Performance Evaluation of the Coulter LH 750 Hematology Analyzer

Transcription

1 ISLH Laboratory Hematology 7: Carden Jennings Publishing Co., Ltd. Official Publication Performance Evaluation of the Coulter LH 750 Hematology Analyzer TRACEY FERNANDEZ, LYNN BESSERT DOMACK, DINAH MONTES, RICHARD PINEIRO, EILEEN LANDRUM, ESTHER VITAL Beckman Coulter, Inc, Diagnostics Division, Miami, Florida ABSTRACT The Coulter LH 750 hematology analyzer is a new fully automated hematology analyzer designed to improve workflow efficiency for high-volume laboratories through elimination of preanalytical sample preparation and sorting as well as reduction of postanalytical manual confirmation of results. The LH 750 can operate in 5 different modes of real-time random access testing, all from a single aspiration of 300 µl whole blood. The random access modes are complete blood count (CBC), CBC + reticulocyte analysis, CBC + 6-part differential including nucleated red blood cells (NRBCs), CBC + 6-part differential + reticulocyte analysis, and reticulocyte analysis only. The LH 750 has extended dynamic ranges for white blood cell (WBC) count ( /L or 10 3 /µl) and platelet count ( /L or 10 3 /µl). Special attention was paid to improved accuracy of low platelet counts at critical decision levels, and these counts showed excellent correlation to reference flow cytometric counts (r = 0.941) for platelet counts between 0 and /L. Thanks to new algorithms, there is a very low incidence of WBC-count interference by giant platelets, platelet clumps, NRBCs, and lyse-resistant RBCs in LH 750 WBC counts compared with flow cytometric WBC counts (r = 0.926). The LH 750 NRBC counts show excellent correlation to manual counts and a predicate device, with r = and 0.918, respectively. Lab. Hematol. 2001;7: Correspondence and reprint requests: Tracey Fernandez, Beckman Coulter, Inc, Diagnostics Division, SW 147th Ave, Miami, FL Received August 20, 2001; accepted August 27, 2001 KEY WORDS: LH 750 hematology analyzer CBC Differential Reticulocytes NRBC INTRODUCTION The LH 750 hematology analyzer (Beckman Coulter, Miami, FL) is a new fully automated hematology analyzer designed to improve workflow efficiency for high-volume laboratories through enhanced flagging and elimination of preanalytical sample preparation and sorting. The LH 750 analyzer enables the laboratory to save time and reagents by optimizing sample throughput using real-time random access by profile type. A sample can be added to a rack at any time without switching the mode of analysis between samples. Sample analysis information, indicating the testing profile required, can be downloaded from the host computer through a bidirectional interface. The bar code on the sample is read, and only the profile requested is performed. The fully automated system can analyze up to 110 samples per hour. Samples from microcollection devices and predilute samples can be analyzed in the manual mode. The LH 750 system has extended dynamic ranges and improved flagging to provide accurate results and reduce the need for timeconsuming and often unnecessary manual review. Other studies have shown the time and cost savings possible as a result of random access and enhanced flagging on the LH 750 analyzer [1]. We evaluated the LH 750 system to fully characterize its performance. In addition to carryover and linearity of the complete blood count (CBC) parameters and precision for CBCs, white blood cell (WBC) differentials, and reticulocytes, we compared results with predicate and/or manual methods. During the evaluation, we focused on parameters expected to improve performance of the 217

