Since preanalytic lysing of erythrocytes remains critical in flow cytometry, we investigated the influence of four lysing procedures on the quantification of leukocyte and CD34+ cells in hematopoietic cell transplants (HCTs). Samples were derived from stem cell–enriched mobilized whole blood collected by apheresis (unselected) and immunologically purified stem cell products (selected) and were measured using the dual-platform (2-PF) method with two flow cytometric systems. Additionally, cells were measured by a volume-based technique (single platform [1-PF]). Results were identical in the 2-PF mode (unselected HCTs, r = 0.998; selected HCTs, r = 0.999). In comparison with the 2-PF results, the single-platform (1-PF) measurements revealed a mean decrease of 59.5% for CD34+ cells (50.8% for CD45+ cells) in unselected HCTs and a mean decrease of 52% for CD34+ cells (49.8% for CD45+ cells) in selected HCTs. In order to check the accuracy of cell quantification using the 1-PF method, leukocyte reference values from hematology counter results were compared with flow cytometric (1-PF)–counted nucleated cells. That analysis revealed good congruency, with r = 0.998 for unselected HCTs and r = 0.999 for selected HCTs. In conclusion, all lysing procedures that we used induced substantial loss of leukocytes and CD34+ cells. As demonstrated by the high accuracy of the 1-PF technique, all erythrocyte lysing procedures caused significant cell loss, which led to inconsistent counting of CD34+ cells in nonvolumetric flow cytometric (2-PF) protocols.
The reconstitution of hematopoiesis after autologous or allogeneic stem cell transplantation (SCT) is performed by progenitor cells (PCs) from bone marrow (BM), umbilical cord blood (UCB), or peripheral blood (PB). PB has virtually replaced BM as the primary source of PCs for autologous and allogeneic transplantation because of the better availability of peripheral blood progenitor cells (PBPCs) and faster hematopoietic reconstitution. However, independent of the source of collected cells, engraftment potential correlates with both the integrity and the number of hematopoietic progenitor cells (HPCs). It is well known that unmanipulated HPC transplants from all sources are highly contaminated with leukocytes, platelets and erythrocytes. These populations may interfere with the accurate determination of HPCs. Accordingly, there is a demand for high-quality standards in the quantification of HPCs, as demonstrated by numerous established protocols and guidelines for determination of CD34+ cells [1–4]. The CD34-glycoprotein comprises a 110-kDa transmembrane protein with three domains and is expressed on the surface of nearly all undifferentiated and differentiated hematopoietic cell populations as well as on endothelial cells [5– 8]. Therefore, the enumeration of CD34-expressing cells is a well suited and established method for the quantification of HPCs. However, various external quality assessments have shown poor interinstitutional reproducibility for flow cytometric quantification of HPCs . Most of these multicenter studies investigated the interinstitutional variability in the determination of CD34+ cells according to common protocols with consensus values of CD34+ cells. However, the quantification of HPC loss during preanalytic procedures and the disturbance variables caused by non–flow cytometric systems (such as hematology counters [HCs]) were not the point of these studies.
From a technical view, two basic flow cytometric systems, known as dual-platform (2-PF) and single-platform (1-PF) techniques, are used for this purpose [10, 11]. Both systems need reference standards to calculate the correct number of cells. Whereas 2-PF devices require an HC result, the common 1-PF devices measure microparticles (beads) in parallel as an internal standard. The diversity of flow cytometric systems and protocols are reported to have a high impact on the interlaboratory variability of counting results [9, 12, 13].
