostics, Abbott Park, IL) using the new Abbott hemoglobin reagent (TriMethyl-Tetra-Decyl-Ammonium chloride free) (squares) compared with the Bayer-Technicon H- (Bayer-Technicon, Tarrytown, NY) (diamonds). No more interference on the hemoglobin determination was observed when using the new free hemoglobin reagent. WBC Viable Fraction (%) Nonviable WBCs (% of Samples) Figure 7 Time course of the WBC viable fraction in stored samples. In the WBC count, the viable fraction decreased significantly (P <.1) within 48 hours, particularly in samples stored at room temperature. Diamonds, samples stored at 4 C; squares, samples stored at room temperature. Figure 8 Nonviable WBC flag in stored samples. A significant increase (P <.1) of samples showing the nonviable flag was observed only in samples stored more than 4 hours, especially at room temperature. Gray bars, samples stored at room temperature; black bars, samples stored at 4 C. analyzer). Because only a few samples with nucleated RBCs passed through the laboratory during this evaluation, a statistically insignificant number of these samples was included in our study (5%), and no interference with comparison was observed. We also found a good correlation with the reference manual differential for all WBC subpopulations except for basophils in the low range (% to <%). Probably, the high coefficients of variation for basophils and the poor correlation 5 Am J Clin Pathol ;113: on 13 February 18

7 Hematopathology / ORIGINAL ARTICLE A B hours 5 hours Percentage hours 48 hours 7 hours Percentage hours 48 hours 7 hours 1 1 Ne Ly Mo Eo Ba Ne Ly Mo Eo Ba Differential Differential Figure 9 Differential WBC count in stored samples. A, Only minor reduction in the proportion of eosinophils was observed in samples stored at room temperature. B, No significant changes were observed in samples stored at 4 C. Ne, neutrophils; Ly, lymphocytes; Mo, monocytes; Eo, eosinophils; Ba, basophils. 6 6 Percentage 4 Percentage Figure 1 Time course of neutrophil subpopulations in samples stored at room temperature. Significant changes (P <.1) were observed in the segmented neutrophil and band fractions: the time course shows a progressive reduction in the proportion of segmented neutrophils and a sideby-side, progressive increase in the proportion of bands. No change in the immature granulocyte fraction was observed. Diamonds, segmented neutrophils; squares, band neutrophils; triangles, immature granulocytes. Figure 11 Time course of neutrophil subpopulations in samples stored at 4 C. In samples stored at 4 C, the time course shows a greater reduction in the fraction of segmented neutrophils and a greater progressive increase in the band fraction than in samples stored at room temperature. No change in the immature granulocyte fraction was observed. Diamonds, segmented neutrophils; squares, band neutrophils; triangles, immature granulocytes. between the instruments are not related primarily to poor performance of the analyzers for the WBC differential count but to the low number of basophils counted. The low correlation with the reference method could be due to the low number and irregular distribution of the cells in the peripheral blood films or, more probably, to the interference of nonviable WBCs, which were present in all the identified outliers. Finally, when analyzing samples with high basophil counts, a dramatic increase of the correlation coefficient was found. Am J Clin Pathol ;113: on 13 February 18