2 218 T. Fernandez et al TABLE 1. LH 700 Series Random Access Profiles* Profile CBC CBC + DIFF (6-part differential) CBC + DIFF + RETIC CBC + RETIC RETIC system as a result of new advanced algorithms. This included WBC and platelet parameters and the addition of the nucleated red blood cell (NRBC) count. It has been customary to evaluate performance of new hematology analyzers by comparing them with predicate methods, usually currently available systems. With significant advances in technology available on new analyzers, it is often difficult to measure their improved performance in this way. Accurate and practical reference methods have not been in place for some parameters. Recent new immunologic methods have been developed, and, in some cases, standards have been proposed that provide very accurate results throughout the range without the problems of interference [2,3]. These new methods and standards are especially important for the platelet and WBC results, because these results are used to make critical clinical decisions, and accurate counts, devoid of interference, are imperative. In our studies, we used both automated analyzers as predicate methods and used manual or flow cytometric reference methods to better evaluate performance of all parameters on the LH 750 system. MATERIALS AND METHODS Parameters WBC, RBC, Hgb, Hct, MCV, MCH, MCHC, Plt, MPV, PCT, PDW CBC parameters plus NE%, NE#, LY%, LY#, MO%, MO#, EO%, EO#, BA%, BA#, NRBC%, NRBC# CBC and 6-part differential parameters plus RETIC%, RETIC#, IRF, MRV, HLR%, HLR#, MSCV CBC and RETIC parameters RETIC parameters *CBC indicates complete blood count;wbc, white blood cell; RBC, red blood cell; Hgb, hemoglobin; Hct, hematocrit; MCV, mean corpuscular volume; MCH, mean corpuscular hemoglobin; MCHC, mean corpuscular hemoglobin concentration; Plt, platelets; MPV, mean platelet volume; PCT, platelecrit; PDW, platelet distribution width; DIFF, differential; NE, neutrophils; LY, lymphocytes; MO, monocytes; EO, eosinophils; BA, basophils; NRBC, nucleated red blood cells; RETIC, reticulocytes; IRF, immature reticulocyte fraction; MRV, mean reticulocyte volume; HLR, high light-scatter reticulocyte; MSCV, mean sphered-cell volume. These parameters are for research use only. Not for use in diagnostic procedures. Instrumentation The LH 750 hematology analyzer is a new fully automated hematology analyzer designed to improve workflow efficiency for high-volume laboratories through enhanced flagging and elimination of preanalytical sample preparation and sorting. The laboratory saves time and reagents by optimizing sample throughput using real-time random access by the profile types listed in Table 1. As in previous Coulter systems, the LH 750 analyzer accepts various sizes and types of specimen containers in its cassettes. The fully automated LH 750 system can analyze up to 110 samples per hour in either the CBC or CBC + 6-part differential mode. Microsamples can be analyzed in the open-vial mode and predilute samples can be processed with multiple dilution ratios, reporting corrected CBC results. The capabilities of the LH 700 series instruments are further expanded with the addition of a SlideMaker and SlideStainer as part of a trackless Coulter LH 755 hematology workcell shown in Figure 1. The LH 750 system uses a total of 5 reagents (3 new), detailed in Table 2. The LH 750 analyzer uses new automationready calibration and control materials, the Coulter S-CAL calibrator and 5C-ES cell control, with extended stability of up to 90 days. These calibration and control materials have disk-based expected values that can be automatically downloaded to the system. For those systems with a SlideStainer, prepackaged reagents are also available, including TruColor Wright, Wright-Giemsa, and May-Grünwald-Giemsa stains. The CBC parameters on the LH 750 analyzer are measured using AccuCount technology, an advanced analytical technique combining the Coulter principle of impedance counting and sizing with new sophisticated mathematical algorithms. The algorithms use a combination of mathematical formulas, including particle counting, wait-time counting, and estimated flight of time of the particle, to improve precision and expand the linear range of WBCs and platelets. AccuCount technology extends the linear range for the WBC count from 0 to cells/µl and the platelet count from 0 to cells/µl. The WBC count is automatically corrected in the presence of interference such as NRBCs, giant platelets, platelet clumps, unlysed RBCs, and RBC fragments. Improved platelet count accuracy, particularly FIGURE 1. The LH 755 workcell. LH 750 with LH Slidemaker and LH SlideStainer.