Preanalytic sample preparation errors can also influence results. Although most protocols for quantification of CD34+ cells are standardized in terms of sample collection, storage, and staining procedures (including antibody clones, dyes, etc.), some other aspects remain critical [1, 2, 4, 5, 14, 15]. Erythrocyte lysing procedures have been identified as significantly influencing the detectability of immunolabeled cells. During the past few years, the use of fixatives and washing procedures were the main subjects of critical discussion [16, 17], but recent studies showed that detergents or high salt concentrations may also affect the integrity of HPCs . In single cases, differences of 100% for identical samples using different lysing reagents for the determination of HPCs in UCB were reported . In a recent study, we demonstrated that different erythrocyte-lysing procedures result in a mean loss of HPCs of 41.3% (range, 34%–45%) in the PB of G-CSF–mobilized patients . This loss was demonstrated by comparing the results of common 2-PF flow cytometry with those of volume-based 1-PF flow cytometry. Therefore, a high loss of HPCs can be verified by a volumetric measurement of identical samples and confirmed by the recovery of nucleated cells. During the selection and purification of HPCs by immunoadsorption techniques, the cellular environment changes from natural unmodified blood to citrate-modified blood and finally to citrate-containing phosphate-buffered solution. This may influence the stability of cellular components, such as HPCs. Thus, the above-mentioned breadboard construction for the analysis of whole blood samples was also performed in the present study to investigate the lysing-induced loss of HPCs in unselected and selected HPC apheresis products. Moreover, the present study was divided into two parts using  unselected PBPCs, which may simulate the influence of contaminating cells (e.g., leukocytes, red blood cells [RBCs], platelets) and plasma on the quantification of PBPCs after the erythrocyte-lysing procedure , and selected (purified) PBPCs without contaminating white blood cells (WBCs), RBCs, and platelets, in order to simulate the influence of lysing procedures specifically on the determination of PBPCs.
Materials and Methods
Collection of Stem Cells
The effect of erythrocyte-lysing procedures on the quantification of leukocytes and HPCs was investigated with two different flow cytometric techniques and systems. A total of 135 flow cytometric determinations of 10 unselected and five positively selected HPC products was performed. All samples were derived from representative aliquots of HPC products and were collected by apheresis (COBE-Spectra cell-separator; GAM-BRO-BCT, Martinsried, Germany, http://www.gambrobct.com) from patients undergoing autologous HPC transplantation or allogeneic transplantation from healthy volunteers. The positive selection of CD34+ cells was performed using the large-scale device CliniMACS (Miltenyi Biotec, Bergisch Gladbach, Germany, http://www.miltenyibiotec.com). Positive selection was provided if required for clinical application. Characteristics of all autologous and allogeneic HPC donors are given in Table 1.
2-PF measurements were performed using the FACSCalibur (Becton, Dickinson and Company, Franklin Lakes, NJ, http://www.bd.com) flow cytometer as the standard method (SM) and the particle-analyzing system (PAS) flow cytometer (Partec, Münster, Germany, http://www.partec.com) as the comparison method (CM) (30 determinations). The numbers of leukocytes (CD45+ cells) in the HPC products were measured by flow cytometry (first platform), and the total leukocyte concentration was measured by the HC (second platform). The absolute number of leukocyte subsets was calculated according to a formula given in the next section for CD34+ cells and described previously . The PAS device was also used for direct (absolute) volumetric determinations of CD45+ and CD34+ cell concentrations (1-PF, without beads) (60 determinations).
Determination of CD34+ and CD45+ Cells.
HPCs (CD34+ cells) were quantified by both the 2-PF and 1-PF methods after labeling with a monoclonal antibody specific for HPCA-2 antigen (class III antibody, clone 8G12; Becton, Dickinson and Company) conjugated with phycoerythrin (PE). For the detection of CD45-expressing leukocytes, cells were labeled with CD45 monoclonal antibodies conjugated to fluorescein isothio-cyanate (FITC) (clone J33; Immunotech, Luminy, France, http://www.immunotech.com). Measurements were performed according to the German consensus protocol established by the German Society for Transfusion Medicine and Immunohematology and the German Society for Hematology and Oncology. This protocol for the quantification of CD34+ cells is similar to the International Society for Hematotherapy and Graft Engineering protocol [4, 14]. The German consensus protocol was chosen as the reference method because it is the most common and accepted method for the determination of CD34+ cells in Germany. The laboratory involved in this study uses this protocol and has passed all its internal and external proficiency tests. In detail, 100 μl of the unselected or selected product was incubated with a CD45-FITC– and a CD34-PE–conjugated monoclonal antibody. After 15 minutes of incubation at room temperature in the dark, cells were fixed and lysed. In short, all CD45+ cells were gated in a side-scatter (SSC)/CD45+ plot. This gate was activated in a plot of SSC/CD34+ cells, and CD34+ cells were marked in a separate gate. Only events within both gates were considered as true CD34+ cells. Both gates were taken for a backgating in a forward-scatter (FSC)/SSC and SSC/CD34+ plot. In both cases, the CD34+ cells should be visible as an endemic population (Fig. 1). In summary, 50,000 events (leukocytes) were collected as recommended by the 2-PF method. Using the volumetric 1-PF method, 200 μl sample volumes were analyzed, corresponding to more than 50,000 events. Whereas in the volumetric 1-PF method, the concentrations of progenitor cells and leukocytes were directly given without further calculations. The concentration of CD34+ cells was calculated after the 2-PF measurement according to the following formula:
The number of CD34+ cells per μl (x) was calculated by taking the number of CD34+ cells measured using the flow cytometer, multiplying this by the absolute number of leukocytes measured using the HC, and dividing by the number of CD45+ cells measured using the flow cytometer (50,000 [total measured events] − number of CD45+ cells).
