How to cite this article: Cherian S, Levin G, Lo WY, Mauck M, Kuhn D, Lee C, Wood BL. Evaluation of an 8-color flow cytometric reference method for white blood cell differential enumeration. Cytometry Part B 2010; 78B: 319–328.
Manual microscopy is the current reference method for white blood cell (WBC) differential counts. However, manual counts are time and labor intensive, difficult in patients with low WBC counts, and can misclassify cells having difficult morphology. We investigated an 8-color, single-tube, lyse no-wash flow cytometric method to perform an extended 8-part differential as a potential replacement reference method for WBC differential enumeration.
Whole blood was stained using a panel of antibodies including CD45APC-Cy7, CD16+CD19FITC, CD33+CD64PE-Cy5, CD123PE, HLA-DRPE-Cy7, CD34+CD117APC, and CD38A594 with the membrane permeant DNA binding dye Hoechst 34580 to generate an 8-part differential (lymphocytes, granulocytes, monocytes, eosinophils, basophils, immature granulocytes, blasts, and nucleated RBCs) using TruCount beads to generate absolute counts for all populations. Manual and instrument differentials were generated for 300 blood samples ranging from normal to complex. Results were compared with the flow cytometric differential.
Although manual microscopy (1) is the current reference method for white blood cell (WBC) differential counts, manual differential counts are time and labor intensive, and can be difficult in patients with low WBC counts. Manual differential counts provide particularly poor enumeration of infrequent populations due to the relatively small number of cells evaluated, suffer from nonuniform distribution of cells on the slide, and can misclassify cells having unusual morphology. Several groups have investigated various flow cytometric methods for performing a differential count (2–5). These groups utilized 3 to 5-color flow cytometric methods that did not allow for positive identification of all populations or enumerated only subsets of the total cell population. We investigated an 8-color, single-tube, lyse, no-wash flow cytometric method to perform an extended 8-part differential as a potential replacement reference method for WBC differential enumeration. This method has many potential applications including validation of new hematology automated analyzers and clarifying cases with difficult morphology. In addition, this method has the potential to supplement or perhaps ultimately replace either the manual differential or current methods used to perform instrument-automated differential counts in the clinical laboratory.
All samples were residual K3-EDTA anticoagulated clinical samples submitted for a complete blood cell (CBC) count and their use was approved by the University of Washington Human Subjects Review Committee (96-2538-V/E22). The study group included 102 random peripheral blood samples meeting criteria for acceptance of an automated instrument differential in the clinical hematology laboratory, 85 samples failing criteria for acceptance of an instrument differential (requiring technologist review or manual differential prior to reporting), and 113 samples collected at random from a hematology/oncology outpatient clinic. For each sample, a WBC and automated differential were obtained using a Sysmex XE2100 (Kobe, Japan) according to standard laboratory procedures, a manual differential was performed per CLSI H20-A2 (200 cell differential on each of two slides), and a flow cytometric differential was obtained as described below.
Whole blood (100 μL) was stained with 75 μL of a cocktail of monoclonal antibodies and 25 μL of a 100 μg/mL solution of the membrane permeant DNA binding dye Hoechst 34580 (Invitrogen, Eugene, OR) within 24 h of collection. Antibodies were obtained from either Becton-Dickinson (BD, San Jose, CA) or Beckman-Coulter (BC, Fullerton, CA) as indicated below. The antibody cocktail consisted of CD45 APC-Cy7 (BD, clone 2D1), CD16+CD19 FITC (BC, clones 3G8 and J4.119, respectively), CD33+CD64 PE-Cy5 (BC, clones D3HL60.251 and 22, respectively), CD123 PE (BD, clone 9F5), HLA-DR PE-Cy7 (BC, clone Immu-357), CD34+CD117 APC (BD, clones 8G12 and 104D2, respectively), and CD38 A594 (BD, clone HB-7). To reduce background in this no-wash assay, each antibody was titered in 100 μL total volume to optimize the signal to noise ratio, and maintain positive intensity near that seen at saturation, which generally resulted in their use below manufacturer's recommended concentration. Following incubation for 15 min at room temperature in the dark, 1.5 mL of buffered NH4Cl containing 0.25% ultrapure formaldehyde (Polysciences, Warrington, PA) was added and incubated for an additional 15 min at room temperature in the dark. The samples were prepared in TruCount tubes (BD) with accurate reverse pipetting of the blood to allow generation of an absolute count for all WBC populations. The samples were acquired on an LSRII (BD, San Jose, CA) with four lasers having excitation maxima at 407, 488, 594, and 635 nm. A goal was set of collecting 100,000 nucleated events per sample. For samples with a low WBC (less than 1 × 103 cells/μL), fewer nucleated events were collected with the number of nucleated events correlating directly with the WBC (R2 = 0.907). For samples with a low WBC, a range of 2,933–78,824 nucleated events were collected per sample (mean 38,374, median 33,686).