3 Performance Evaluation of the LH 750 Hematology Analyzer 219 TABLE 2. LH Series 700 Reagent System LH 700 series diluent LH 700 series PAK reagent system LYSE S III Diff Lytic Reagent LH 700 series RETIC PAK Coulter CLENZ cleaning reagent Isotonic electrolyte solution for diluting whole blood samples. New formulation that contains no formaldehydeproducing compounds. The reagent allows the leukocytes and nucleated red blood cells to be differentiated into their subpopulations through the volume, conductivity, and light-scatter measurements. Reagent for lysis of erythrocytes (RBCs) and release of hemoglobin (Hgb) with minimal interference from leukocytes. The RETIC PAK contains a clearing solution and New Methylene Blue (NMB) for random access analysis. Cleaning agent that cleans and rinses the internal surfaces of the instrument components. Daily use prevents protein buildup and eliminates the need for routine aperture bleaching. at clinical decision levels, has been achieved by improved algorithms [4]. The NRBC count is an integral part of the differential count and does not require additional reagent. Using VCS technology (assessing volume [V], high-frequency conductivity [C], and laser light scatter [S]) along with an advanced algorithm applied to the WBC count, NRBCs are now identified by the LH 750 analyzer in their signature position in the differential data plot and in the far-left region of the WBC histogram. If cells are present in both locations, the NRBC count is derived from the WBC histogram and reported. Figure 2 illustrates this process. The LH 700 series WBC 6-part differential and reticulocytes are measured using VCS technology with Coulter s advanced IntelliKinetics and AccuGate technologies. Reticulocyte analysis is fully automated and requires no sample prepreparation. To increase efficiency, the LH 750 analyzer offers useradjustable flagging for reference ranges, action limits, critical limits, definitive flags, and sensitivity of suspect flags. Based on their patient population and requirements, any laboratory can input all of the criteria or decide to use the built-in default criteria. Using customized decision rules, laboratories can easily program appropriate follow-up action based on patient results. FIGURE 2. Nucleated red blood cell (NRBC) enumeration is based on NRBCs present in the NRBC signature position on the VCS scatterplot and interference at the white blood cell (WBC) lower discriminator. The NRBC count appears beneath the basophil (BA) count on the Diff screen.

4 220 T. Fernandez et al TABLE 3. LH 750 Carryover* High-to-Low Carryover Unit of LH 750 Parameter Set 1 Set 2 Set 3 Measure Specification WBC % 2.0 RBC % 1.0 Hgb % 2.0 Plt % 2.0 DIFF Events 200 RETIC Events 600 *See Table 1 legend for abbreviation definitions. Specimens The data for this evaluation were collected at Beckman Coulter, Miami, Florida, by the Systems and Applications Support Department. The sample population for this study included both normal and abnormal specimens collected in K 2 - or K 3 EDTA. The normal specimens were obtained from healthy subjects from a blood donor program. Abnormal specimens were residual material from patient specimens drawn for clinical testing at a large tertiary care hospital. All specimens were maintained at room temperature and analyzed by all methods within 1 hour of each other and within 24 hours of phlebotomy. TABLE 4. LH 750 Within-Run Precision: CBC and WBC Differential Parameters* LH 750 Specification Parameter Mean 2 SD CV 2 SD CV WBC N/A 1.7 RBC N/A 0.8 Hgb N/A 0.8 MCV N/A 0.8 RDW N/A 2.2 Plt (range, ) N/A 14.0 Plt (range, ) N/A 3.3 MPV N/A 2.2 NE% N/A LY% N/A MO% N/A EO% N/A BA% N/A *CV indicates coefficient of variation; N/A, no specifications are available. See Table 1 legend for other abbreviation definitions. TABLE 5. LH 750 Within-Run Precision: Reticulocyte Parameters* LH 750 Specification Parameter Mean 1 SD CV 1 SD CV RET% (range, ) RET% (range, ) RET% (range, ) RET% (range, ) MRV N/A N/A IRF N/A N/A *CV indicates coefficient of variation; N/A, no specifications are available. See Table 1 legend for other abbreviation definitions. Methods The statistics used to compare results from the methods were Deming regression analysis and mean difference analysis. Deming linear regression accounts for the errors of measurement of both methods. Carryover Carryover was measured by analyzing 3 consecutive samples of normal whole blood (H1, H2, H3) followed by 3 consecutive samples of LH 700 series diluent (L1, L2, L3). This sequence was repeated 3 times. Carryover for each directly measured CBC parameter (WBC, RBC, hemoglobin [Hgb], platelets [Plt]) was calculated for each individual sequence (L1, L2, L3, and H1, H2, H3). The following formula was used to calculate high-to-low carryover percentage (H/L%): H/L% = ([L1 - L3]/[H3]) 100. Instrument high-to-low carryover for the WBC differential and reticulocyte parameters is tested to a limit of 200 events for the WBC differential parameters and 600 events for the reticulocyte parameter. Within-Run Precision Within-run precision was measured by replicate testing (N = 31) of normal whole blood specimens. For CBC and WBC differential parameters, results are reported for a single TABLE 6. LH 750 Within-Run Precision: CBC Parameters in Predilute Mode, Dilution 1:5* Parameter Mean 2 SD CV WBC RBC Hgb MCV RDW Plt MPV *No specifications are available. CV indicates coefficient of variation. See Table 1 legend for other abbreviation definitions.