Nuclear DNA was stained with a commercially available dye as a ready-to-use staining solution (CyStain UV; Partec). The buffer contains 4′,6′-diamino-2-phenylindole (DAPI) and is able to fluorescently label nucleated cells in a one-step procedure. We added 0.9 ml of the staining solution to 0.1 ml of each sample to be analyzed. Samples were measured volumetrically with the PAS flow cytometer directly after the addition of the staining buffer (45 determinations). DAPI-containing nucleated cells were fluorescently excited by ultraviolet light derived from a mercury high-pressure lamp (Osram Sylvania, Munich, Germany, http://www.sylvania.com) .
Determination of Leukocytes
The Sysmex K-1000 HC (Sysmex, Kobe, Japan, http://www.sysmex.co.jp) was used as a reference method (RM) for counting the leukocytes in unselected and selected HPC products and as the second platform method for the 2-PF evaluation with both flow cytometric devices. The manufacturer of the HC indicates a maximum deviation of 3% from the accurate value and precision with a maximum coefficient of variation of 3% for leukocytes in the range of 1.0–99.9 × 103 leukocytes/μl. The accuracy of the HC was regularly subjected to internal and external proficiency testing. In addition, a recent study proved a high correlation for counting results of leukocytes using this HC and the flow cytometer that we used .
The influence of four different commercially available erythrocyte-lysing reagents on the recovery of HPCs and leukocytes was investigated in this study. We compared FACS-Lysing Solution (Becton, Dickinson and Company) (lysis I, a lysis agent based on ethylene glycol), Ortho-mune (Ortho–Clinical Diagnostic Systems, Neckargmünd, Germany, http://www.orthoclinical.com) (lysis II, a lysis agent based on ammonium chloride), Uti Lyse (DakoCytomation, Glostrup, Denmark, http://www.dakocytomation.com) (lysis III, a lysis agent not further specified), and Cylyse (Partec) (lysis IV, a lysis agent based on a detergent). The erythrocyte-lysing reagents were applied to identical samples and were used according to the manufacturers' recommendations.
Identical samples treated with four different lysing reagents were measured. The results of two flow cytometry systems and the HC were compared. Pearson correlation coefficients and paired t-tests with a confidence level of 95% (p < .05) were used for comparison. The regression coefficient was calculated for the comparison of both 2-PF methods. Statistical analyses were performed using SPSS 12.0 software (SPSS, Chicago, http://www.spss.com).
The influence of different erythrocyte-lysing procedures on the concentration of HPCs from unselected and selected apheresis products was investigated. The flow cytometric analysis and gating strategy are exemplarily explained using the measurement of an unselected stem cell product (Fig. 1). The following parameters were analyzed and compared.
Enumeration of CD34+ Cells in the 2-PF Mode
For the unselected samples, a high consistency between the two devices was demonstrated, with a correlation coefficient of r = 0.998 for the range 334–7,712 CD34+ cells/μl (Fig. 2A). The same high consistency was shown for the selected samples (r = 0.999) for the range 2,959–17,425 CD34+ cells/μl (Fig. 2B).
There was no statistically significant difference between the two determinations in unselected (p = .075) and selected (p = .409) products (Wilcoxon test). In conclusion, both systems resulted in identical values for both groups when running in the 2-PF mode. Therefore, it was concluded that the comparison method (PAS) was comparable with the standard method (FACSCalibur) in the 2-PF mode.
