Improved flow cytometric method to enumerate residual cells: Minimal linear detection limits for platelets, erythrocytes, and leukocytes



Increasing demand for quality control of blood products requires more sensitive methods to enumerate residual cells. Presently, the reported threshold (in cells per microliter) is 400 for red blood cells, 30–500 for platelets, and 1 for leukocytes. To examine precision and linearity in enumerating residual platelets and red blood cells, EDTA-anticoagulated blood from healthy donors was serially diluted with serum, stained in TruCount tubes using a no-lyse/no-wash procedure and a monoclonal antibody cocktail against the CD42a (FL1) and glycophorin-A (FL2) epitopes, and analyzed by flow cytometry. Leukocyte counts were determined in separate tubes. Cell preparation and analysis were performed once for 20 blood samples each and 20 times using the same specimen. Acquisition from the same tube was performed separately for platelets (threshold on FL1) and red blood cells (threshold on FL2). Multiparameter analysis was used for data evaluation. Linear results were obtained for platelets per microliter between 3,410 and 5 and for red blood cells per microliter between 54,000 and 3. For the lower cell concentrations, the coefficient of variation was 16.7% for platelets and 10.9% for red blood cells. The presented method allows the distinction between physiologically intact and ghost red blood cells. The method represents a reliable, sensitive, and accurate approach to quantify platelets and red blood cells in diluted blood. It can be applied to enumerate residual cells in plasma products and meets the increasing demand for quality control in blood components. Cytometry (Clin. Cytometry) 50:231–237, 2002. © 2002 Wiley-Liss, Inc.

Residual leukocytes (WBC) and platelets (PLT) in plasma preparations (fresh-frozen plasma [FFP]) may enhance the risk of exposure to cell-associated pathogens like viruses or the agent of the variant of Creutzfeld Jakob Disease (CJD) (vCJD;1). During storage of blood components, WBC and PLT release cytokines and growth factors that can cause transfusion reactions in recipients (2, 3). In addition, there is some evidence that as few as 0.24 × 105 red blood cells (RBC) are sufficient to induce primary Rh alloimmunization (4). The use of cell-free plasma should avoid such complications after transfusion.

Currently, the Council of Europe guidelines require that FFP contains <100 WBC per microliter, <50,000 platelets per microliter, and <6,000 RBC per microliter (5). Although these cell numbers can be detected by conventional techniques, there is a strong need for improved technology for the accurate and sensitive quantification of residual cells. Fifteen years ago, the development of new apheresis systems like spinning membrane technology (Fenwal Autopheresis C) enabled the production of virtually cell-free plasma (6, 7). Increasing demand for quality control of these plasma products requires more sensitive and objective methods to monitor the efficiency of cell depletion. The regulatory background for preparing blood products is currently changing due to a possible transmission of the vCJD by blood components. As a consequence, the new European standard will most likely demand a reduction in leukocyte number to <1 × 106 cells per donation in PLT and RBC concentrates (5) and in RBC products. This can be achieved by the use of suitable procedures, such as by filtration methods (5). However, these filtration steps might affect the concentration of other residual cells, predominantly PLT (8), which will require advanced and more sensitive techniques for whole blood and plasma cell analyses.

Cell chamber counting represents a conventional, but subjective and time-consuming, method to measure low numbers of WBC, RBC, or PLT. To improve the lower limit of detection, samples have to be concentrated by low-speed centrifugation (9). Based on the addition of a known number of fluorescent beads, a flow cytometric single-platform technique was introduced recently for CD34 enumeration (10). It provides a reproducible and objective tool to count residual WBC, RBC, and PLT in blood preparations. Currently, the reported linear threshold (in cells per microliter) is 400 for RBC, 30–500 for PLT, and 1 for WBC (9–11). Lysis of disturbing RBC is necessary to reach a linear detection limit of 30 PLT per microliter (9).

We report a highly sensitive no-lyse/no-wash single-platform flow cytometric method for the simultaneous enumeration of residual PLT and RBC in diluted blood, but also in plasma products (unpublished data), or in other blood components, with a linear detection limit of 5 PLT and 3 RBC per microliter. In plasma products, the presented method allows the distinction between physiologically intact and ghost RBC.


Blood Samples

For the precision and linearity studies, the numbers of PLT, RBC, and WBC were determined in undiluted, EDTA-anticoagulated blood from healthy donors (n = 20) using an automated hematology analyzer (Advia 120, Bayer, Leverkusen, Germany). Thereafter, the blood samples were diluted serially, using either human serum (HS) (Biseko, Biotest Pharma GmbH, Dreieich, Germany) or cell-free fetal calf serum (FCS), from 1:10 to 1:1,968,000. The dilutions and the measurements by fluorescence cytometry were performed one time each for 20 blood samples and 20 times within 1 day using the same specimen. We used diluted blood due to the lack of plasma containing a standardized target cell number. In addition to these analyses, we examined human plasma products obtained from different collection procedures (unpublished data).

The Advia 120 is calibrated regularly by the appropriate standards. The linearity values (cells per microliter ± SD) for the measurement of the different cells types as provided by the company are 5–208 × 103 (±10 × 103) for PLT, 0.0–2.65 × 106 (±0.03 × 106) for RBC, and 0.02–21.8 × 103 (±0.46 × 103) for WBC.

Cytofluorometric Staining of the Cells

PLT, RBC, and WBC were stained in TruCount tubes (BD Biosciences, San Jose, CA), which contain a calibrated number of fluorescent beads. For immunologic labeling of PLT and RBC, 100 μl of the different blood dilutions were stained with 10 μl of a pretested monoclonal antibody (mAb) cocktail against CD42a (fluorescein isothiocyanate [FITC]; 1:10 in phosphate-buffered saline [PBS]-sodium-azide, BD Biosciences) and glycophorin-A (GPA) (phycoerythrin [PE]; 1:5 in PBS-sodium-azide, Dako, Glostrup, Denmark) using a no-lyse/no-wash procedure (10). After 30 min of incubation at 4°C, 300 μl PBS was added to provide a sufficient volume for fluorescence-activated cell sorting (FACS) acquisition.

WBC were quantified using the propidium iodide (PI)-containing LeucoCount reagent, according to the manufacturers protocol (BD Biosciences). To address the manufacturer's warning that the use of this reagent is not recommended with EDTA-anticoagulated blood, we compared serum-diluted blood samples (n = 5) drawn in EDTA and citrate tubes (S-Monovette Sarstedt, Nuembrecht, Germany). In addition, we examined EDTA and citrate-anticoagulated blood specimens (n = 25) from patients undergoing allogeneic bone marrow transplantation after myeloablative conditioning, whose leukocyte counts were below 100 WBC per microliter.

Flow Cytometry and Absolute Cell Count

Cell analyses were performed on a FACSCalibur (BD Biosciences). Depending on the amount of target cells (higher or lower), each measurement contained 20,000–35,000 total events, comprising predominantly beads and target cells. Acquisition from the same tube was performed independently for PLT (threshold on fluorescence 1 [FL1]) and for RBC (threshold on fluorescence 2 [FL2]). An additional threshold was set on side scatter (SSC) to remove debris (Fig. 1). This type of subsequent acquisition of one list-mode file for each cell type was mandatory to increase the sensitivity for detecting the cell type of interest (because the other cell type was not acquired). Logarithmic amplification was used for all parameters. WBC were acquired from a separate tube (threshold on FL2) as suggested by the manufacturer. Multiparameter analysis with the Paint-A-Gate software (BD Biosciences) was used for data evaluation (Fig. 1).

Figure 1.

Multiparameter analysis using the Paint-A-Gate software. PLT and RBC were quantified by combining the definite cell population gate in a FCS versus SSC dot plot, with the respective fluorescence gate in an FL1 versus FL2 dot plot. PLT (3) appear in FL1 (CD42a) and RBC (4) in FL2 (GPA). WBC events (2) were gated on FL2 in an FL1 versus FL2 dot plot. Fluorescent beads (1) appear in the upper right position of FL1 versus FL2 dot plots. PLT (threshold on FL1) and RBC (threshold on FL2) analysis was performed separately from the same tube. An additional threshold was set on SSC to remove debris. Green dots represent debris and nontarget events.

To determine the number of beads, a FL1 versus FL2 dot plot with a gate (Gate 1, blue dots) in the upper right position was painted (Fig. 1). The number of PLT and RBC, respectively, was quantified by combining the definite cell population gate (red dots) in a forward light scatter (FCS) versus orthogonal light scatter/SSC dot plot with the respective fluorescence gate (red dots) in an FL1 versus FL2 dot plot. PLT appear in FL1 (CD42a) and RBC in FL2 (GPA) (Fig. 1). WBC events were gated (red dots) on FL2 in an FL1 versus FL2 dot plot (Fig. 1). Based on bead number and sample volume, the absolute cell counts for PLT, RBC, and WBC per microliter were calculated from the absolute counts as follows: Cells per microliter = number of cell events × number of beads per tube/number of bead events × sample volume (μl).

Cell Sorting

From plasma products containing two different populations of RBC (detected by phenotype analysis), the respective populations were sorted into tubes (FACSVantage, BD Biosciences) and spun onto slides. RBC were then evaluated microscopically after May-Gruenwald staining.

Statistical Analysis

The coefficient of variation (CV) values were calculated from the mean and SD of 20 independent experiments for the same specimen. For comparison of PLT, RBC, and WBC numbers between the hematology analyzer and fluorescence cytometer, median and range values were calculated from experiments of 20 blood samples.


FACS Versus Hematology Analyzer

The cell counts calculated after FACS analysis of 1:100 (PLT and RBC) and 1:10 (WBC) blood dilutions differed <20% (2.6–19.8%, median 12.9% for PLT; 1.6–19.6%, median 10.5% for RBC; 1.6–11.4%, median 8.1% for WBC) from those obtained from undiluted EDTA blood analysis on the hematology analyzer (Table 1).

Table 1. Comparison of PLT, RBC, and WBC Numbers*
 Advia 120 Median (range) Cells/μlSample dilutionFACSCalibur Median (range) Cells/μlAdvia vs. FACS Median of difference (range) (%)
  • *

    Cell counts were determined from the original blood samples using an automated hematology analyzer (Advia 120) and from 1:100 (PLT and RBC) or 1:10 (WBC) blood dilutions using flow cytometric cell enumeration.

PLT2,660 (2,120–3,600)1/1002,509 (1,705–3,340)12.9 (2.6–19.8)
RBC45,400 (41,200–56,100)1/10039,783 (35,158–50,822)10.5 (1.6–19.6)
WBC571 (479–935)1/10636 (518–948)8.1 (1.6–11.4)

Regarding the warning of the manufacturer of LeukoCount not to use the reagent together with EDTA-anticoagulated blood, we found no significant difference in residual cell counts between EDTA and citrate-anticoagulated blood samples that had either been diluted with serum or were obtained from patients with <100/WBC per microliter of blood (data not shown).

Linearity of PLT, RBC, and WBC Counts by FACS

The expected PLT and RBC numbers were calculated from FACS results of 1:100 sample dilutions, the expected WBC from 1:10 dilutions (Table 1). When three to nine cells per microliter were expected, the observed cell counts from 20 experiments deviated between 8% and 100% for PLT and between 10% and 60% for RBC. Regarding WBC quantification, FACS counts from 0.6 to 1.2 WBC per microliter (n = 5) differed between 0% and 45% (n = 5) from calculated cell numbers (Fig. 2A).

Figure 2.

Deviation (A: percentage of expected cell numbers) and linearity (B: cells/μl) of PLT, RBC, and WBC counts from five experiments performed with different blood samples. Expected cell concentrations were calculated from FACS analyses of 1:100 sample dilutions (PLT and RBC) and of 1:10 dilutions (WBC). One representative experiment is shown for the linearity (B).

Linear results of flow cytometric cell measurements were obtained between 5 and 3,410 PLT per microliter (range: 5–14/1,705–3,410), between 3 and 54,000 RBC per microliter (range: 3–10/32,833–54,000), and between 0.7 and 636 WBC per microliter (range: 0.6–1.2/518–948). Whereas RBCs were identified clearly at the lowest cell concentration tested (three RBC per microliter), PLT became detectable at the fourth lowest cell concentration (five PLT per microliter, Table 2). Below this level, PLT events were not discriminated from background. Above this concentration, there was no overlap between the 95% confidence intervals (CI) of the neighboring measurements (10). A representative example of residual cell analysis is shown in Figure 2B. For the lower cell concentrations, a clear gate of the respective cell population (red dots) was defined and the definite events were 231 (CD42a-FITC) for seven PLT per microliter, 377 (GPA-PE) for nine RBC per microliter, and 62 (PI-positive) for one WBC per microliter (Fig. 3).

Table 2. Precision of RBC and PLT Counting by Flow Cytometric Cell Enumeration*
Expected cell number (calculated cells/μl)Measured cell number (N = 20)
Mean (cells/μl)SD (cells/μl)±2 SD (cells/μl)CV (%)
  • *

    Stable linearity was observed for five PLT and three RBC per microliter when one blood sample was diluted, stained, and analyzed independently 20 times. Above these concentrations, the 2 SD values of neighboring analyses did not overlap.

Figure 3.

Minimal linear detection of PLT, RBC, and WBC by fluorescence cytometry. Dot plots of FL1 versus FL2 for WBC and of FSC versus SSC for PLT and RBC are shown for residual cell analysis. At cell concentrations of one WBC, seven PLT, and nine RBC per microliter, a clear gate of the respective cell population was defined and appropriate numbers of absolute events were counted for cells (red) and beads (blue). Only one of several dot plots used for data evaluation is depicted for each cell type.

Precision of PLT, RBC, and WBC Counts at Low Cell Concentrations

Table 2 presents the variability of the established method for residual cell counting performed within 1 day. We calculated a CV of 16.7% for five PLT per microliter and of 10.9% for three RBC per microliter (Table 2) from 20 independent experiments using the same specimen.

Phenotype Analysis of Residual PLT and RBC in Human Plasma

Residual PLT and RBC in diluted EDTA blood samples show a typical appearance in the respective dot plots (red dots in Figs. 1 and 3). When analyzing residual cells in human plasma products and in platelet concentrates, we observed two RBC phenotypes. Apart from the typical phenotype of intact RBC (red dots), we detected a distinct second RBC population (cyan dots, Fig. 4). This cell type appeared at varying amounts in plasma and PLT preparations, but was the exclusive phenotype after freezing/thawing of plasma samples. Microscopic evaluation of sorted cells revealed that this phenotype represented RBC membrane covers without hemoglobin (ghosts). PLT did not change their phenotype during the procedure of human plasma preparation or during freezing/thawing.

Figure 4.

Intact versus ghost RBC in human plasma products. Dot plots of FSC versus SSC and of FL2 versus SSC show the two different RBC phenotypes detected in blood components. As shown by microscopic evaluation of FACS-sorted cells, the red and cyan cell populations represent intact RBC and ghost RBCs, respectively.


Development of apheresis technologies enabled the production of virtually cell-free plasma products (6, 7). Their quality safeguard requires a simple, objective, and accurate method to detect very low RBC, PLT, and WBC numbers. This study paper introduces a no-lyse/no-wash single-platform fluorescence cytometric method to determine residual PLT and RBC from one single tube, with a linear detection limit of five PLT and three RBC per microliter.

No-lyse/no-wash procedures provide a rapid tool to enumerate residual cells without losing cells due to lysing or washing steps. Whereas for quantification of WBC a commercial test kit based on PI staining is available with a linear detection limit of one WBC per microliter (LeucoCount kit, BD Biosciences), the reported linear threshold using a no-lyse/no-wash procedure is 400 RBC and 500 PLT per microliter (12). The roughly 100-fold higher linear sensitivity of the current method may be partially due to exclusion of nonspecific binding of mAbs (CD42a, GPA). The major point, however, is the independent acquisition of PLT and RBC from the same tube. Sequential acquisition of PLT and RBC, combined with a threshold set on FL1 for PLT and on FL2 for RBC, and the additional threshold set on SSC led to the acquisition of less debris and more events of the respective cell population. The fact that RBC were excluded when PLT were acquired increased the sensitivity of PLT enumeration and vice versa. An additional increase in sensitivity was obtained by multiparameter data evaluation using the Paint-A-Gate software. The fluorescence cytometric method described by Neumuller et al. (9) provides a good linearity of PLT enumeration. However, to reach a linear detection limit of 30 PLT per microliter, the authors suggest a lysing step to remove disturbing RBC. In contrast, our no-lyse/no-wash method allows simultaneous measurement of RBC and PLT without a possible cell loss due to washing steps and provides a threefold higher linear sensitivity for PLT quantification.

Regarding cellular components in plasma products, WBC and PLT are activated during storage and secrete cytokines and growth factors (2, 3), which may consequently cause transfusion reactions in the recipient (2, 13–15). In addition, PLT and mononuclear cells contain appreciable internal pools of prion proteins (PrPc;16, 17). Although it is not yet known whether or not the distribution of normal PrPc necessarily correlates with that of the abnormal isoform, scientific evidence suggests that the endogenous expression of PrPc is necessary for propagation of the CJD-causing PrPc isoform (18). A few studies reported the presence of infectivity in the blood of CJD patients (19–22) and of experimentally infected rodents (23–25). Although epidemiologic case–control and look-back studies have shown no evidence of clinical transmission of CJD by blood, blood components, or manufactured plasma derivates (26–31); it cannot be excluded that blood supplies might be contaminated with the vCJD agent (32). There is also evidence that as few as 0.24 × 105 RBC of rhesus-positive donors are sufficient for anti-D-production in rhesus-negative recipients (4). Primary Rh alloimmunization can occur after transfusion of low numbers of RBC, as well as after plasma exchange (31, 33). Therefore, the use of plasma products with the lowest possible content of residual cells is recommended. Although appropriate methods like the spinning membrane technology or filtration procedures (6, 7, 34) allow the production of virtually cell-free plasma products, a relatively high number of residual cells is tolerated by the guidelines of the European council (<100 WBC, <50,000 PLT, and <6,000 RBC per microliter). Due to recent adaptations of the quality requirements, the WBC content of whole blood components has to be reduced to less than 1 × 106 cells per donation (5) by a suitable procedure (e.g., by filtration methods). These filtration steps might effect the concentrations of residual cells, predominantly PLT in whole blood and in plasma prepared from leukocyte-reduced whole blood (8). The presented method provides an improved and highly sensitive FACS procedure for residual cell measurement. Linear detection with good precision of five PLT and three RBC per microliter provides a rapid, objective, and accurate tool to monitor the efficiency of cell depletion technologies. This method distinguishes physiologically intact from ghost RBC in plasma products as determined by cell sorting. With the exception of the latter, our method is a tool to quantify intact cells, but it cannot provide evidence about cell debris possibly produced during the preparation of blood components. For optimized quality control, this method should be combined with the analysis of biochemical cell parameters (data unpublished). Along with such biochemical analyses, our technique will support efforts to increase the quality of blood components.