5 Performance Evaluation of the LH 750 Hematology Analyzer 221 FIGURE 3. Linearity scatter plots for white blood cells (WBC) (A), red blood cells (RBC) (B), hemoglobin (HGB) (C), and platelets (PLT) (D). Expected value for each dilution (x-axis) is plotted against the average count (y-axis). sample. Within-run precision for platelet and reticulocyte parameters was determined on multiple samples throughout the clinical range, including important low-level clinical decision points. To measure precision on platelets in the low range, blood specimens were altered to obtain the target values. Reticulocyte-percentage precision was measured at 4 levels across the clinical range. Within-run precision was also measured for predilute samples in the manual mode on the LH 750 analyzer. The predilute sample was prepared by manual dilution of whole blood with an isotonic diluent at a ratio of 1:5. The SD and coefficient of variation (CV) were calculated for the directly measured parameters and compared to those of the instrument specification. Linearity Linearity was tested using a set of 11 dilutions prepared using a method based on gravity or adjusted weights. Dilutions for WBC linearity were prepared using monologs. Platelet, RBC, and Hgb linearity dilutions were made from normal whole blood. The dilutions of blood and surrogate materials were tested using the mean of 3 replicate measurements at each level. The mean values obtained were compared to the expected values for each dilution. Results were compared by regression analysis and the mean difference analysis. Method Comparison CBC Parameters. In this study, a total of 415 clinical specimens, with values covering the expected range of performance, TABLE 7. LH 750 Linearity* LH 750 Specification Parameter Linearity Range Mean Diff Mean% Diff Slope Intercept Coefficient of Correlation WBC ( 10 3 /µl) ±0.2 ± N/A ±9.0 RBC ( 10 6 /µl) ±0.05 ± Hgb (g/dl) ±0.2 ± Plt ( 10 6 /µl) ±10.0 ± *See Table 1 legend for abbreviation definitions.

6 222 T. Fernandez et al TABLE 8. Method Comparison: CBC, WBC Differential, and Reticulocyte Parameters, Mean Difference Analysis, LH 750 Versus Gen S* Population Range LH 750 Specification Parameter No. of Samples Low High Mean Diff Mean% Diff Mean Diff Mean% Diff WBC (range, /µl) ±0.2 ±3.5 WBC (range, > /µl) N/A ±9.0 RBC, 10 6 /µl ±0.05 ±2.0 Hgb, g/dl ±0.2 ±3.0 MCV, fl N/A ±2.0 RDW% ±0.5 ±5.0 Plt, 10 6 /µl ±10.0 ±7.0 MPV, fl N/A ±5.0 NE% ±2.0 N/A LY% ±1.5 N/A MO% ±1.0 N/A EO% ±0.5 N/A BA% ±0.5 N/A RETIC% ±0.5 N/A MRV, fl N/A N/A IRF, ratio N/A N/A *N/A indicates no specifications are available. See Table 1 legend for other abbreviation definitions. were analyzed on both the LH 750 and Gen S analyzers [5]. Results obtained for the CBC parameters, WBC, RBC, Hgb, mean corpuscular volume (MCV), hematocrit (HCT), red cell distribution width (RDW), Plt and mean platelet volume (MPV) were compared by regression analysis and the mean difference between results obtained by each method was calculated. The data set for each parameter included only unflagged results. Sample sets were eliminated from the calculations if the following flags were present on results from either analyzer: nonnumeric flag, partial aspiration flag, WBC Interference flag, Plt R flag, and RDW R flag. LH 750 analyzer results for the platelet parameter were also compared against the platelet reference method proposed by the International Council for Standardization in Haematology (ICSH) and the International Society for Laboratory Hematology (ISLH) Task Force on Platelet Counting [2]. This comparator method was performed on a Coulter EPICS XL-MCL flow cytometer using a combination of fluorescein isothiocyanate (FITC)-labeled anti-cd41 and anti- CD61 monoclonal antibodies for platelet enumeration. Results were compared by regression analysis and the mean difference between methods. To characterize the LH 750 system accurate WBC count, accuracy analysis was performed on those samples for which the predicate Gen S system generated a * flag for the WBC parameter. The performance of the LH 750 system accurate WBC count was tested by comparison with flow cytometric counts. The flow cytometric WBC count was obtained on an XL-MCL flow cytometer using a combination of the leukocyte marker monoclonal antibody CD45-PC5 and the platelet marker monoclonal antibody CD41-FITC [3]. Results were compared by regression analysis and the mean difference between methods. An additional study was conducted to compare results for whole blood samples with those analyzed in the LH 750 system predilute mode. A total of 147 predilute samples were prepared by manual dilution of whole blood with an isotonic TABLE 9. Method Comparison: CBC, WBC Differential and Reticulocyte Parameters, Deming Regression Analysis, LH 750 Versus Gen S* Coefficient of Parameter Slope Intercept of Correlation WBC, 10 3 /µl RBC, 10 6 /µl Hgb, g/dl MCV, fl RDW% Plt, 10 3 /µl MPV, fl NE% LY% MO% EO% BA% RETIC% MRV, fl IRF, ratio *See Table 1 legend for abbreviation definitions.

