Journal of Immunological Methods 338 (2008) 5862

Contents lists available at ScienceDirect

Journal of Immunological Methods

j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / j i m

Research paper

Determination of blood leukocyte concentration with constant volume acquisition on a ow cytometer is comparable to individualized single platform testing with beads as internal reference standard
Susan Hansen, Ronald Dahl, Hans Jrgen Hoffmann
Department of Respiratory Diseases, Aarhus University and Hospital, Building 2b, Nrrebrogade 44, DK 8000 Aarhus C, Denmark

a r t i c l e

i n f o

a b s t r a c t
Flow cytometers have a constant ow rate. This enables ow cytometers to measure leukocyte concentrations in a determined volume by acquiring data at a xed rate over a xed time and is called constant volume acquisition (CVA). The volume aspirated by a FACS Calibur ow cytometer in 4 min at a high rate has a median of 163 l (IQR 156170) with TruCount tubes. Leukocyte concentrations of 26 healthy volunteers were measured twice on up to four occasions with a BrkerTrk chamber, by single platform technology (SPT) with TruCount tubes and on the same data set using CVA. Total leukocyte concentrations determined by CVA correlated better with measurements in a BrkerTrk (BT) chamber than with SPT. Concentrations determined with CVA were 1.86% higher than with BT whereas SPT data were 5.35% higher than BT (p b 0.001), and 3.36% higher than CVA (p b 0.001). At leukocyte concentrations b 6 million/ml SPTcorrelated better with BT than CVA. The SPT measurement may be more variable because it depends on measurement of the number of beads aliquoted, the number of beads and leukocytes aspirated, where both BT counting and CVA measurements only depend on the number of leukocytes counted. CVA with PanLeukoGating can be established using microscopy as a reference, and is comparable to BT chamber and SPT determination. Leukocyte concentrations can be measured with CVA on ow cytometers in research and clinical settings. 2008 Elsevier B.V. All rights reserved.

Article history: Received 27 April 2008 Received in revised form 4 July 2008 Accepted 10 July 2008 Available online 3 August 2008 Keywords: Flow cytometry Single platform testing CD45 PanLeukoGating

1. Introduction Blood leukocyte concentrations are an essential complement to the fractional changes of leukocyte subsets. The best known blood leukocyte parameter is the CD4 concentration during infection with the human immunodeciency virus (HIV) that becomes critical when it decreases below 200 cells/ l (Yarchoan et al., 1991). We have recently investigated the frequency of regulatory T cells (Hoffmann et al., 2007) and changes in basophil granulocyte, regulatory T cell (T reg) and plasmacytoid (pDC) and myeloid dendritic cell subsets during

Abbreviations: BT chamber, BrkerTrk chamber; CVA, constant volume acquisition; SPT, single platform testing. Corresponding author. Department of Respiratory Diseases, Aarhus University and Hospital, Building 2b, Nrrebrogade 44, DK 8000 Aarhus C, Denmark. Tel.: +45 89492107; fax: +45 89492110. E-mail address: hans.jurgen.hoffmann@ki.au.dk (H.J. Hoffmann). 0022-1759/$ see front matter 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.jim.2008.07.010

allergen provocation (Hoffmann et al., 2006), and used a two platform method to determine absolute leukocyte changes. Single platform determination of leukocyte concentration has been developed and is still being rened for monitoring of HIV status (Schnizlein-Bick et al., 2000; Martini et al., 2007). Typically, beads are added to the blood sample before staining, and a lyse-no-wash method is employed (Nicholson et al., 1997). However, even though only one tube is employed, three measurements are made to calculate leukocyte concentration by this method; measurements are based on determination of numbers of leukocytes, aliquoted beads and acquired beads. Each measurement comes with some error, which is compounded as the terms are multiplied together. The present generation of ow cytometers has stable ow and constant volume (Bergeron et al., 2003). The PanLeukoGating procedure using CD45 (Glencross et al., 2002) has been developed to calibrate ow cytometers to analyse leukocytes in a constant volume dened by a constant acquisition time (Storie et al., 2003; Walker et al., 2006).