8 Grimaldi and Scopacasa / EVALUATION OF THE ABBOTT CELL-DYN 4 HEMATOLOGY ANALYZER Value neutrophil nuclear morphologic features (loss of lobularity) in aged samples, as evident when examining the blood films. Finally the WBC viable fraction, the nonviable WBC flag, and the confidence fraction are sensitive intrasample quality checks, and they could contribute to increased accuracy of the counts, particularly when the discrimination between mature granulocytes and bands is of clinical interest. In particular, the confidence fraction, which indicates the probability that a cell population was classified correctly, in our study was always.76 or more in samples that were not stored, whereas the confidence fraction was always less than.76 in stored samples for all WBC subsets. In fact, in the presence of high band counts, a WBC viable fraction of 95% or less and/or a nonviable WBC flag and/or a confidence fraction less than.76 could indicate sample deterioration and, therefore, the need for a fresh sample. Figure 1 Time course of the confidence fraction in stored samples. In samples stored at room temperature and samples stored at 4 C, the confidence fraction (ie, the probability that the cells were classified correctly) for segmented neutrophils, bands, and immature granulocytes was always less than.76, whereas the fraction was always.76 or more in samples that were not stored. An excellent correlation for monocytes was found with the H- analyzer (r =.99) and a moderate correlation with the reference method (r =.83). This finding is in agreement with those of other reports 8,9 and seems to be due to artifact distribution for these cells during preparation of the blood films and to difficulties in differentiating small monocytes from large lymphocytes. In the lipid interference study, we observed, as expected, a significant (P <.1) inaccuracy of hemoglobin determination on both instruments starting from a lipid value of 45 mg/dl, while very high lipid amounts (4,8 mg/dl) showed no interference with the WBC, RBC, and platelet counts performed by the CD 4. These data suggest the adequacy of the Abbott CELL-DYN 4 algorithm for discriminating high amounts of hyperlipidemic particles and excluding them from counts as well. Finally, the Abbott TTAB free hemoglobin reagent improves the accuracy of the hemoglobin determination by the CD 4 analyzer, including in samples with very high lipid amounts. In stored samples, the increase in the percentage of nonviable cells was much slower in refrigerated samples (4 C), while storage at higher temperatures induced an acceleration of the process. The apparent change in the relationship of segmented neutrophils to bands (Figures 1 and 11) with storage time is opposite what might be expected with the passage of time. This apparent change should be due to the changes in Conclusions Our evaluation data show that the Abbott CELL-DYN 4 is a precise device, and it is at least as accurate as the Bayer-Technicon H-. The results are linear to extreme counts, and no significant carryover was observed. Derived from a DNA fluorescence analysis, the availability of quantitative and qualitative viability information for WBCs is a new feature on a hematology analyzer and is a very useful tool for analyzing aged blood specimens. In addition, we found that ease of use contributes to the CELL- DYN 4 performance, which may prove very useful in the clinical laboratory. The Abbott CD 4 analyzer represents a substantial advance in hematology instrumentation and has considerable potential for extending the clinical role of hematologic studies. From the Department of Laboratory Medicine, Università Federico II, Naples, Italy. The CELL-DYN 4 analyzer, reagents, and calibration and control samples were provided free of charge for this evaluation by Abbott Diagnostics, Abbott Park, IL. Address reprint requests to Dr Grimaldi: Dipartimento Assistenziale di Medicina di Laboratorio, Policlinico Universitario Federico II, Via S Pansini 5, 8131 Napoli, Italy. References 1. CELL-DYN 4 [product manual]. Santa Clara, CA: Abbott Diagnostics; Ormerod MG. Estimating cell viability, application in cell biology. In: Ormerod MG, ed. Flow Cytometry: A Practical Approach. New York, NY: IRL Press at Oxford University Press; 199: Am J Clin Pathol ;113: on 13 February 18

9 Hematopathology / ORIGINAL ARTICLE 3. D Onofrio G, Zini G, Tommasi M, et al. Integration of fluorescence and hemocytometry in the CELL-DYN 4: reticulocyte, nucleated blood cell and white blood cell viability study. Lab Hematol. 1996;: National Committee for Clinical Laboratory Standards. Reference Leukocyte Differential Count (Proportional) and Evaluation of Instrumental Methods; Approved Standard. Villanova, PA: National Committee for Clinical Laboratory Standards; 199. NCCLS document H-A. Vol Broughton PMG, Gowenlock AH, McCormach JJ, et al. A recommended scheme for the evaluation of instruments for automated analysis in the clinical biochemistry laboratory. J Clin Pathol. 1969;: Shinton NK, England JM, Kennedy DA. Guidelines for the evaluation of instruments used in haematology laboratories. J Clin Pathol. 198;35: International Council for Standardization in Haematology Expert Panel on Cytometry. Guidelines for the evaluation of blood cell analyzer including those used for differential leukocyte and reticulocyte counting and cell marker applications. Clin Lab Haematol. 1994;16: Lebeck LK, Amst BJ, Houwen B. Flow cytometric white blood cell differentials: a proposed alternate reference method. Sysmex J Int. 1993;3: Goossens W, Van Hove L, Verwilghen RL. Monocyte counting: discrepancies in results obtained with different automated instruments. J Clin Pathol. 1991;3: Am J Clin Pathol ;113: on 13 February 18