7 Performance Evaluation of the LH 750 Hematology Analyzer 223 FIGURE 4. Regression plots for LH 750 complete blood count parameters. Gen S results (x-axis) are plotted against the LH 750 results (y-axis) for white blood cells (WBC) (A), red blood cells (RBC) (B), hemoglobin (HGB) (C), mean corpuscular volume (MCV) (D), red cell distribution width (RDW) (E), platelets (PLT) (F), and mean platelet volume (MPV) (G). diluent in a ratio of 1:5. Results were compared by regression analysis and the mean difference between modes. WBC Differential. The WBC differential results from the LH 750 analyzer were compared with results from the Gen S analyzer used as a predicate device. A total of 192 samples were analyzed. The comparison included WBC differential results for neutrophils, lymphocytes, monocytes, eosinophils, and basophils. Sample sets were eliminated from the calculations if any of the following flags were present on results from either analyzer: nonnumeric flag,

8 224 T. Fernandez et al TABLE 10. Method Comparison: Platelet and WBC Accuracy, LH 750 Versus Flow Cytometric Count* Population Range Coefficient of Parameters Number of Samples Low High Mean Diff Mean% Diff Slope Intercept Correlation Plt, 10 6 /µl Plt, 10 6 /µl WBC, 10 3 /µl *See Table 1 legend for abbreviation definitions. International Council for Standardization in Haematology/International Society for Laboratory Hematology platelet reference method. Flow cytometric WBC count. partial-aspiration flag, system-alarm flag, Verify Diff flag, low event-number flag and differential-suspect flag. Flagged samples should be followed up with a manual blood-film review or differential count. The WBC differential results were compared by regression analysis and the mean difference between methods. Reticulocyte Parameters. The reticulocyte results from the LH 750 analyzer were compared with results from the Gen S analyzer. A total of 383 samples were analyzed. In addition to the reticulocyte parameter, the comparison included parameter results for mean reticulocyte volume (MRV) and immature reticulocyte fraction (IRF). The data set for each parameter included only results from samples that were unflagged. Sample sets were eliminated from the calculations if any of the following flags were present on results from either analyzer: nonnumeric flag, partial-aspiration flag, system-alarm flag, Verify Retic flag, and Abnormal Retic flag. Due to differences in the mean volume of reticulocytes when one instrument reports 0% reticulocytes and another reports a small percentage that is greater than zero, results with 0% reticulocytes were also removed from analysis. The reticulocyte results were compared by regression analysis and the mean difference between methods. NRBC Parameters. The NRBC results from the LH 750 analyzer were compared with results given by the manual comparative method and a predicate method. The manual comparative method consisted of a visual microscopic NRBC analysis performed by 2 readers, in which each reader counts the number of NRBCs seen per 100 WBCs. The average count of the 2 readers is the manual comparative NRBC count. The data set (N = 55) included only results for which the LH 750 analyzer generated a NRBC count. The NRBC results were compared by regression analysis and the mean difference between methods. RESULTS Carryover Results for carryover are shown in Table 3. The carryover results for all parameters tested were well within the specifications for the system. Within-Run Precision Within-run precision results for CBC, WBC differential, and reticulocyte parameters are shown in Tables 4 and 5. The results for all parameters tested were well within the instrument specifications. Within-run precision results for predilute samples are shown in Table 6. There are no published specifications for predilute samples. FIGURE 5. Platelet accuracy for the LH 750 was examined using the International Council for Standardization in Haematology/International Society for Laboratory Hematology flow cytometry platelet reference method. A, Results of all samples. B, Comparison of the LH 750 for samples with platelet counts of <50,000/µL.