Enumeration of CD34+ Cells in the 1-PF Mode
Comparing the 2-PF results with the 1-PF results, the mean loss of CD34+ cells was 59.5% for unselected apheresis products. Exact figures were: lysis I, 47% (standard deviation [SD], 5.8%); lysis II, 45% (SD, 14.8%); lysis III, 46% (SD, 6.3%); lysis IV, 100% of the 2-PF results (Fig. 3A). In selected apheresis products, the mean loss of CD34+ cells was 52%. Exact figures were: lysis I, 36% (SD, 10.8%); lysis II, 35% (SD, 8.8%); lysis III, 37% (SD, 1.9%); lysis IV, 100% of 2-PF results (Fig. 3B). The mean absolute numbers of CD34+ cells/μl and SDs are given in Table 2.
Determination of Leukocytes by Flow Cytometry and by HC
Similar to the kinetics of CD34+ cell loss found by comparing the results of the 2-PF and 1-PF enumerations, CD45+ cells decreased in a like manner (compared with the reference HC values), with a mean loss of cells in unselected products of 50.75%. Exact figures were: lysis I, 33% (SD, 3.1%); lysis II, 34% (SD, 9.0%); lysis III, 36% (SD, 4.2%); lysis IV, 100% (Fig. 4A). The selected products, with a purity of HPCs of 99.7%, had a mean CD45+ cell loss of 49.75%. Exact figures were: lysis I, 33% (SD, 8.8%); lysis II, 32% (SD, 6.3%); lysis III, 34% (SD, 4.6%); lysis IV, 100% (Fig. 4B). The mean absolute numbers of CD45+ cells/μl and SDs are given in Table 3.
Determination of Nuclear-Stained Leukocytes by the CM and HC
In order to verify the volumetric counting method, aliquots of unselected and selected products were fluorescently labeled with DAPI, and the nucleated cells were counted using the CM. Total numbers of leukocytes were measured in parallel using the HC and CM. A high consistency between the two methods was shown, with a correlation coefficient of r = 0.998 (p = .098) for the unselected products and r = 0.999 (p = .810) for the selected products.
In summary, the following results were obtained:
If both flow cytometric systems (FACSCalibur and PAS) were used in the 2-PF mode, results for the enumeration of CD34+ cells were identical (unselected cell products, r = 0.998, p = .075; selected cell products, r = 0.999, p = .409).
In relation to the standard system (FACSCalibur 2-PF), the 1-PF method demonstrated a mean loss of 59.5% (SD, 1.0%)—including one lysing reagent (lysis IV) with a loss of 100%—for CD34+ cells in unselected stem cell products. The same was true for selected products, with a mean loss of 52% (SD, 1.5%)—including one lysing reagent (lysis IV) with a loss of 100%.
If the same unmanipulated (unlysed) samples were counted with the HC in parallel with the CM after nuclear staining with DAPI, the results were identical: r = 0.998 (p = .098) for unselected products and r = 0.999 (p = .810) for selected products.
The success of reconstituting the hematopoietic system after myeloablative therapy and subsequent SCT is closely related to the administration of sufficient and functioning HPCs. Traditionally, these parameters have been evaluated using the colony-forming test, but these methods are handicapped by a lack of reproducibility and a long assay time [1, 21]. During the past decades, flow cytometry has earned an outstanding role in the enumeration and characterization of HPCs. Numerous protocols have been developed for flow cytometric quantification of CD34+ cells in different compartments [2, 4, 5, 14]. These protocols and guidelines have distinctly improved intra- and interlaboratory reproducibility (coefficient of variations), but some critical issues still remain. Most protocols disregard the influence of such sensitive and aggressive protocol steps as cell fixation, centrifugation, and the lysing of erythrocytes, which may have a high impact on cell enumeration [17, 19, 21, 22]. The last sample treatment, in particular, results in a loss of accuracy that has not been investigated in as much detail as has been done for the precision and sensitivity in the determination of CD34+ cells. Furthermore, most studies focus on mobilized whole blood and waive the modified conditions during the HPC purification process. Therefore, this study represents, for the first time, a systematic investigation of pre- and perianalytic cellular loss in unmanipulated apheresis and purified HPC products.