S. Hansen et al. / Journal of Immunological Methods 338 (2008) 5862

59

We adapted this method to our project, investigating ux of leukocyte subsets due to the allergen injection in specic immunotherapy (Bousquet et al., 1987), and compared leukocyte concentrations determined by SPT measurement with constant time acquisition and measurement in a BrkerTrk chamber as our predicate method. 2. Materials and methods

brightly uorescent polystyrene beads were used to determine the absolute concentrations of leukocytes, by using the formula: Total cell concentration = (number of cell events counted pr tube number of beads counted pr tube) / (total number of beads in tube sample volume pr tube). According to manufacturer's suggestions, bead doublets were included in the bead gate. 2.5. Statistics

2.1. Participants The Ethics Committee of Aarhus County approved the project. Adult patients (n = 20) to be treated with venom injection for wasp allergy by a standardized protocol (Bousquet et al., 1987) and controls not treated with allergy vaccination (n = 6) were recruited. Two patients withdrew because of lack of interest. Cells were counted in 4 ml heparinised blood before SCIT and 30 min after the last venom injection at the 1st, 4th, 7th visits and at maintenance dose. In controls, cells were counted at similar intervals. For each of 18 patients and 6 controls, two blood samples were analysed at each visit. Each participant visited the clinic up to four times, at intervals of 47 weeks. Patients were seen on 44 days from January to October 2007. 2.2. Microscopy Blood leukocyte concentrations were determined by microscopic counting, using a BrkerTrk counting chamber (0.0025 mm2, 0.5 l). 475 l dilution uid was mixed with 25 l heparinised blood and approximately 2 glass pearls were added. The dilution factor was 1/20. The blood leukocyte concentration can be calculated as number of leukocytes counted 0.04 cells/ml. 2.3. Flow cytometry Total leukocytes and T cells were measured with TruCount tubes (Catalogue Nr 340334) according to manufacturer's instructions. Heparinised blood (50 l) was labelled within 1 h of venipuncture with titrated amounts of antibodies. To identify leukocytes CD45 PerCP (10 l, Catalogue Nr 345809) was used in a tube designed to identify and determine the concentration of leukocytes (CD45+), plasmacytoid dendritic cells (CD45+Lin1CD123+HLA DR+) and basophil granulocytes (CD45+Lin1CD123+HLA DR) using Lin1 FITC, CD123 PE and HLA DR APC (Hoffmann et al., 2006). FACS Lysing Solution (450 l) was added to the TruCount tubes for 15 min, and the cells were counted with a FACS Calibur for 4 min at a high ow rate acquisition using Cell Quest Software. High ow rate analysis gives the lowest intra-assay CVs for cell counts (Storie et al., 2003). Compensation controls were generated each day with Comp beads (Catalogue nr 51-90-9001291) and all antibodies. All reagents and equipment for ow cytometry were from Becton-Dickinson, San Jose, CA, USA. 2.4. Data analysis Data were compensated with compensation beads prepared on each day, and analysed in batch with FlowJo version 8.2 (Fig. 1). TruCount tubes containing a known number of
Fig. 1. Representative gating procedure for determining bead and leukocyte numbers. Fresh heparinised blood was labelled with titrated amounts of CD45 PerCP in a TruCount tube, and was incubated and lysed with BD lysis reagent. The sample was acquired with a FACS Calibur ow cytometer at high speed set to trigger on PerCP for 4 min.

Counting uncertainty was assumed to equal the square root of the number of counted events. Statistical signicance of difference between results was assessed using a paired ttest if data was normally distributed, or a Wilcoxon Signed Rank test. p b 0.05 was considered statistically signicant. Association between leukocytes measured with microscope and ow cytometer analysed with the Pearson's Correlation Coefcient (SPSS 11.0.0, 2001). Missing data was excluded from the analyses. 3. Results 3.1. Calculation of constant volume acquisition (CVA) volume The mean rate of bead acquisition was 63.34 3.15 beads/s (CV = 4.97%) and the median volume aspirated in analysis of leukocytes was 163 l (IQR 156170 l). The aspirated volume contained 14.8 l blood. 3.2. Reproducibility of determinations by microscope, SPT and CVA The reproducibility of repeated BT counting (n = 7) and SPT enumerations (n = 2, four repeats acquired consecutively on the same day) was determined on blood of independent

60

S. Hansen et al. / Journal of Immunological Methods 338 (2008) 5862

Table 1 Comparison of statistical strength of the measurements Method Microscopy TruCount, bead number CVA and SPT, cell number Material Donor 1 Donor 2 Beads Donor 3 Repeats 7 7 4 4 Number of events 290 129 14,160 89,712 SD 16 3 454 5669 CV (%) 5.52 2.33 3.21 6.32 100/n 5.87 8.80 0.84 0.33 Leukocyte concentration 11,611 5142 6239

Vastly different numbers of events were counted when counting cells by microscopy, and TruCount beads and cells in the ow cytometer. The effect of this difference on precision is illustrated here; the counting method and material counted are listed, and the number of repeats, and the mean number of events counted with each method are given. The CV is calculated from the observed variation, and the theoretical CV is calculated as 100/n. The leukocyte number determined is also given, as this may inuence the actual number of events evaluated.

persons (Table 1). The observed precision was comparable for all measurements, but was less than expected from statistical theory. 3.3. Validation of SPT and CVA algorithms against microscopy CVA and SPT were compared to BT by linear regression; CVA = BT 0.9527 + 349,455 cells/ l ( R 2 = 0.6868, n = 147, p b 0.001) and SPT= BT 0.7864+ 1,092,629 cells/l (R2 =0.7895, n = 147, p b 0.001) (Fig. 2a). In a BlandAltman analysis (Fig. 2b, c and d) the spread of the data was similar for CVA and SPT compared to BT counting (Fig. 2b and d). The mean difference of leukocyte concentrations determined by the CVA and SPT methods and BT counting were 1.86% (IQR 6.779.85%, ns) and 5.35% (IQR 2.4214.04%, p b 0.001), respectively. The difference between CVA and SPT determinations was signicant (3.36%, IQR 1.37%9.03%; p b 0.001). The median difference from the predicate BT counting method to SPT was signicantly larger than to CVA at high leukocyte concentrations, but was less at low leukocyte concentrations (Fig. 2e). The binned net differences between the predicate BT counting method and CVA or SPT were symmetrical around the line of equality (Fig. 2f). 4. Discussion In ow cytometric analysis, cell concentrations are often given as percentage change without regard to the absolute cell concentrations. For economic reasons, absolute leukocyte concentrations have primarily been the domain of clinical immunology and infections disease, as a decrease of absolute CD4 T cell concentrations to below 200 cells/l is a critical threshold in the progression of HIV infection (Yarchoan et al., 1991). Here we show that CVA can be applied in a research setting to determine changes in blood leukocyte concentrations to give results comparable to both BT counting and SPT. In a clinical experiment changes in basophil granulocytes and pDC were investigated during specic immunotherapy by venom injection of persons allergic to wasp sting (Hansen, Dahl and Hoffmann, in preparation). Total leukocyte concentrations were determined as quality control tool by BT counting, SPT and CVA. 4.1. Calculation of constant volume acquisition (CVA) volume The ow rate of ow cytometers is constant enough to allow aspiration of a xed volume in a predetermined time (Bergeron et al., 2003; Storie et al., 2003). The CV of the rate determined here, 4.97%, was comparable to that reported by

previously for a Calibur; b 6% (Bergeron et al., 2003). To utilize CVA one may adopt an approach similar to the one described here where a ow cytometric method is standardized against an accepted predicate method, avoiding the complexity of adjusting for viscosity (Storie et al., 2003; Walker et al., 2006). The ow rate of the cytometer(s) should be checked regularly (N four times on one day to establish a mean event number collected with CV, then weekly followed by monthly, and after each service of the instrument) with leukocytes of known concentration (as has been suggested by Walker et al., 2006) or with a lot of beads reserved for this purpose. The method requires identication of CD45+ leukocytes for PanLeukoGating (Glencross et al., 2002) for comparison to leukocyte concentration determined by BT counting. The volume analysed here in 4 min was 163 l. Others have found that 100 l were analysed in 2 min (Storie et al., 2003), suggesting that it is important to calibrate each ow cytometer individually with sufcient readings and to monitor it continuously. Digital ow cytometers have better algorithms to identify and discard doublets. The ow rate should thus be set to minimize doublet generation, yet maximize the number of events counted as this improves the counting accuracy. As long as both calibration and experimental measurements are done at the same ow rate in the same medium, the improved doublet discrimination should not affect accuracy. In a series of eight acquisitions of the same sample of blood to determine reproducibility, the bead and cell number did not decrease linearly with time. Data for this analysis had been acquired over 10 months, and the instrument had been serviced (preventive maintenance) twice in this period. There was no clear association between service and variation. The blood volume analysed was 14.8 l. This is nearly 30 times more than evaluated in the BT chamber. It should thus be possible to enumerate small cell populations. The normal range of leukocyte concentrations is from 300010,000 cells/l; the present data ranged from 3320 to 14,040 cells/l. 4.2. Reproducibility of determinations by microscope, and ow cytometry with SPT or with constant volume There was little correlation between the expected variation (proportional to n/n) and the determined variation (b 6%) of data collected in a BrkerTrk chamber, and of repeated analyses of TruCount tubes (3%). In other trials to determine the reproducibility of CD4 counts for monitoring of HIV patients, the intralaboratory variation was 12%, and the interlaboratory variation was 3% for CD4 T cell counts (Schnizlein-Bick et al., 2000).

S. Hansen et al. / Journal of Immunological Methods 338 (2008) 5862

61

Fig. 2. Comparison of procedures for evaluation of leukocyte concentrations (a) SPT and CVA determinations are plotted as function of the predicate BT chamber counts. (bd) BlandAltman plots comparing CVA with BT counts and SPT. The median and CV are indicated by solid and dotted lines. (e) Distribution of data at low (30006000 cells/l), medium (600012,000 cells/l) and high (N 12,000 cells/l) leukocyte concentrations was determined for the three methods compared as medians and quartiles, (f) Histogram of differences of all measures between the predicate method, BT counting, and SPT and CVA.

4.3. Validation of SPT and CVA algorithms against microscopy BT counting chamber is the predicate method in our laboratory. This is different to the approach that a Clinical Im-

munology laboratory would have; they would compare data with a hematology instrument or with a SPT method (Nicholson et al., 1997; Schnizlein-Bick et al., 2000; Storie et al., 2003). Calculating leukocyte concentrations on the basis of a constant

62

S. Hansen et al. / Journal of Immunological Methods 338 (2008) 5862

volume were more similar to microscopic leukocyte counts than concentrations determined with a daily TruCount bead number. As the calculations were done on the same data set, the only difference between CVA and SPT determinations was that the volume in SPT was based on the both the actual number of beads in an individual tube and the number of beads acquired with the cells, compared with an average volume used in CVA. Thus the variation introduced in the SPT tubes is likely to be due to variation in precision of bead measurements. The fact that the number of beads in each TruCount tube may vary, and may vary more than the volume aspirated by a ow cytometer over constant time, has not been proposed before. As CVA used the same data les as the SPT, there were no differences in pipetting or viscosity. This conrms and extends the result of earlier studies (Bergeron et al., 2003; Storie et al., 2003; Walker et al., 2006) in which the volume sampled during a xed amount of time was sufciently accurate to calculate absolute cell concentrations. The single platform test using TruCount tubes has previously been shown to be different from other single platform methods (Nicholson et al., 1997; Schnizlein-Bick et al., 2000), but is thought to be more accurate than two platform evaluation in interlaboratory comparisons (Schnizlein-Bick et al., 2000). The difference between SPT and two platform measurements was attributed to hematology instruments that were part of the dual platform. 5. Conclusion Flow cytometry is a versatile and powerful technique that allows statistical analysis at the single cell level. The expansion of detection systems to more than 15 simultaneous uorochomes has received much coverage in scientic literature, and it is surprising that the studies by Bergeron et al. (2003) and Storie et al. (2003) on which this work is based, have not gained as much popularity. In this study we show that it is possible to determine leukocyte concentration by acquiring data for a xed time provided the acquisition volume of the ow cytometer is monitored. Combined with the digital revolution in computing and the expansion of the number of parameters detected, CVA will make ow cytometry even more versatile, and more precise. CVA is the method of choice as it is signicantly cheaper than SPT, and less labour intensive than BT counting.

Acknowledgements The technical assistance of Anne Marie Toft, and the participation of the immunotherapy patients and control persons, are appreciated. The project was funded by Aarhus University Hospital, the Velux Foundation and the Aarhus University Research Foundation. References
Bergeron, M., Lustyik, G., Phaneuf, S., Ding, T., Nicholson, J.K.A., Janossy, G., Shapiro, H., Barnett, D., Mandy, F., 2003. Stability of currently used cytometers facilitates the identication of pipetting errors and their volumetric operation: time can tell all. Cytometry B Clin. Cytom. 52, 37. Bousquet, J., Mller, U.R., Dreborg, S., Jarisch, R., Malling, H.J., Mosbech, H., Urbanek, R., Youlten, L., 1987. Immunotherapy with Hymenoptera venoms. Position paper of the Working Group on Immunotherapy of the European Academy of Allergy and Clinical Immunology. Allergy 42, 401. Glencross, D., Scott, L.E., Jani, I.V., Barnett, D., Janossy, G., 2002. CD45-assisted PanLeucogating for accurate, cost-effective dual-platform CD4+ T-cell enumeration. Cytometry 50, 69. Hoffmann, H.J., Skjold, T., Raithel, M., Adolf, K., Hilberg, O., Dahl, R., 2006. Response of respiratory our allergics in an ingested our challenge may involve plasmacytoid dendritic cells, CD25+ T cells and CD152+ T cells. Int. Arch. Allergy Immunol. 140, 252. Hoffmann, H.J., Malling, T.M., Topcu, A., Ryder, L.P., Nielsen, K.R., Varming, K., Dahl, R., Omland, O., Sigsgaard, T., 2007. CD4(dim)CD25(bright) Treg cell frequencies above a standardized gating threshold are similar in asthmatics and controls. Cytometry A 71, 371. Martini, F., D'Ofzi, G., Girardi, E., Vitale, A., Cimini, E., Mollicone, B., Vlassi, C., Aiuti, F., 2007. The 2005 Italian Quality Control Study for the evaluation of CD4 cells in centers involved in the treatment of HIV type 1 patients. AIDS Res. Hum. Retroviruses 23, 777. Nicholson, J.K., Stein, D., Mui, T., Mack, R., Hubbard, M., Denny, T., 1997. Evaluation of a method for counting absolute numbers of cells with a ow cytometer. Clin. Diagn. Lab. Immunol. 4, 309. Schnizlein-Bick, C.T., Spritzler, J., Wilkening, C.L., Nicholson, J.K., O'Gorman, M.R., 2000. Evaluation of TruCount absolute-count tubes for determining CD4 and CD8 cell numbers in human immunodeciency virus-positive adults. Site Investigators and The NIAID DAIDS New Technologies Evaluation Group. Clin. Diagn. Lab. Immunol. 7, 336. Storie, I., Sawle, A., Goodfellow, K., Whitby, L., Granger, V., Reilly, J.T., Barnett, D., 2003. Flow rate calibration I: a novel approach for performing absolute cell counts. Cytometry B Clin. Cytom. 55, 1. Walker, C.L., Whitby, L., Granger, V., Storie, I., Reilly, J.T., Barnett, D., 2006. Flow rate calibration. III. The use of stabilized biostandards to calibrate the ow rate and calculate absolute CD4+ T-cell counts. Cytometry B Clin. Cytom. 70, 154. Yarchoan, R., Venzon, D.J., Pluda, J.M., Lietzau, J., Wyvill, K.M., Tsiatis, A.A., Steinberg, S.M., Broder, S., 1991. CD4 count and the risk for death in patients infected with HIV receiving antiretroviral therapy. Ann. Intern. Med. 115, 184.