Evaluation of the body fluid mode of automated hematology analyzer XN ‐series for extremely low peripheral white blood cell counts

Sir, The white blood cell (WBC) count remains low (0.000–0.500 × 109/L) in hematopoietic stem cell transplantation (HSCT), and its subtle changes are important indices of infection risk and success of engraftment and are useful in the management and treatment of the patients 1, 2. Confirmation of the WBC count by a manual method is necessary especially in the extremely low count range (≤0.100 × 109/L) below the limit of measurement of automated hematology analyzers. Newly developed automated hematology analyzers XN-Series (Sysmex Corp., Kobe, Japan) have a body fluid (BF) mode, as a new built-in function, in addition to the low WBC (LW) mode, which is intended for analyzing samples with low WBC counts 3, 4. The BF mode has been developed for high-accuracy assay of specimens with very low WBC counts like cerebrospinal fluid through the automatic background checking and the use of a sample volume about 10 times more as normally used in the whole-blood (WB) mode. The measurement principle of the BF mode for WBC and red blood cell (RBC) is flow cytometry and impedance methods, respectively. Reportable parameters in the BF mode are the total and differential (mononuclear cells: MN and polymorphonuclear cells: PMN) WBC counts, with the minimum display value at 0.001 × 109/L. The differential counts of neutrophils, lymphocytes, monocytes, and eosinophils are given as research parameters. When the analyzer is switched to the BF mode after analysis of whole blood, it performs automatic background checking before the measurement is started. Up to three such automatic checks are made until the WBC count comes down to ≤0.001 × 109/L and the RBC count to ≤0.003 × 1012/L. This minimizes the influence of carry-over and ensures accurate measurements of the total and differential WBC counts in the low count range. We evaluated the performance of the BF mode of the XN-2000 (XN) for measuring total and differential WBC counts of peripheral blood samples with low WBC counts. Samples used in the study were submitted to the clinical laboratory of Tokai University Hospital for a complete blood count test and were taken with addition of EDTA-2K as an anticoagulant. The study was approved by Institutional Review Board for Clinical Research of Tokai University Hospital. Within-run reproducibility in five replicates in each of the total and differential (neutrophils, lymphocytes and monocytes) WBC counts by the BF and LW modes on five samples with WBC counts ≤0.500 × 109/L was comparably studied. Within-run reproducibility of the total WBC by the BF was better than that by the LW mode, as reflected in smaller coefficient of variations (CV), which were 2.0–9.2% and 0.0–39.1% for the BF and the LW mode, respectively (Table 1). Even for a sample with the lowest WBC count (0.014 × 109/L), the CV with the BF mode was good at 9.2%. Within-run reproducibility of absolute counts and percentages of differential WBC were also better in the BF mode than the LW mode, as reflected in smaller CV, which were respectively 0.7–29.9% and 2.6–26.5% vs. 5.5–136.9% and 2.6–136.9% for neutrophils, 5.1–11.0% and 3.4–7.5% vs. 0.0–37.3% and 2.9–39.9% for lymphocytes, and 0.0–18.2% and 6.1–19.0% vs. 0.0–46.5% and 3.6–47.5% for monocytes. This is believed to be because of difference in precision arising from the difference in sample volume used in the BF and LW modes (respectively 10 and 3 times that used in the WB mode). However, we found some cases in which CV with the LW mode was better than the BF mode. For the BF and LW modes, the minimal reportable units of each leukocyte absolute count are 0.001 × 109/L and 0.010 × 109/L, respectively, and the LW mode gives the data by a unit of 0.010 × 109/L. Therefore, the difference in the reportable digit number might be concerned in such an inverted phenomenon. With the BF mode, we can expect a good CV of not more than 10% when the WBC count is 0.010 × 109/L or more. In samples with WBC count >0.100 × 109/L, the CV for the differential (neutrophil, lymphocyte and monocyte) counts was not more than 15%, which represented good reproducibility. The CV for total WBC with the BF mode similarly determined for five samples with WBC counts ≤0.100 × 109/L was 3.6–14.1%. They were not more than 15.0% in all the samples, including one with WBC count 0.010 × 109/L, and not more than 10% in all the samples except one (data not shown). Its within-run reproducibility was better than the manual counting, as reflected in smaller CV than 24.6–44.2% with the latter on three samples with WBC count ≤0.100 × 109/L, as determined using a standard procedure with Turk's solution and a Fuchs–Rosenthal chamber. In samples with extremely low WBC count ≤0.100 × 109/L, almost all the differential WBC counts were ≤0.016 × 109/L and their CV was in the range 8.7–32.3%, showing considerable sample to sample variation. The CV for absolute counts and percentages were respectively 8.7–17.9% and 4.7–17.5% for neutrophils, 3.7–27.2% and 2.9–21.8% for lymphocytes, and 21.5–32.3% and 18.0–32.4% for