9 Performance Evaluation of the LH 750 Hematology Analyzer 225 FIGURE 6. The LH 750 accurate white blood cell (WBC) counts are compared with the flow cytometric WBC counts. Linearity Linearity plots are shown in Figure 3, with the expected value for each dilution on the x-axis and the average LH 750 analyzer parameter measurements on the y-axis. The results for all parameters tested were well within the instrument specifications shown in Table 7. Correlation coefficients for all parameters were >0.99. Method Comparison CBC Parameters. The CBC results from the LH 750 analyzer were compared with results from the Gen S system by regression analysis and the mean differences between methods. Results are shown in Tables 8 and 9, and the regression plots are shown in Figure 4. The specifications for accuracy for the LH 750 analyzer are stated as mean difference; all CBC parameters were well within these specifications. In addition, WBC, RBC, Hgb, MCV, Plt, and RDW showed coefficients of correlation of >0.98, and MPV showed a coefficient of correlation of >0.96. The improved platelet count accuracy was measured by regression analysis and the mean difference between methods, and the results are shown in Table 10. The regression plots for platelet counts are shown in Figure 5, with the counts obtained by the ICSH/ISLH platelet reference method on the x-axis and the LH 750 system results on the y-axis. Figure 5A shows platelet results for the entire operating range, with a coefficient of correlation of >0.99. Figure 5B shows platelet results for which the count obtained by the ICSH/ISLH platelet reference method was less than 50,000/µL. The performance of the LH 750 accurate WBC count was measured by regression analysis and the mean difference between methods, and the results are shown in Table 10. The regression plot for the WBC count is shown in Figure 6. The flow cytometric results are on the x-axis, and the LH 750 system results are on the y-axis. The coefficient of correlation for WBC was >0.92. Accuracy for the LH 750 system CBC parameters in the predilute mode was defined as the agreement between samples run in the LH 750 system predilute mode using a 1:5 dilution versus whole blood samples run in the LH 750 system standard analysis or whole blood mode. Estimates of agreement were made by regression analysis and mean difference analysis, and the results are shown in Table 11. WBC Differential. The WBC differential from the LH 750 analyzer was compared with that from the Gen S system by regression analysis and the mean differences between methods. Results are included in Tables 8 and 9, and the regression plots are shown in Figure 7. The specifications for accuracy for the LH 750 analyzer are stated as mean difference; all WBC differential parameters were well within these specifications. In addition, coefficients of correlation for neutrophils, lymphocytes, and eosinophils were >0.99; for monocytes, >0.97; and for basophils, >0.81. Reticulocyte Parameters. The reticulocyte results from the LH 750 analyzer were compared with the Gen S system results by regression analysis and the mean differences between methods. Results are included in Tables 8 and 9, and the regression plots for the reticulocyte parameters are shown in Figure 8. The specification for accuracy of the reticulocyte parameter for the LH 750 analyzer is stated as mean difference; the reticulocyte parameter was well within this specification, and the coefficient of correlation for reticulocytes was >0.90. There are no published specifications for TABLE 11. Method Comparison: CBC Accuracy in Predilute Mode, Dilution 1:5 Versus Whole Blood* Parameter Number of samples Mean Difference Mean% Difference Slope Intercept Coefficient of Correlation WBC RBC Hgb MCV RDW Plt MPV *No specifications are available. See Table 1 legend for abbreviation definitions.

10 226 T. Fernandez et al FIGURE 7. Regression plots for the LH 750 white blood cell differential parameters for neutrophils (NE%) (A), monocytes (MO%) (B), lymphocytes (LY%) (C), eosinophils (EO%) (D), and basophils (BA%) (E). Gen S results (x-axis) are plotted against LH 750 results (y-axis). MRV and IRF. The coefficient of correlation for MRV was >0.82 and for IRF was >0.88. NRBC Parameter. The NRBC parameters were compared by mean difference and regression analysis, and the results are shown in Table 12. The regression plots for the NRBC parameters are shown in Figure 9. In Figure 9A, the manual comparative NRBC counts are on the x-axis, and the LH 750 system results are on the y-axis. In Figure 9B, the predicate method results are on the x-axis, and the LH 750 system results are on the y-axis. Correlation with the manual comparative smear was >0.93, and correlation with the predicate method was >0.91. DISCUSSION The results of this evaluation demonstrate the LH 750 analyzer s performance for carryover, linearity, precision, and accuracy of the directly measured CBC, 6-part differential, and reticulocyte parameters. Our evaluation not only demonstrated the LH 750 analyzer s performance against specifications and predicate methods, but also confirmed its ability to provide very accurate results at clinical decision points. New technology available on this system insures accurate results and eliminates unnecessary follow-up. Using AccuCount technology, platelet counts on the LH 750 analyzer proved to be accurate over the full extended