This was done by comparing the results of a volume-based 1-PF flow cytometer (CM), measuring cell count in a defined volume, with the results from the most often used system in this area, known as the 2-PF flow cytometer (SM) (set as 100%) . The 2-PF SM was performed according to the established consensus protocol for the determination of CD34+ cells . The same 2-PF calculations were performed using results from the CM, by the inclusion of HC results. This procedure thus offered a good way to validate the accuracy of the CM measurements. Both devices gave identical results (unselected products, r = 0.998; selected products, r = 0.999) when running in the 2-PF mode, but results differed distinctly when the CM was used in the volumetric 1-PF mode. The differences cannot be a result of false volumetric counts because aliquots of the unselected and selected products were counted with the reference method (HC) and the results were compared with those of the CM after nuclear staining. The high congruency of both methods revealed the reliability of the CM. Furthermore, the reliability and accuracy of volumetric 1-PF flow cytometry in counting leukocytes and immunolabeled cells has already been shown in other trials [20, 24].
The CM technique was already used in a previous study to quantify lysing-induced cell loss in G-CSF–mobilized whole blood . Compared with mobilized blood (loss of CD45+ cells, 34.5%), the stem cell products revealed a higher loss of leukocytes (50.75% in unselected products, 49.75% in selected products) and a higher loss of CD34+ cells (mobilized blood, 41.3%) in unselected products (59.5%) and in selected products (52%). The results of lysis IV were included in these calculations, but by using this lysing procedure, the cells (i.e., CD34+ cells and CD45+ cells) were no longer detectable. Interestingly, this total loss was not observed in the previous study using mobilized whole blood .
Unselected and selected stem cell products include different fractions and populations. In the latter group, cells are out of their natural environment. This, and the addition of citrate as an anticoagulant in apheresis products, may be responsible for the greater sensitivity of selected CD34+ cells to particular components of lysing buffers, and results in a complete loss of immunologic detectability when using detergent-based lysing procedures (lysis IV). The cell destructive properties of citrate in combination with detergents have already been described and used in another context to isolate nuclei from cells or tissues . However, the mechanisms of the reduced expression or complete absence of formerly detectable CD34 and CD45 antigens after erythrocyte lysing have not yet been resolved. The erythrocyte-lysing procedure probably affects membrane integrity . This supposition is also supported by a previous study on the viability of neutrophils, showing a disturbed plasma membrane in more than 90% of cells, as indicated by cellular influx of propidium iodide after the erythrocyte-lysing procedure . Several publications have critically discussed the lysing process as the main disturbing factor in the accurate quantification of CD34+ cells [17–19, 27]. Some authors recommend the use of no-lyse protocols [22, 28], whereas others prefer to dispense with a lysing procedure, especially if pure HPC products are counted . However, all these studies agree that erythrocyte lysing has a high impact on the enumeration of HPCs, and this is underlined by the results of this article.
We have recently established a reliable protocol to count immunolabeled lymphocyte subsets like CD4+ T-cells in unlysed blood samples [29, 30]. We found 10% lower numbers of CD4+ T cells in lysed blood samples than in the same no-lyse blood samples. Therefore, we recommend the use of a simplified but reliable no-lyse protocol that may also be suitable for quantifying CD34+ cells. Because of the high practical and clinical consequences of lysing effects on the determination of CD34+ stem cells in transplants, this protocol is currently under investigation in a larger comparative study with other established protocols.
In conclusion, both unselected and selected HPC products are strongly influenced by lysing procedures. Moreover, it was found that the change in HPC environment from blood to a buffer-based solution induces a higher susceptibility to single components of lysing reagents. The present results give an explanation for the high interlaboratory variability found in different studies and confirm the accuracy problems of the 2-PF method. This has to be taken into account for flow cytometric analysis and for further guidelines focused on the determination of CD34+ cells in different compartments.
The authors indicate no potential conflicts of interest.
Table Table 1.. Overview of patient data
Table Table 2.. Absolute number (±standard deviation) of CD34+ cells/μl in unselected and selected hematopoietic progenitor cell products
Table Table 3.. Absolute number (±standard deviation) of CD45+ cells/μl in unselected and selected hematopoietic progenitor cell products
We thank Gerlind Bellmann, Alexandra Mertens, and Nikola Gözelmann for excellent technical assistance.