Among the nucleated cells, the antibody panel listed above was used as follows to generate an 8-part differential including lymphocytes, granulocytes, monocytes, eosinophils, basophils, immature granulocytes, blasts, and nucleated RBCs (Figure 1) using software developed in our laboratory (WoodList, BLW). First, monocytes were sequentially identified using CD33+CD64, HLA-DR, and CD16+CD19. Monocytes were then excluded, and of the remaining cells basophils and plasmacytoid dendritic cells were identified and then excluded using expression of CD123 and HLA-DR. Of the remaining cells, CD45 and side scatter were used to generate a rough lymphoid cell gate from which CD33+CD64 positive cells were excluded. The lymphocytes were then separated into B cells (CD19+CD16 and HLA-DR), NK cells (CD19+CD16 without HLA-DR), and T cells (lymphoid cells lacking CD19 and CD16), which were combined to give the lymphocyte enumeration. The method does not allow separation of NK/T cells expressing CD16 from true NK cells, but is expected to be result in only marginal error that is unlikely to be clinically significant. The lymphocytes were then excluded leaving mature neutrophils, immature granulocytes, eosinophils, blasts, and nucleated red blood cells. Mature neutrophils were identified on the basis of CD16 expression. Eosinophils were separated from immature granulocytes and blasts on the basis of side scatter, absence of CD16+CD19, presence of CD33+CD64, and high CD45 expression. Eosinophils were then excluded, and the immature granulocytes were isolated from the remaining cells on the basis of side scatter versus expression of CD33+CD64 with forward scatter low debris being excluded from the immature granulocyte gate. Blasts were identified using CD45 versus side scatter characteristics in conjunction with expression of CD34+CD117. Finally, nucleated red cells were isolated from the remaining nucleated cells on the basis of expression of DNA binding dye combined with low forward scatter characteristics induced by the lysing reagent. This method also allows identification of plasma cells (bright CD38) and mast cells (bright CD117). Plasmacytoid dendritic cells, mast cells, plasma cells, and in most cases nucleated red cells were present in too few numbers in the validation groups studied to adequately assess these parameters. The absolute concentration (cells per μL) of each population was determined in conjunction with enumeration of TruCount beads per manufacturer's recommendation, and the sum of the absolute count for all populations was used to generate the WBC count. Precision studies using this method generated a CV of 3.5–3.7 for populations present at greater than 1,000 cells per microliter, a CV of 7.1–8.8 for populations present at 8 to 50 cells per microliter, and a CV of 12.6 to 32.8 for populations present at less than eight cells per microliter. Furthermore, the incidence of doublet formation was not significant until the population of white blood cells exceeded 80,000 cells per microliter (data not shown).
Validation of the Flow Cytometric Differential
The flow cytometry differential was validated in a series of 102 random samples with characteristics allowing acceptance of the instrument generated WBC and differential. In this group, the mean WBC was 9.59 × 103 cells/μL with a range of 4.25–20.8 × 103 cells/μL. A comparison between the flow cytometric method and the Sysmex XE2100 generated WBC and differential showed a linear relationship for WBC, neutrophils, lymphocytes, monocytes, eosinophils, basophils, and immature granulocytes (Pearson correlation coefficients 0.961, 0.973, 0.985, 0.933, 0.957, 0.852, and 0.900, respectively) with a mild proportional bias to lower counts for monocytes (slope 0.8995) and higher counts for basophils (slope 1.185) and immature granulocytes (slope 1.144) by the flow cytometric method (see Table 1 and Figure 2). A comparison of the flow cytometric method with the manual microscopic differential revealed a strong linear relationship for neutrophils (Pearson 0.967) and lymphocytes (Pearson 0.926), a weaker linear relationship for monocytes (Pearson 0.812) and eosinophils, (Pearson 0.817) and a poor relationship for basophils (Pearson 0.29) and immature granulocytes (Pearson 0.378). Basophils showed decreased numbers by the flow cytometric method in comparison with morphology (slope 0.615) while flow cytometry generated a somewhat increased number for immature granulocytes (slope 1.172) as compared with morphology. The Sysmex and manual differential methods showed findings similar to those seen between the flow cytometry and manual methods (see Table 1).
Table 1. Validation of Flow Cytometry Differential: Uncomplicated Samples Meeting Criteria for Acceptance of an Instrument Generated Automated Differential
N = 101, 1 case with congenital absence of CD16 on neutrophills excluded
Flow cytometric method is designated as Y in the regression equations.
Sysmex XE 2100 method is designated as Y in the regression equations.
WBC (×103 cells/μl)
y = 1.0237x − 131.85
R2 = 0.9614
y = 1.0488x − 151.61
y = 1.0252x − 286.72
y = 0.9821x − 138.42
R2 = 0.9732
R2 = 0.967
R2 = 0.9919
y = 0.9854x − 28.8
y = 0.9839x + 8.6263
y = 0.9998x + 41.79
R2 = 0.9846
R2 = 0.9262
R2 = 0.9407
y = 0.8995x + 35.455
y = 1.0241x + 123.65
y = 1.1097x + 120.18
R2 = 0.9327
R2 = 0.8122
R2 = 0.7755
y = 0.9816x + 4.8984
y = 0.921x + 24.477
y = 0.9274x + 21.279
R2 = 0.9567
R2 = 0.8169
R2 = 0.8085
y = 1.185x + 2.6684
y = 0.6145x + 22.042
y = 0.4557x + 16.37
R2 = 0.8524
R2 = 0.29
R2 = 0.2384
y = 1.1437x − 9.2245
y = 1.1717x + 29.598
y = 0.9647x + 34.427
R2 = 0.9002
R2 = 0.378
R2 = 0.3718
The greatest systematic discrepancies between the flow cytometric differential and morphologic differential lay in enumeration of basophils and immature granulocytes. For both these populations, the correlation was similarly poor between both flow cytometry and Sysmex XE2100 enumeration as compared with morphology, while flow cytometry correlated well with Sysmex XE2100 enumeration of basophils and immature granulocytes. Both basophils and immature granulocytes were present in very low numbers in the samples evaluated in this group (basophils, mean 0.031 × 103 cells/μL; immature granulocytes, mean 0.053 × 103 cells/μL) therefore, the discrepancy in enumeration is favored to represent the ability of flow cytometry, and the Sysmex XE2100 to evaluate many more cells than is possible by microscopy.
One sample was excluded from analysis due to absence of CD16 expression on neutrophils. This sample showed similar numbers of neutrophils by morphology and instrument but showed essentially no neutrophils and a dominant eosinophil population by flow cytometric analysis. Examination of the smear showed many mature neutrophils with only few eosinophils (see Figure 3), and flow cytometry demonstrated an absence of CD16 expression on all myeloid cells leading to their classification as eosinophils. CD16 was expressed on NK cells in this case.
Performance of the Flow Cytometry Differential in Specimens Requiring Manual Review or a Manual Differential Prior to Reporting
The flow cytometry differential was then evaluated in a series of 85 samples with problems identified on initial instrument screening such as abnormal flags or white blood counts outside the linear range of the instrument (see Table 2). In this group, the mean WBC was 9.87 × 103 cells/μL with a range of 0.17–79.76 × 103 cells/μL. The correlation between the flow cytometric method and the manual differential shows a generally linear relationship for most populations (Pearson correlation coefficients of 0.989, 0.964, 0.939, and 0.997 for neutrophils, monocytes, eosinophils, and blasts respectively), with the exception of lymphocytes (Pearson 0.654), basophils (Pearson 0.135), and immature granulocytes (Pearson 0.746). The correlation coefficient was lower than expected for lymphocytes due to a single outlier from a patient with many nucleated red blood cells (18.1 × 103 cells/μL by flow cytometry). In this case, the instrument-generated WBC and nRBC count correlated well with that generated by flow cytometry, suggesting morphology was overestimating lymphocytes. When this outlier was excluded, the Pearson correlation coefficient was 0.916 for lymphocytes between the flow cytometric and manual methods. As seen in the prior data set, the correlation was poorest between the flow cytometric and manual method when evaluating cells present at low numbers including basophils and immature granulocytes. Similarly, blasts showed a proportional bias to lower values by flow cytometry, likely due to their presence at low numbers. Monocytes showed a proportional bias to higher values by flow cytometry in comparison with morphology (slope 1.652), suggesting some difficulty in accurately identifying monocytes by morphology in this patient population.
Table 2. Validation of Flow Cytometry Differential: Samples that Require Manual Slide Review
N = 85
Flow cytometry versus manual differential
In this group the WBC count generated by the instrument correlated well with that generated by flow cytometry (R2 = 0.9964, y = 1.0287x − 291.78). The flow cytometric method is designated as Y in the regression equation.
If the one sample with an nRBC count of 18.1 × 103 cells/μl by flow cytometry is excluded, the Pearson correlation coefficient is 0.9156 (y = 0.9156x + 23.263) for lymphocytes between the flow cytometric and manual methods.
Performance of Flow Cytometric Differential Versus the Manual Differential in Specimens Collected at Random from an Outpatient Hematology/Oncology Laboratory
The flow cytometry differential was then evaluated in a series of 113 samples collected at random from an outpatient hematology/oncology laboratory (see Table 3). In this group, the mean WBC was 9.48 × 103 cells/μL with a range of 0.04–105.10 × 103 cells/μL. The flow cytometric method and the manual differential showed a linear correlation for all populations except basophils (Pearson correlation coefficients of 0.996, 0.998, 0.839, 0.966, 0.137, 0.811, and 0.920 for neutrophils, lymphocytes, monocytes, eosinophils, basophils, immature granulocytes, and blasts, respectively). Basophils also showed a strong bias to lower values by flow cytometry (slope 0.457), likely due to their low numbers. Immature granulocytes showed a proportional bias to higher values by flow cytometry (slope 1.570) and somewhat reduced linear correlation (Pearson 0.811), also reflecting variably lower numbers. Monocytes also showed somewhat reduced linear correlation (Pearson 0.839), but without bias, suggesting some increased variability in how this population is identified by the two methods. In general, there was good correlation for blasts between the two methods (see Figure 4); however, in two samples, blasts were overestimated by morphology. Both of these samples were from patients who had history of a B cell lymphoma (chronic lymphocytic leukemia and mantle cell lymphoma) in which a subset of the atypical cells had dispersed chromatin and prominent nucleoli giving a blastoid morphologic appearance. Additionally, in two samples evaluated, eosinophils were overestimated by morphology. Upon review, slides from both these samples were noted to contain increased numbers of neutrophils showing toxic granulation and a myeloid left shift raising the possibility that some neutrophils with toxic granulation and/or immature myeloid forms were being misclassified by morphology as eosinophils. Another difficulty encountered in this clinical group was a poor correlation between the nRBC count generated by flow cytometry and the manual method when the nRBC count was greater than 0.5 × 103 cells/μL. In one case, the flow cytometric method gave a higher number (4,112 cells/μL vs. 2,107 cells/μL) while in the other case, flow cytometry gave a lower number (1,115 cells/μL vs. 1,642 cells/μL). An instrument-generated nRBC was available in former case (4,220 cells/μL) and showed better correlation with the number generated by flow cytometry.