11 Performance Evaluation of the LH 750 Hematology Analyzer 227 FIGURE 8. Results of the regression analysis for the LH 750 reticulocyte parameters are plotted against the Gen S results for reticulocyte percentage (RETIC%) (A), mean reticulocyte volume (MRV) (B), and immature reticulocyte fraction (IRF) (C). range of the system. Accuracy was further confirmed by comparison with the ICSH/ISLH reference flow cytometry [2] method for platelets. We included samples in the low platelet count range, which is of clinical importance for platelet transfusion decisions, and found very good performance. Other studies [4,6,7] confirm this finding. These studies also show a reduction in platelet flags on the LH 750 analyzer, allowing more results to be reported directly from the system. We demonstrated the accuracy of WBC counts on the LH 750 analyzer throughout the extended linearity range, up to /µl. Automatic correction of the WBC counts in the presence of interference provides WBC counts that are accurate compared with counts obtained using very specific flow cytometric methods. Other investigators have confirmed the performance of this enhanced WBC technology [8]. They also confirmed that this enhanced technology reduces incidence of flags and need for follow-up testing. Enumeration of NRBCs can also be performed by the LH 750 system. This new parameter is measured using VCS technology with advanced algorithms, without requiring additional reagents or reflex or repeat testing. Our studies showed good performance when we compared results from the LH 750 analyzer with results from both manual and automated methods. In conclusion, our studies showed that the LH 750 hematology analyzer provides accurate and reliable results over a wide range of normal and abnormal sample types. Increased linearity reduces the need to prepare diluted samples and TABLE 12. Method Comparison: NRBC Parameter* Population Range Coefficient of Parameter Number of Samples Low High Mean Difference Slope Intercept Correlation NRBC, % NRBC, % *NRBC indicates nucleated red blood cell. Manual comparative method. Predicate method.

12 228 T. Fernandez et al FIGURE 9. LH 750 nucleated red blood cell (NRBC) results are compared with the manual NRBC count (A) and a predicate method (B). reanalyze them. With advanced technology and algorithms, platelet and WBC results are very accurate, even in the presence of interference. NRBC counts are now available along with accurate WBC counts. Accuracy is insured and more results are reported. There is less need for time-consuming, labor-intensive, and costly confirmation of results. These enhancements on the LH 750 hematology analyzer decrease the need for operator intervention and increase overall laboratory efficiency. REFERENCES 1. Brown W, Kaplan S, Keeney M, Johnson K, Wolfe N, Chin Yee I. Workflow improvement with random access and enhanced flagging on the new Coulter LH 750 hematology analyzer. Lab Hematol. 2001;7: International Council for Standardization in Haematology Expert Panel on Cytometry and International Society for Laboratory Hematology Task Force on Platelet Counting. Platelet counting by the RBC/platelet ratio method: a reference method. Am J Clin Pathol. 2001;115: Schlenke P, Frohn C, Kluter H, et al. Evaluation of a flow cytometric method for simultaneous leukocyte phenotyping and quantification by fluorescent microspheres. Cytometry. 1998;33: Charie L, Harrison P, Smith C, et al. Accuracy in the low platelet count range: a comparison of automated platelet counts on Beckman Coulter high-volume hematology analyzers with the ISLH/ICSH platelet reference method. Lab Hematol. 2001; 7: Coulter Gen S Hematology Analyzer Reference Manual. Miami, Fla: Beckman Coulter, Inc. 6. Keeney M, Brown W, Chin Yee I. Comparison of platelet counts from the Coulter LH 750 hematology analyzer with counts from the ICSH/ISLH platelet reference method. Lab Hematol. 2001;7: Kaplan S, Johnson K, Wolfe N. Validation of low platelet counts from Coulter LH 750 and Coulter Gen S analyzers. Lab Hematol. 2001;7: Chin Yee I, Keeney M, Johnson K, Brown W, Wolfe N, Kaplan S. Leukocyte flagging rates of the Coulter LH 750 analyzer compared with the Coulter Gen S hematology analyzer. Lab Hematol. 2001;7: