Cytometry (Communications in Clinical Cytometry) 46:254 –261 (2001)

Enumeration of CD34ⴙ Cells in Cord Blood: A

Variation on a Single-Platform Flow Cytometric
Method Based on the ISHAGE Gating Strategy
Anne M. Brocklebank and Rosemary L. Sparrow*
Development Unit, Australian Red Cross Blood Service-Victoria, Southbank, Victoria, Australia

Single-platform flow cytometric absolute cell counting protocols provide increased robustness for CD34ⴙ
cell enumeration by limiting potential sources of imprecision. However, samples with any cellular fragmen-
tation or debris, such as cord blood samples, provide challenges for these assays. We describe a simple,
robust absolute CD34ⴙ cell counting protocol, suitable for cord blood, using TRUCOUNT™ absolute count
tubes (BD Biosciences, San Jose, CA) and a modified ISHAGE (International Society for Hematotherapy and
Graft Engineering) gating strategy. An advantage of TRUCOUNT tubes is that each tube is supplied with a
known number of lyophilized fluorescent beads. The method includes no-wash fixative-free ammonium
chloride red blood cell lysis and the viability dye, 7-amino actinomycin D, to exclude dead cells. The
threshold was set on CD45 expression in the FL1 channel and an exclusion gate in the forward scatter
channel reduced debris. No manual adjustment of the gating regions was required, even for samples in less
than optimal condition. Comparison of the TRUCOUNT-ISHAGE protocol with the original dual-platform
ISHAGE assay (n ⴝ 30) and the single-platform ISHAGE protocol using Flow-Count™ Fluorospheres
(Beckman Coulter, Fullerton, CA; n ⴝ 22) showed high correlation (R2 ⴝ 0.949 and 0.989, respectively) and
no significant difference or bias for samples ranging from 22 to 600 CD34ⴙ cells per microliter. Results are
presented that demonstrate the detrimental effect of a fixative-containing lysis reagent when used in a
lyse-and-wash procedure. The TRUCOUNT-ISHAGE protocol combines the attributes of TRUCOUNT tubes and
the ISHAGE gating strategy to provide a single-platform protocol capable of achieving readily standardization
of CD34ⴙ cell enumeration. Cytometry (Comm. Clin. Cytometry) 46:254 –261, 2001.
© 2001 Wiley-Liss, Inc.

Key terms: CD34; cord blood; ISHAGE; single platform; stem cells

Human umbilical cord blood is a rich source of hema- by the International Society for Hematotherapy and Graft
topoietic stem cells (HSC) that is being used increasingly Engineering (ISHAGE; 8) were a significant attempt to-
as an alternative to bone marrow for HSC transplantation wards standardization of the CD34 assay. These guidelines
(1–3). The cellular content, in particular the number of were based on a dual-platform method whereby the per-
HSC, is a principal factor in assessing the adequacy of cord centage of CD34⫹ cells was determined by flow cytom-
blood units for clinical transplantation. HSC are contained etry and the leukocyte count by an automated hematology
within the CD34⫹ cell population, which represents typ- analyzer. A lyse-and-wash method for sample preparation,
ically less than 1% of the total leukocytes in cord blood a gating strategy based on a forward scatter threshold,
(4). This places CD34 analysis in the category of “rare- CD45 staining to define the total leukocyte population,
event” analysis and has presented a challenge to the de- and light scatter characteristics were recommended. Nev-
velopment of standardized, robust methods for the quan- ertheless, the use of two instruments that were used to
titation of HSC. Cord blood poses additional challenges. determine the absolute CD34⫹ cell number, the potential
These include the presence of variable numbers of nucle-
loss of cells during the lysis and wash steps, and the
ated red blood cells (RBC; which can lead to overestima-
tion of the leukocyte count) and the age of the samples,
which is often greater than 24 h old. Increased apoptosis,
cell death, and debris can potentially confound the anal- *Correspondence to: Dr. Rosemary Sparrow, Australian Red Cross
Blood Service-Victoria, P.O. Box 354, South Melbourne, Victoria 3205,
ysis. Australia.
CD34⫹ cell enumeration has relied predominantly on E-mail: [email protected]
flow cytometric procedures (5–7). Guidelines published Received 27 March 2001; Accepted 8 May 2001

© 2001 Wiley-Liss, Inc.

 SINGLE-PLATFORM CD34 ASSAY 255
influence of the choice of lysis reagent were sources of Using an alternative source of fluorescent beads,
variability and obstacles to standardization of the assay. Keeney et al. (15) refined the original ISHAGE method by
The development of single-platform methods enable the incorporating Flow-Count™ Fluorospheres (Beckman
absolute CD34⫹ cell count to be determined by a single Coulter, Fullerton, CA), along with a no-wash, fixative-free
instrument. Most of these new methods rely on flow lysis method for sample preparation and inclusion of a
cytometry. An exception is a microvolume fluorometry viability dye, 7-amino actinomycin D (7-AAD). This
method that utilizes a novel instrument, the IMAGN威 2000 method formed the basis of the Stem-Kit™ CD34 HPC
with the STELLer™ CD34 assay kit (BD Biosciences, San enumeration kit (Beckman Coulter). A significant advan-
Jose, CA). This instrument is fully automated. However, tage of this approach is the robustness and sensitivity of
the STELLer CD34 assay has not been optimized for cord the ISHAGE gating strategy. However, Flow-Count Fluoro-
blood and samples in less than pristine conditions, such as spheres are supplied as a bead suspension that necessi-
cord blood samples greater than 24 h old, are problematic tates careful handling to ensure accuracy. Recent data
(Sparrow, unpublished observations). Furthermore, it can- from a multicenter trial conducted by the CD34 Task
not distinguish live from dead cells. Force of the European Working Group of Clinical Cell
The flow cytometric-based single–platform methods Analysis demonstrated good interlaboratory agreement us-
center on the incorporation of a known number of fluo- ing the Flow-Count procedure (16). This trial was de-
rescent beads into the sample. Using a lyse-no-wash pro- signed specifically to ensure a high level of standardization
cedure for sample preparation enables the ratio of CD34⫹ across the participating laboratories. Stabilized blood sam-
cells to beads to be determined and the absolute CD34⫹ ples configured to represent mobilized peripheral blood
cell count to be calculated. The ProCOUNT™ assay (BD stem cell specimens were used with a standardized pro-
Biosciences) was the first commercial kit based on this tocol and targeted training. A protocol using TRUCOUNT
approach. This three-color assay utilizes TRUCOUNT™ tubes was also assessed and yielded comparable results
absolute count tubes (BD Biosciences), which contain a with good interlaboratory agreement.
predispensed lyophilized pellet of a known number of We describe an absolute cell counting procedure for
4.2-␮m fluorescent beads. Due to the small beads, a for- CD34 analysis of cord blood using TRUCOUNT tubes and
ward scatter threshold cannot be used, necessitating the a modification of the ISHAGE gating strategy. The method
threshold to be set on a fluorescence channel. For the includes fixative-free ammonium chloride RBC lysis and
ProCOUNT assay, a proprietary nucleic acid dye is used to elimination of nonviable CD34⫹ cells using 7-AAD. The
discriminate nucleated cells from erythrocytes, platelets, threshold is set on CD45 expression in the FL1 channel
and debris and to set the threshold in the FL1 channel. and debris is removed from the CD34 gating region using
CD34⫹ and CD45⫹ events are detected in the FL2 and an exclusion gate in the forward scatter channel. This
FL3 channels, respectively. Acquisition and analysis are TRUCOUNT-based method compared favorably with the
performed according to the gating strategy recommended original dual-platform ISHAGE method and single-platform
by the manufacturer or by using proprietary software (BD method using Flow-Count Fluorospheres. In addition, we
Biosciences). Most evaluations of the ProCOUNT assay show that the commercial lyse-and-fix reagent FACSLyse
have been favorable (9,10). However, significant discrep- (BD Biosciences) has a detrimental effect on CD34⫹ cell
ancies have been observed in some instances (11,12). enumeration when using a lyse-and-wash procedure. This
Furthermore, sample variability may necessitate manual new protocol combines the advantages of TRUCOUNT
adjustment of the gating regions, thus lessening the ro- tubes and the ISHAGE gating strategy in a simple, robust
bustness of the assay. The assay has not been optimized assay.
for cord blood or bone marrow samples and, for single-
laser flow cytometers, cannot be used to assess viability or MATERIALS AND METHODS
analyze subpopulations. Cord Blood Specimens
Hübl et al. (13) described a modification of the Pro- Umbilical cord blood samples (n ⫽ 30) were collected
COUNT assay that is applicable to cord blood. In this from full-term deliveries with informed consent from the
modified protocol, the nucleic acid dye was replaced with mothers and approval of the institutional ethics commit-
a viability dye, YO-PRO-1 (FL1 channel), and the threshold tee. Selection of this number of patients was based on a
was set on CD45 staining (FL3 channel). Nevertheless, power analysis and sample size determination (using Stat-
manual adjustment of the regions was still required for graphics statistical software) that indicated that 16 –30
each sample, potentially reducing the robustness of the samples was sufficient to compare the CD34 methods,
assay. with a 95% confidence level. All procedures used for
Fornas et al. (14) described a no-lysis, no-wash whole collecting and processing cord blood were based on the
blood method using TRUCOUNT tubes. The method relies standard operating procedures of the Australian Red Cross
on a vital nucleic acid dye, SYTO-13, to discriminate eryth- Blood Service-Victoria Cord Blood Bank. Briefly, cord
rocytes and debris from viable nucleated cells and to set blood was collected into a modified blood collection pack
the threshold in the FL1 channel. This protocol eliminates (Fenwal, Baxter Healthcare, Morelos, Mexico) containing
the lysis step. However, efflux of the SYTO-13 dye, par- 23 mL citrate-phosphate-dextrose-adenine anti-coagulant
ticularly in P-glycoprotein– expressing cells, necessitated solution and processed within 36 h of collection. Cord
the addition of cyclosporin A to the assay mixture. blood was centrifuged to obtain the leukocyte-rich buffy
256 BROCKLEBANK AND SPARROW

coat layer. An aliquot of this cellular fraction was used for Quest 3.1 software (BD Biosciences). The instrument was
CD34⫹ cell analysis. All assays were performed immedi- aligned and calibrated daily using a three-color mixture of
ately as samples became available. Blood cell counts were CaliBRITE beads (BD Biosciences) with FACSComp soft-
performed on either a Sysmex NE-8000 or Abbott Cell-Dyn ware (BD Biosciences).
3200 hematology analyzer. Proportion of nucleated RBC Gating strategy: dual-platform ISHAGE protocol.
was also determined by examining blood films. The original dual-platform ISHAGE gating strategy and
calculation to obtain the number of CD34⫹ cells per
CD34 Staining microliter was exactly as described previously (8).
Gating strategy: single-platform TRUCOUNT-ISHAGE
Dual-platform ISHAGE protocol. The original dual-
protocol. Figure 1 shows the gating strategy for CD34
platform ISHAGE protocol was performed as described
analysis using TRUCOUNT tubes. This protocol followed
previously (8). Cord blood samples were diluted in phos-
closely that of Keeney et al. (15) for single-platform
phate-buffered saline (PBS), pH 7.2, to obtain a leukocyte
ISHAGE analysis using Flow-Count Fluorospheres, with
count of 10 –20 ⫻ 109/L, as recommended by Keeney et
the following modifications. Because TRUCOUNT beads
al. (15). All samples were run in duplicate. Briefly, 100 ␮L
produce a very low forward scatter signal, a forward
of cord blood sample was incubated with 10 ␮L of FITC-
scatter threshold cannot be used. Instead, an exclusion
conjugated anti-CD45 (clone HuLyM4; Silenus, Mel-
gate (Fig. 1, Plot 9) was drawn to simulate a forward
bourne, Australia) and 20 ␮L of phycoerythrin (PE)-conju-
scatter threshold and eliminate signals arising from debris,
gated anti-CD34 (clone 581, Beckman Coulter) for 20 min
while allowing collection of all the bead data. An FL1
at 4°C. Initial experiments included samples incubated
threshold was set on CD45 expression, with care to en-
with an isotype control PE-labeled IgG1 (Beckman
sure that CD45low/CD34⫹ cells were not excluded from
Coulter) as a negative control for nonspecific binding.
analysis (Fig. 1, Plot 1). The live leukocytes were identified
RBC were lysed with 2.0 mL of bicarbonate-buffered am-
as 7-AAD–negative events (Fig. 1, Plot 2). The gating out of
monium chloride solution (RBC lysis solution; 0.15 M
dead cells (region R7) also served to gate out the TRU-
NH4Cl, 0.01 M NaHCO3, 1.0 mM EDTA) for 15 min at
COUNT beads from further CD34⫹ cell analysis. The
room temperature. The cells were washed twice, resus-
TRUCOUNT beads, evident as a population of bright
pended in 1.0 mL PBS, stored on ice in the dark, and
events in each fluorescence channel, were enumerated in
analyzed within 1 h.
region R6 (Fig. 1, Plot 8). To exclude nonbead events, R6
For some experiments, RBC were lysed for 10 min using
was drawn tightly around the beads in line with the
a fixative-containing RBC lysis reagent FACSLyse (BD Bio-
manufacturer’s instructions. CD34⫹ events were enumer-
sciences), washed twice, and resuspended in PBS.
ated in G4 (Fig. 1, Plot 6). A total of 100,000 viable
Single-platform protocol using TRUCOUNT tubes.
leukocytes (G1) were collected, or a minimum of 100
CD34 and CD45 antigen staining was performed as de-
CD34⫹ events and 2,000 bead events. The number of
scribed above, except TRUCOUNT tubes (BD Bio-
CD34⫹ cells per microliter was calculated by the follow-
sciences) were used. The tubes contain a known number
ing formula:
of lyophilized 4.2-␮m fluorescent beads. Immediately after
the addition of 2.0 mL of RBC lysis solution, 4 ␮L of 0.5
mg/mL 7-AAD in methanol (Molecular Probes, Eugene, (No. of CD34⫹ cells [G4]
OR) was added. To ensure accuracy, reverse pipetting was ⫻ No. of beads per TRUCOUNT tube)
used to dispense volumes greater than 20 ␮L. The tubes (No. of beads counted [G6]
were incubated at room temperature for 15 min, then ⫻ sample volume [100 ␮L])
placed on ice in the dark and analyzed within 1 h. Samples
were mixed gently immediately prior to analysis. ⫻ sample dilution factor.
Single-platform protocol using Flow-Count Fluo-
rospheres. Antigen staining and RBC lysis/7-AAD treat- Gating strategy: Flow-Count–ISHAGE protocol.
ment were performed as described above, except plain The gating strategy described by Keeney et al. (15) was
tubes were used (Falcon, BD Biosciences). Following lysis used with the modification that the bead data were col-
treatment, 100 ␮L of Flow-Count Fluorospheres (Beckman lected by forward scatter versus time parameters. The
Coulter) was added by reverse pipetting. These beads are number of CD34⫹ cells per microliter was calculated as
supplied as a suspension of a defined concentration of described previously (15).
10-␮m fluorescent beads in an aqueous medium contain-
ing surfactant and 1% formaldehyde. The beads must be Statistical Analysis
well mixed and equilibrated to room temperature prior to
Comparison of the number of CD34⫹ cells obtained by
use. Following addition of the beads, the sample was
the different methods was determined by the nonparamet-
mixed gently and analyzed immediately.
ric Wilcoxon test for paired samples and by Pearson’s
correlation coefficient, using Minitab software (Minitab,
Flow Cytometry State College, PA). Significance was defined as P ⬍ 0.05.
Samples were analyzed on a FACScan flow cytometer Linear regression analysis was performed using Excel 97
(BD Biosciences) with a 488-nm argon laser and Cell software (Microsoft, Redmond, WA). Agreement between
 SINGLE-PLATFORM CD34 ASSAY 257

FIG. 1. Modified ISHAGE gating strategy using TRUCOUNT tubes with FITC-anti-CD45, PE-anti-CD34, and 7-AAD on cord blood. Listmode data were
acquired on a FACScan (BD Biosciences) and analyzed with Cell Quest software. Plot 1: Threshold on FL-1 channel; all events collected are shown in
this ungated plot. Plot 2: Discrimination of live cells from dead cells. Events falling in R7 are gated out of further analysis. Plot 3: First ISHAGE plot,
gated on live cells. R1 defines all CD45⫹ leukocytes and R5 defines the lymphocyte population. Plot 4: Second ISHAGE plot, gated on live CD45⫹
leukocytes. R2 defines possible CD34⫹ cells. Plot 5: Third ISHAGE plot, gated on live CD45⫹CD34⫹ events. R3 defines clustering of CD34⫹ events.
Plot 6: Fourth ISHAGE plot, gated on live CD45⫹CD34⫹ events. R4 defines clustering CD34⫹ events which have similar light scatter properties and
is used to enumerate CD34⫹ cells. Plot 7: Live lymphocytes. The left side of R4 is drawn to include cells no smaller than a small lymphocyte, and assists
to position the exclusion gate in Plot 9. Plot 8: R6 defines TRUCOUNT beads; this plot assists to define the lower limit of CD45 expression of CD34⫹
cells. Plot 9: Scatter plot. R8 defines an exclusion gate, which removes most of the platelets and debris while allowing TRUCOUNT beads to be collected.