Table 3. Validation of Flow Cytometry Differential: Samples From an Outpatient Oncology Clinic
N = 113
Flow cytometry versus manual differential
In this group the WBC count generated by the instrument correlated well with that generated by flow cytometry (R2 = 0.9979, y = 1.0323x + 119.73). The flow cytometric method is designated as Y in the regression equations.
y = 1.0425x − 131.98
R2 = 0.9955
y = 1.0331x − 109.86
R2 = 0.9976
y = 1.0537x + 145.02
R2 = 0.8392
y = 1.0207x + 1.573
R2 = 0.9661
y = 0.457x + 35.56
R2 = 0.1366
y = 1.5696x + 93.526
R2 = 0.811
y = 1.0359x − 16.456
R2 = 0.9202
Performance of Flow Cytometric Differential Versus the Manual Differential in Specimens with a Very Low WBC (Less Than 1 × 103 cells/μL)
In our entire cohort, 23 patients had a WBC of less than 1 × 103 cells/μL (mean 0.44 × 103 cells/μL, range 0.04–0.98 × 103 cells/μL). In this group, the white blood count as measured by flow cytometry correlated well with the instrument generated white blood cell count (R2 = 0.949, y = 0.9269x + 14.468). Manual differential counts in specimens with such low numbers of white blood cells can be tedious and time consuming. In this group, comparison of the flow cytometric method and the manual differential showed variably reduced correlation consistent with low cell numbers (Pearson correlation coefficients of 0.947, 0.967, 0.736, 0.770, 0.487, 0.128, and 0.973 for neutrophils, lymphocytes, monocytes, eosinophils, basophils, immature granulocytes, and blasts, respectively) (see Table 4).
Table 4. Validation of Flow Cytometry Differential: Samples with WBC Less Than 1.0 × 103 cells/μl
N = 23
Flow cytometry versus manual differential
In this group the WBC count generated by the instrument correlated well with that generated by flow cytometry (R2 = 0.9492, y = 0.9269x + 14.468). The flow cytometric method is designated as Y in the regression equations.
y = 0.7936x + 14.046
R2 = 0.9466
y = 1.0846x − 10.175
R2 = 0.9674
y = 0.9175x − 0.0845
R2 = 0.7359
y = 1.3757x − 1.7129
R2 = 0.7703
y = 1.0791x + 2.0207
R2 = 0.4865
y = 1.7691x + 8.3067
R2 = 0.1281
y = 1.2193x + 3.0741
R2 = 0.9733
We describe herein a novel single-tube flow cytometric method for generating a WBC count and 8 part differential that performs well in comparison to instrument and manually generated differential counts with both relatively normal and difficult patient samples including samples from patients with hematologic malignancies, patients recovering from chemotherapy, and patients with extremely low white blood cell counts. The flow cytometric differential showed the best correlation with morphology in enumeration of neutrophils, lymphocytes, mononcytes, eosinophils, and blasts. Poorest correlations were seen in enumeration of basophils and immature granulocytes. These latter populations were present in relatively low numbers in all studied groups, therefore, difficulties in identifying these two populations are likely due to bias by morphology as only 400 cells were counted to determine the morphologic differential. Similar difficulties in enumerating immature granulocyte (3, 5) and basophil (3, 5) populations by morphology have been reported in prior studies with improved correlation coefficients seen when selecting for patients with higher numbers of these populations (6).
The flow cytometric differential as described in this study, performs with similar correlation coefficients when compared to morphology as reported in prior studies (3, 5). Two recent methods have been published and validated for performing a flow cytometric differential. Faucher et al. (3) describe a “6 marker/5 color method” for performing an extended white blood cell differential by flow cytometry using the following combination: CD36-FITC/CD2-PE+CRTH2-PE/ CD19-ECD/CD16-Cy5/CD45-Cy7. Using this combination, various lymphocyte subpopulations, monocytes (CD16 positive and negative), neutrophils, myeloid precursors, eosinophils, and basophils are identified with similar correlation to that seen in the current study as compared with morphology. As compared with this prior study, our method is similar in that we use CD45 versus side scatter to guide initial gating strategies, CD16 to identify mature neutrophils, have the capability of distinguishing various lymphocyte subsets, and can identify CD16 positive monocytes. In addition, rather than using CD36 to identify monocytes, a marker that is not lineage specific and suffers from the effects of platelet adherence, we utilize a more specific strategy for monocyte identification (using markers including CD33, CD64, and HLA-DR). Further, we use a more specific method for basophil identification, CD123 without HLA-DR, which, in prior studies, has demonstrated great accuracy and precision in identifying basophils (7). The inclusion of CD123 with HLA-DR also allows identification of plasmacytoid dendritic cells, which have been noted to vary in the peripheral blood in various reactive (8–10) and malignant conditions (11). As compared with the method described by Faucher et al., the current method has the added benefit of using specific markers (CD34 and CD117) to identify blast populations thereby allowing a more definite blast enumeration, subject to certain constraints (see below). Finally, in this study, a DNA binding dye is included allowing more accurate identification of the nucleated red blood cells. However, in contrast to this prior reported method, the technique described herein does not use a marker for identification of T cells, instead relying on exclusion of B and NK cells within the lymphocyte population.
Bjornsson et al. (5) describe a single tube method to perform a 10-part differential applicable to either the peripheral blood or marrow using the following combination: CD36-FITC/CD203-PE/ CD138-PE/CD45-ECD/CD16-Pcy5+CD56-Pcy5 with DRAQ5 as a membrane permeant DNA binding dye. This strategy allowed identification of platelets as well as the following nucleated cell subsets: lymphocytes (with separation of NK cells), monocytes, neutrophils, immature granulocytes, eosinophils, blasts, basophils, and plasma cells. As compared with this prior study, our method is similar in that we use CD45 versus side scatter to guide initial gating strategies, CD16 to identify mature neutrophils, specific markers for isolating basophils, and have the capability of distinguishing NK cells from background lymphocytes. Our method also allows identification of plasma cells using expression of CD38, which is expressed at a very high level on plasma cells and is more stable than CD138 over time at room temperature (12). In contrast to this prior study, our method allows more specific identification of blasts using CD34 and CD117 and, as noted above, allows enumeration of the plasmacytoid dendritic cell subset. However, in contrast to the current study, Bjornsson et al. have optimized their method for use in the marrow as well as blood. The advantages of the current approach over either of these methods is the positive identification of more cell populations in a manner likely to provide more accurate enumeration in difficult samples, such as regenerating blood after chemotherapy and patients with hematopoietic neoplasia. However, a direct comparison of all three methods on a large number of suitable samples would be required to demonstrate this, and is outside the scope of the current study.
Two notable sources of discordance between flow cytometry and morphologic classification were identified in this study. First, two samples were noted in which atypical lymphocytes were misclassified as blasts by morphology. In both cases, the samples came from patients with a B cell lymphoma leading to leukocytosis. The first patient had chronic lymphocytic leukemia with increased numbers of prolymphocytes and in this case, prolymphocytes were misclassified as blasts by morphology. The second sample was from a patient with mantle cell lymphoma. As a subset of the neoplastic cells in this case had a blastoid appearance, these neoplastic lymphocytes were misclassified as blasts by morphology. Both these cases demonstrate pitfalls in the morphologic differential that are easily overcome by antibody driven specific cell identification methods characteristic of flow cytometry. The second area of notable discordance was in enumerating nucleated red blood cells by morphology as compared to flow cytometry. In one case, morphology gave a higher nucleated red blood cell count while in two cases flow cytometry gave a higher nucleated red blood cell count. The specific reason for these discrepancies was not clear but these discrepancies were seen in samples with high levels of circulating nucleated red blood cells (>500 cells/μL) and in the two cases where data was available, the flow cytometric nRBC count correlated better with an Sysmex XE2100 generated nRBC than morphology did.
One potential pitfall encountered in evaluating the flow cytometric differential using the described method relates to the use of CD16 for neutrophil identification. In one patient evaluated, the neutrophils were uniformly CD16 negative while CD16 was expressed on NK cells. This finding suggests the presence of a polymorphism leading the absence of CD16 expression on neutrophils (13). This uncommon polymorphism is estimated to be present in ∼1/1,000 in the Caucasian population. As CD16 is a GPI linked protein, paroxysymal nocturnal hemoglobinuria (PNH) would also lead to decreased CD16 expression on neutrophils. PNH like populations can be seen in some patients with MDS and aplastic anemia (14), therefore the described method may also encounter problems when attempting to assess specimens from patients with such diseases. Consequently, samples showing large number of eosinophils without many neutrophils should be considered suspect and evaluated further.
In addition, expression of some antigens we have incorporated into this assay have been described as having altered levels of expression in various clinical states. For instance, neutrophil CD64 expression has been shown to be upregulated in infection/sepsis (15) while CD16 expression has been noted to be decreased in patients with severe traumatic injury (16) and following high intensity exercise (17). The impact of such findings on the assay described herein is unclear and is beyond the scope of this current study. Future investigations will evaluate the use of the flow cytometric differential in various states of disease and health.
There are a few methodological issues that may improve the performance of the assay or allow its use on other instrument platforms. After the study was completed, it was noted that the use of 0.25% formaldehyde causes a modest decrease in the intensity of the two FITC labeled antibodies against CD16 and CD19. While reasons for this are not entirely clear, reduction of the formaldehyde concentration to 0.1% significantly improves their performance. It was also noted that the Hoescht dyes appear to cause a mild reduction in the intensity of the PE tandem fluorochromes for unclear reasons, and use of another class of permeant DNA binding dyes might improve assay performance. Finally, CD38 was coupled to Alexafluor 594, which requires a yellow laser for excitation, not usually present on clinical flow cytometers. However, use of CD38 on a second violet laser fluorochrome such as Pacific Orange is easily done and would make the assay amenable to use on a standard clinical instrument.
The identification and enumeration of blasts remains a potential problem for all assays described. Although the inclusion of CD34 and CD117 significantly improves the positive identification of normal or regenerating blasts, as well as most acute leukemias, a subset of acute leukemias are known to lack expression of these antigens, in particular acute monocytic leukemia and subsets of acute lymphoblastic leukemia commonly of T cell lineage. Cases of this type were not present within the current study population. The use of CD45 vs. SS gating in conjunction with the DNA binding dye and other antigens present in the cocktail would likely allow their enumeration, particularly when their numbers are elevated, but would require a modification to the described gating strategy. Further work to identify a reagent combination that is more generically useful for blast identification is warranted and is underway in our laboratory.
We describe herein, a novel, single-tube, flow cytometric method for performing a WBC count and 8-part differential that performs well with both relatively normal and difficult patient samples. This method compares well with previously described methods for determining a flow cytometric differential and, in part because it is adopted to a 10-color instrument, allows more specific identification of populations such as blasts and allows identification of additional populations such as plasmacytoid dendritic cells. In the future, we hope to continue validation of this method in different patient populations and to adapt it for use in bone marrow specimens.
The authors appreciate the assistance of the staff in the Hematopathology and Hematology Laboratories at the University of Washington and the Hematology Laboratory at the Seattle Cancer Care Alliance for their assistance in specimen procurement.