monocytes. Thus, generally the CV increased with decrease in the cell count. The BF mode showed assay linearity of WBC counting in the range of 0.009–0.948 × 109/L, when evaluated for a dilution series of two samples with initial concentration of 0.100 × 109/L and 1.000 × 109/L using the dilution reagent specific to XN (the CELLPACK) as a diluent (y = 1.003x−0.003 and y = 0.991x + 0.008, respectively). Although the BF mode can display results down to 0.001 × 109/L, the reliability of measurements including assay linearity in the even lower range of 0.001–0.010 × 109/L is to be ascertained. These results suggest that the performance of the BF mode in measuring the total and differential WBC counts should be taken into account when using such counts measured in the low count range. Method comparison of the BF and LW modes in measuring the total and differential WBC counts using 37 samples from HSCT patients showed good correlation coefficients (r) ≥0.940 (data not shown). As the reference for comparison, differential WBC counts were determined by the manual method, where 10–100 cells were counted in May–Grunwald–Giemsa-stained smears of samples having 0.003 × 109/L or more WBC counts. When absolute and differential WBC counting of 31 of these samples by the BF and LW modes were compared with the manual method, the former showed higher correlation with the manual method, the correlation coefficients (r) being respectively 0.970 and 0.840 for WBC, 0.963 and 0.866 for neutrophil, and 0.935 and 0.813 for lymphocyte (Figure 1a). Generally, the LW mode gave higher measured values than the manual method, suggesting a positive error. One reason for this may be that a background check is carried out to strictly prevent carry-over when the system is switched from the ordinary mode to the BF mode, which does not happen with the LW mode. Their difference in the minimum display value also appears to be a contributing factor. Monitoring the WBC and neutrophil counts of two patients under HSCT by the BF mode revealed changes, similar to those seen by the LW mode and the manual method, in these counts in response to treatment and the clinical course (Figure 1b). Among the three methods, the LW mode gave the highest values throughout. The BF mode showed changes in cell counts as closely as detected by the manual method. In WBC measurement in a low count range, such as of HSCT patients, very minute carry-over of cells can cause a problematic false high value. To ensure reliability of the measured WBC count, the manual counting is necessary depending on the previous results and the scattergram patterns. Manual methods, however, have drawbacks like considerable time and labor requirements and low assay precision in the extremely low count range. Although the LW mode had smaller CV for within-run reproducibility than in the manual method in the low count range, the falsely high measured values caused by carry-over in such ranges were an issue. Therefore, we believe that the use of the BF mode, which has better precision and accuracy in the low count range, would enable us to obtain more reliable total and differential WBC data than before. Because of its development for intended use for body fluids, the BF mode has a limitation in that it cannot eliminate the interference of nucleated RBCs in WBC counting, in contrast to the LW mode. Although nucleated RBCs are very rarely seen in HSCT patients, operators should examine manual counts in cases they judge the scattergram as having an abnormal pattern, as well as in those display an alarm message of the abnormal distribution scattergram, which indicates that WBC and non-WBC particles cannot be discriminated. With XN's automatic retesting function, retesting is possible in the LW mode according to the results of a first test 3. Although the BF mode can be set only manually, there is an in-built stringent automatic background check to prevent carry-over, enabling measurement with higher accuracy than the LW mode. The volume of sample aspirated in the BF mode is 88 μL, still a very small amount. Although the switchover takes about 1.5 min, the analyzer can be made ready to start measurements without sample preparation and change of the reagents. These two modes may be used selectively to exploit their features, or in combination, to assay samples with low WBC counts and to improve reliability of the test results and efficiency of laboratory testing by reducing retests. In conclusion, the BF mode of the XN-Series analyzer was capable of measuring extremely low WBC counts and differentials in peripheral blood samples, with better precision and accuracy than the LW mode. The BF mode would be useful in measuring the WBC and the neutrophil counts at the time when the WBC remains extremely low like at post-HSCT and could possibly replace the manual method. We thank Kazutoyo SAKAIRI, Noriko WADA, Kazumi GONDO, and Takayuki SETO in Clinical Laboratory of Tokai University Hospital for their contribution, by technical support, to performing the research in the measurement of clinical samples. We also thank Fumiaki Hayashi in Sysmex Corporation for his assistance in preparing the manuscript.