the methods was assessed by analysis of the relationship tocol, showed that the FACSLyse method gave signifi-
between the differences and the mean of the differences, cantly lower (P ⬍ 0.006) estimates of CD34⫹ cell number
as recommended by Bland and Altman (17). Close agree- than the ammonium chloride method (median CD34⫹
ment is indicated by the mean of the differences ap- cells per microliter with FACSLyse ⫽ 95, 95% confidence
proximating zero and the scatter of the data points interval [CI] 51–247; median CD34⫹ cells per microliter
being randomly distributed around the mean of the with ammonium chloride ⫽ 172, CI 93–375). Although a
differences. high correlation coefficient was achieved (R2 ⫽ 0.942),
the slope was 0.65 and intercept –11.2 (Fig. 2A). Analysis
RESULTS of the differences of the means of the two methods, as
Influence of RBC Lysis Reagent described by Bland and Altman (17), gave a mean bias of
Our original method for CD34⫹ cell analysis was based 85.1 (Fig 3A), indicating significant disagreement between
on the dual-platform ISHAGE protocol and included lysis the methods. We also analyzed mobilized peripheral
of RBC with a proprietary fixative-containing lysis reagent, blood samples supplied for an interlaboratory CD34 as-
FACSLyse (BD Biosciences), followed by two washes (8). sessment program (18). Comparison with results from
Initial experiments were performed to compare FACSLyse other laboratories indicated closer agreement with the
with a nonfixative-containing ammonium chloride RBC fixative-free ammonium chloride method and underesti-
lysis solution. The results from cord blood samples (n ⫽ mation using FACSLyse (results not shown). All subse-
10), each prepared simultaneously using the two lysing quent experiments used the ammonium chloride RBC
reagents and analyzed by the dual-platform ISHAGE pro- lysis reagent.
258 BROCKLEBANK AND SPARROW

Single-Platform TRUCOUNT-ISHAGE Protocol channel and the threshold was set on CD45⫹ events in
The original ISHAGE gating strategy was modified to the FL1 channel. Acquisition time to acquire 100,000 live
include a polygonal exclusion gate in the forward scatter CD45⫹ events was typically 5 to 6 min and the entire
assay could be completed in less than 1 h. The cord blood
samples analyzed had a range of 22– 600 CD34⫹ cells per
microliter and the mean coefficient of variation of repli-
cate samples was 7% (range 0 –15 %). A negative control
was included in initial experiments but was not necessary,
as has been shown for the ISHAGE protocol (8,15). No
manual adjustment of the gating regions was required for
any of the samples.
Comparison of CD34 Protocols
Figure 2B shows the linear correlation of the absolute
CD34⫹ cell counts of cord blood samples (n ⫽ 30) ana-
lyzed simultaneously by the single-platform TRUCOUNT-
ISHAGE protocol and the dual-platform ISHAGE protocol.
The two methods showed good correlation (R2 ⫽ 0.949,
slope 0.97, and intercept 5.0) and good agreement, as
demonstrated by a mean bias of 0.11 (Fig. 3B). There was
no significant difference between the median CD34⫹
cells per microliter obtained for the two methods (TRU-
COUNT median ⫽ 152, CI 110 –220; dual-platform
ISHAGE median ⫽ 155, CI 111–219; P ⫽ 0.86).
Similarly, comparison of the single-platform method us-
ing Flow-Count Fluorospheres and the dual-platform
ISHAGE protocol (n ⫽ 22) showed good correlation (R2 ⫽
0.954, slope 0.89, and intercept 19.4; Fig. 2C). There was
good agreement between the methods as determined by a
mean bias of 1.2 (Fig. 3C). The median CD34⫹ cells per
microliter values were not significantly different (Flow-
Count median ⫽ 160, CI 107–234; dual-platform ISHAGE
median ⫽ 155, CI 104 –222; P ⫽ 0.66).
Finally, the single-platformTRUCOUNT and Flow-Count
methods were compared. The results (n ⫽ 22) showed
good correlation (R2 ⫽ 0.989, slope 1.0, and intercept
-9.2; Fig. 2D) and good agreement between the methods
as determined by a mean bias of -5.7 (Fig. 3D). The median
CD34⫹ cells per microliter values were comparable (TRU-
COUNT median ⫽ 150, CI 102–217; Flow-Count me-
dian ⫽ 160, CI 107–234).
DISCUSSION
The increasing clinical utilization of HSC transplanta-
tion, the establishment of cord blood banks, and the
inclusion of HSC processing facilities within the control of
national regulatory authorities have hastened the need for
accurate and robust assays for CD34⫹ cell enumeration.
Cord blood has posed additional challenges, particularly

4™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™
FIG. 2. Comparison of methods. A: Dual-platform ISHAGE with FAC-
SLyse versus dual-platform ISHAGE with ammonium chloride lysis re-
agent. B: Single-platform TRUCOUNT versus dual-platform ISHAGE with
ammonium chloride lysis reagent. C: Single-platform Flow-Count versus
dual-platform ISHAGE with ammonium chloride lysis reagent. D: Single-
platform TRUCOUNT versus single-platform Flow-Count. Dashed line
represents theoretical perfect agreement between the methods. Solid
line represents calculated linear regression line of the observed data.
Regression equation for the observed data is shown.
SINGLE-PLATFORM CD34 ASSAY 259
due to the presence of varying numbers of nucleated RBC
and the extended age of the samples at the time of CD34⫹
cell enumeration. Single-platform absolute CD34⫹ cell-
counting protocols offer the greatest potential for the
development of robust assays capable of analytical accu-
racy and standardization. In this study, we described a
modification of the single-platform ISHAGE protocol that
incorporates an alternative source of fluorescent beads,
presented in the format of TRUCOUNT tubes, which is
suitable for analysis of cord blood samples.
Initial experiments were performed to ascertain the
influence of the RBC lysis reagent. These experiments
utilized the original dual-platform ISHAGE lyse-and-wash
protocol. Our results demonstrated that the fixative-con-
taining FACSlyse reagent gave significantly lower esti-
mates of the CD34⫹ cell number compared with the
fixative-free bicarbonate-buffered ammonium chloride
RBC lysis solution. These results were consistent with
those reported by others (19 –24). Loss of CD34⫹ cells
following treatment with fixative-containing solutions has
been suggested to be a consequence of increased sticki-
ness of fixed cells (24). Fixed cells may be lost from the
cell suspension either by increased adhesion to the sur-
face of the assay tubes or greater loss during washing.
Gratama et al. (24) recommended that the wash steps be
omitted to minimize cell loss associated with fixative-
containing RBC lysis reagents.
Other investigators have cautioned the use of ammo-
nium chloride RBC lysis reagents due to the observation
that CD34⫹ cells undergo light scatter changes character-
istic of apoptosis when exposed to such reagents (21). In
the experiments described here, sample analysis was com-
pleted typically within 30 min following the addition of
the ammonium chloride lysis solution. If the samples were
not analyzed immediately, they were placed on ice follow-
ing the 15-min lysis step. In some instances, samples were
analyzed up to 60 min following the addition of lysis
solution (results not shown). No change in the light scat-
ter characteristics was observed with delayed sample anal-
ysis, seemingly a benefit of placing the tubes on ice prior
to analysis (23). Based on these results, we adopted a
fixative-free bicarbonate-buffered ammonium chloride so-
lution as the preferred RBC lysis reagent.
In order to have in place a more robust CD34 assay for
routine analysis of HSC products, in particular cord blood,
we investigated the options of single-platform flow cytom-
etry protocols. We preferred a method that utilized the
ISHAGE gating strategy because of its robustness, sensitiv-
ity, and widespread use. Although the single-platform

4™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™™
FIG. 3. Bland-Altman plots of differences between the methods versus
their mean. A: Dual-platform ISHAGE with ammonium chloride versus
dual-platform ISHAGE with FACSLyse. B: Single-platform TRUCOUNT
versus dual-platform ISHAGE with ammonium chloride. C: Single-plat-
form Flow-Count versus dual-platform ISHAGE with ammonium chloride.
D: Single-platform TRUCOUNT versus single-platform Flow-Count. Solid
line represents the mean of the differences or bias. Dashed lines repre-
sent mean of the difference (bias) ⫾ 2 SD. A mean bias equal to zero
represents perfect agreement.
260 BROCKLEBANK AND SPARROW

ISHAGE method using Flow-Count Fluorospheres is well In conclusion, we described a protocol that combines
established (15), there are disadvantages with the use of the attributes of TRUCOUNT tubes and the ISHAGE gating
Flow-Count or with similar fluorescent bead suspensions. strategy to provide a robust, accurate single-platform ab-
These include (1) ensuring a homogenous bead suspen- solute viable CD34⫹ cell-counting method. Although we
sion prior to dispensing; (2) precise pipetting of the fluo- established this assay for cord blood samples, this proto-
rospheres, as well as the sample; (3) presence of preser- col would be equally applicable to other types of HSC
vative in the bead suspension, which can have an adverse samples as indicated by our preliminary results with mo-
effect on cell viability (25), necessitating that the beads be bilized peripheral blood and leukapheresis samples. This
added immediately prior to flow cytometric analysis; and TRUCOUNT-ISHAGE protocol provides an alternative ro-
(4) comparing the practicality of pack size versus cost- bust, time-efficient assay that is capable of readily achiev-
effectiveness (i.e., the current 200-test pack size of Flow- ing standardization across clinical laboratories.
Count Fluorospheres combined with a 1-month expiry
upon opening may not be cost-effective for many labora- ACKNOWLEDGMENTS
tories). We thank Dr. Jenny Cauchi (Douglas Hocking Research
We were attracted to using TRUCOUNT tubes, as this Institute, Geelong) and Lanny Ramadi for assistance in
approach eliminates a potential source of error associated preparing the samples. We thank Robert Sutherland (On-
with dispensing the beads. This product is ready to use, cology Research, The Toronto Hospital, Toronto, Canada)
does not contain preservatives that interfere with cell for his generous and helpful suggestions regarding flow
viability, and each 25-test pack has a 2-month expiry upon cytometry gating strategies.
opening. We modified the ISHAGE protocol by perform-
ing CD34 and CD45 labeling, RBC lysis, and 7-AAD stain-
LITERATURE CITED
ing in the TRUCOUNT tubes. Because the TRUCOUNT
1. Cairo MS, Wagner JE. Placental and/or umbilical cord blood: an
beads were small, modifications were required to the alternative source of hematopoietic stem cells for transplantation.
single-platform ISHAGE gating strategy compared with Blood 1997;90:4665– 4678.
those described previously (15). Unlike Flow-Count Fluo- 2. Rubinstein P, Carrier C, Scaradavou A, Kurtzberg J, Adamson J, Migli-
accio AR, Berkowitz RL, Cabbad M, Dobrila NL, Taylor PE, Rosenfield
rospheres, a forward scatter threshold cannot be used for RE, Stevens CE. Outcomes among 562 recipients of placental-blood
TRUCOUNT beads. Consequently, the threshold must be transplants from unrelated donors. N Engl J Med 1998;339:1565–
set on a fluorescence channel that defines the denomina- 1577.
3. Gluckman E. Current status of umbilical cord blood hematopoietic
tor cell population. We chose CD45 expression in the FL1 stem cell transplantation. Exp Hematol 2000;28:1197–1205.
channel to set the threshold, allowing the FL3 channel to 4. Sutherland DR, Keating A, Nayar R, Anania S, Stewart AK. Sensitive
detection and enumeration of CD34⫹ cells in peripheral and cord
be used for the viability dye, 7-AAD. In order to reduce blood by flow cytometry. Exp Hematol 1994;22:1003–1010.
debris in the acquisition region, a polygonal exclusion 5. Gratama JW, Orfao A, Barnett D, Brando B, Huber A, Janossy G,
gate was set in the forward scatter channel between the Johnsen HE, Keeney M, Marti GE, Preijers F, Rothe G, Serke S,
Sutherland DR, Van Der Schoot CE, Schmitz G, Papa S. Flow cyto-
bead events and cells. The results obtained from cord metric enumeration of CD34⫹ hematopietic stem and progenitor
blood samples, which had a wide range of CD34⫹ cell cells. Cytometry 1998;34:128 –142.
concentrations, showed good agreement compared with 6. Brando B, Barnett D, Janossy G, Mandy F, Autran B, Rothe G, Scarpati
B, D’Avanzo G, D’Hautcourt J-L, Lenkei R, Schmitz G, Kunkl A,
the original dual-platform ISHAGE protocol and the single- Chianese R, Papa S, Gratama JW. Cytofluorometric methods for as-
platform Flow-Count–ISHAGE protocol. We found this sessing absolute numbers of cell subsets in blood. Cytometry 2000;
42:327–346.
TRUCOUNT-ISHAGE method to be technically simple and 7. Keeney M, Sutherland DR. Stem cell enumeration by flow cytometry:
time efficient. As has been noted by others (5,6), particu- current concepts and recent developments in CD34⫹ cell enumera-
lar care must be taken in dispensing the sample and RBC tion. Cytotherapy 2000;2:395– 402.
8. Sutherland DR, Anderson L, Keeney M, Nayar R, Chin-Yee I. The
lysis reagent to ensure accuracy and reproducibility of the ISHAGE guidelines for CD34⫹ cell determination by flow cytometry.
results. Reverse pipetting using calibrated pipettes is es- J Hematother 1996;5:213–226.
sential. Coefficients of variation of less than 10% were 9. McNiece I, Kern B, Zilm K, Brunaud C, Dziem G, Briddell R. Minimi-
zation of CD34⫹ cell enumeration variability using the ProCOUNT
readily achievable. The TRUCOUNT-ISHAGE gating strat- standardized methodology. J Hematother 1998;7:499 –504.
egy described proved to be robust and efficient. Data 10. Sims LC, Brecher ME, Gertis K, Jenkins A, Nickischer D, Schmitz JL,
Sparks S, Wright P, Bentley S. Enumeration of CD34-positive stem
acquisition was achieved within 5– 6 min per sample and cells: evaluation and comparison of three methods. J Hematother
no manual adjustment of the gating regions was required, 1997;6:213–226.
even for samples that were in less than pristine condition. 11. Gutensohn K, Carrero I, Krueger W, Kroeger N, Schäfer P, Luede-
mann K, Kuehnl P. Semi-automated flow cytometric analysis of CD34-
We also assessed the TRUCOUNT-ISHAGE protocol for expressing hematopoietic cells in peripheral blood progenitor cell
the enumeration of CD34⫹ cells in mobilized peripheral apheresis products. Transfusion 1999;39:1220 –1226.
blood and leukapheresis samples. The results obtained 12. Barbosa IL, Sousa ME, Godinho MI, Carvalhais A. Single- versus dual-
platform assays for human CD34⫹ cell enumeration. Cytometry
were comparable with those achieved using the dual- 1999;38:274 –279.
platform ISHAGE protocol (results not shown), indicating 13. Hübl W, Iturraspe J, Martinez GA, Hutcheson CE, Roberts CG, Fisk
DD, Sugrue MW, Wingard JR, Braylan RC. Measurement of absolute
that this method is applicable for samples other than cord concentration and viability of CD34⫹ cells in cord blood and cord
blood. We did not observe the “vanishing counting bead” blood products using fluorescent beads and cyanine nucleic acid
phenomenon (26) in our small series of leukapheresis dyes. Cytometry 1998;34:121–127.
14. Fornas O, Garcia J, Petriz J. Flow cytometry of CD34⫹ cells in whole
samples. However, the recommendation of ensuring suf- blood. Nature Med 2000;6:833– 836.
ficient protein in diluting media is notable. 15. Keeney M, Chin-Yee I, Weir K, Popma J, Nayar R, Sutherland DR.
 SINGLE-PLATFORM CD34 ASSAY 261
Single platform flow cytometric absolute CD34⫹ cell counts based on sis/necrosis on exposure to an ammonium chloride-based red blood
the ISHAGE guidelines. Cytometry 1998;34:61–70. cell lysing reagent. J Hematother 1998;7:315–318.
16. Barnett D, Granger V, Kraan J, Whitby L, Reilly JT, Papa S, Gratama 22. Menéndez P, Redondo O, Rodriguez A, Lopez-Berges MC, Ercilla G,
JW. Reduction of intra- and interlaboratory variation in CD34⫹ stem López A, Durán A, Almeida J, Pérez-Simón JA, San Miguel JF, Gratama
cell enumeration using a stable test material, standard protocols and JW, Orfao A. Comparison between a lyse-and-then-wash method and
targeted training. Br J Haematol 2000;108:784 –792. a lyse-non-wash technique for the enumeration of CD34⫹ hemato-
17. Bland JM, Altman DG. Statistical methods for assessing agreement poietic progenitor cells. Cytometry 1998;34:264 –271.
between two methods of clinical measurement. Lancet 1986;1:307– 23. Keeney M, Chin-Yee I, Nayar R, Sutherland DR. Effect of fixatives on
310. CD34⫹ cell enumeration. J Hematother Stem Cell Res 1999;8:327–
18. Chang A, Ma DDF. The influence of flow cytometric gating strategy 329.
on the standardization of CD34⫹ cell quantitation: an Australian 24. Gratama JW, Menéndez P, Kraan J, Orfao A. Loss of CD34⫹ hemato-
multicenter study. J Hematother 1996;5:605– 616. poietic progenitor cells due to washing can be reduced by the use of
19. Bossuyt X, Marti GE, Fleisher TA. Comparative analysis of whole fixative-free erythrocyte lysing reagents. J Immunol Methods 2000;
blood lysis methods for flow cytometry. Cytometry 1997;30:124 – 239:13–23.
133. 25. Mifsud NA, Wall DM, Sparrow RL. An automated cell counter com-
20. Cassens U, Gutensohn K, Garritsen H, Kelsch R, Kuehnl P, Sibrowski pared with flow cytometry for obtaining accurate leukocyte and
W. The influence of different erythrocyte lysing procedures on flow lymphocyte subset estimations. Transfusion Med 1998;8:157.
cytometric determination of CD34⫹ cells in umbilical cord blood 26. Brando B, Göhde W Jr, Scarpati B, D’Avanzo G. The “vanishing
transplants. Transfusion Med 1998;8:111–118. counting bead” phenomenon: effect on absolute CD34⫹ cell count-
21. Serke S, van Lessen A, Pardo I, Huhn D. Selective susceptibility of ing in phosphate-buffered saline-diluted leukapheresis samples. Cy-
CD34-expressing cells to acquire flow cytometric features of apopto- tometry 2001;43:154 –160.