Variable product purity and functional capacity after CD34 selection: a direct comparison of the CliniMACSR (v2.1) and IsolexR 300i (v2.5) clinical scale devices

Abstract

The two clinical scale devices currently available for CD34+ cell selection from peripheral blood stem cells (PBSC) apheresis products, the CliniMACS and the Isolex 300i, were compared directly by pooling and splitting two PBSC harvests collected on sequential days from 10 patients and processing half of each pooled harvest on each device. The CliniMACS product had significantly higher median CD34+ purity (90%vs 78%; P = 0.004) and lower median T-cell content (0.06%vs 0.44%; P = 0.003) compared with the Isolex 300i product. The median CD34+ yields were similar (64% and 60% respectively). However, when the functional capacities of the products were compared, the median recovery of colony-forming units was significantly greater from the Isolex 300i product (48%vs 38%; P = 0.035), as was expansion of cells in either erythroid or granulocytic lineage-specific liquid culture (2.1-fold more erythroid and 1.5-fold more granulocytic lineage progenitors on d 9 (P = 0.03 and 0.03 respectively). This was due to a higher proportion of apoptotic cells in the CliniMACS product (28%vs 18%; P = 0.007, annexin V binding). Hence, although the CliniMACS device yielded a higher purity product with fewer T cells, the Isolex 300i product contained fewer apoptotic cells and consequently had greater functional capacity in culture.

Full text

Variable product purity and functional capacity after CD34
selection: a direct comparison of the CliniMACS

(v2Æ1)
and Isolex

300i (v2Æ5) clinical scale devices
Michael J. Watts, Tim C. P. Somervaille, Stuart J. Ings, Forhad Ahmed, Asim Khwaja, Kwee Yong
and David C. Linch Department of Haematology, Royal Free and University College London Medical School, London,
UK
Received 8 October 2001; accepted for publication 12 December 2001
Summary. The two clinical scale devices currently avail-
able for CD34
+
cell selection from peripheral blood stem
cells (PBSC) apheresis products, the CliniMACS and the
Isolex 300i, were compared directly by pooling and split-
ting two PBSC harvests collected on sequential days from
10 patients and processing half of each pooled harvest on
each device. The CliniMACS product had significantly
higher median CD34
+
purity (90% vs 78%; P¼0Æ004)
and lower median T-cell content (0Æ06% vs 0Æ44%;
P¼0Æ003) compared with the Isolex 300i product. The
median CD34
+
yields were similar (64% and 60%
respectively). However, when the functional capacities of
the products were compared, the median recovery of col-
ony-forming units was significantly greater from the Isolex
300i product (48% vs 38%; P¼0Æ035), as was expansion
of cells in either erythroid or granulocytic lineage-specific
liquid culture (2Æ1-fold more erythroid and 1Æ5-fold more
granulocytic lineage progenitors on d 9 (P¼0Æ03 and
0Æ03 respectively). This was due to a higher proportion of
apoptotic cells in the CliniMACS product (28% vs 18%;
P¼0Æ007, annexin V binding). Hence, although the
CliniMACS device yielded a higher purity product with
fewer T cells, the Isolex 300i product contained fewer
apoptotic cells and consequently had greater functional
capacity in culture.
Keywords: CliniMACS

, (v2Æ1), Isolex

, 300i (v2Æ5), CD34
+
selection, CD34
+
product functional capacity, apoptosis.
Peripheral blood stem cells (PBSCs), mobilized with a variety
of growth factor and chemotherapeutic regimens, are now
used extensively in autologous and allogeneic transplanta-
tion (Schmitz et al, 1996; Gratwohl et al, 2001). Many
PBSC harvests are used directly without ex vivo manipula-
tion. However, where there is a requirement either for T-cell
depletion of the harvest to reduce the risk of graft-versus-
host disease following haploidentical and other allogeneic
transplants (Urbano-Ispizua et al, 1997; Kawano et al,
1998), or for purging of malignant cells from the harvest
(Cagnoni & Shpall, 1996), selection of CD34
+
progenitor
cells before transplantation may be carried out. Addition-
ally, CD34
+
cell selection is a prerequisite for some gene
therapy procedures (Devereux et al, 1998), as well as other
experimental approaches involving clinical scale ex vivo
expansion of haemopoietic cells for transplantation (Reiffers
et al, 1999; Ardeshna et al, 2000).
Two devices are currently available for clinical scale
CD34
+
cell selection from PBSC apheresis products: the
CliniMACS (Miltenyi Biotec GmbH, Bergisch Gladbach,
Germany) and the Isolex 300i (Baxter Healthcare, Deerfield,
IL). These devices differ in the manner in which the CD34
+
cells are isolated. The CliniMACS device utilizes a monoclo-
nal anti-CD34 antibody (QBEnd10), directly conjugated to
an iron oxide/dextran particle approximately 50 nm in size.
The conjugated antibody binds CD34
+
cells and these are
captured using high-gradient magnetic fields created by pla-
cing a column of small ferromagnetic beads between the poles
of rare earth magnets. To enhance the purity of the CD34
cells, three rounds of magnetic selection are used. The CD34
+
cells in the resultant product retain the ferromagnetic
particles following selection. The Isolex 300i utilizes a
different murine monoclonal anti-CD34 antibody (9C5),
which is unconjugated. Following coating of the CD34
+
cells in the harvest with the antibody, cells are rosetted with
magnetic beads (4Æ5lm in size) coated with polyclonal
Correspondence: Michael J. Watts, Department of Haematology,
Royal Free and University College Medical School, Department of
Haematology, Royal Free and University College London Medical
School, 98 Chenies Mews, London WC1E 6HX, UK. E-mail:
michael.watts@ucl.ac.uk
British Journal of Haematology, 2002, 118, 117–123
2002 Blackwell Science Ltd 117

sheep anti-mouse immunoglobulin (Dynabeads). Addition-
ally, DNAase may be added to reduce cell clumping and
increase yield and purity (Fritsch et al, 2000; Martin-Henao
et al, 2000). The cells are captured by an open field
magnetic system using rare earth magnets and then washed
three times by capture and release. The anti-CD34 magnetic
bead complex is subsequently removed from cells by the
excess addition of an octapeptide (PR34
+
stem cell releasing
agent), which binds with high affinity to the epitope
recognized by the 9C5 anti-CD34 monoclonal antibody.
Studies have been carried out to evaluate the perform-
ance of these devices, many of which have utilized earlier
versions of the software running the devices (Stainer et al,
1998; Schumm et al, 1999; Croop et al, 2000; Despres et al,
2000; Hildebrandt et al, 2000; Martin-Henao et al, 2000).
None, however, has made a direct performance comparison
using identical PBSC products. Consequently, conclusions
regarding the relative merits of these devices are difficult to
make in view of differences between both the study centres
and, more importantly, the apheresis products processed on
the machines. In this study, we have directly compared
these two devices by selecting CD34
+
cells on each device
simultaneously. This was done by pooling and splitting
equally two PBSC harvests collected on successive days from
10 separate patients and processing half on the CliniMACS
device and half on the Isolex 300i device. The differences
observed in the performance of the two devices are now
reported with respect to ease of use, product purity, CD34
+
yield and residual T- and B-cell content. Additionally, we
have investigated the in vitro functional capacity of the two
CD34 purified products and report the differences observed
in the yield of colony-forming units (CFUs) and the
expansion of lineage-committed cells in liquid culture.
PATIENTS AND METHODS
Patients studied and PBSC collection. Following approval
from the local ethics committee and with informed patient
consent, 10 subjects were studied. Eight patients had a
diagnosis of multiple myeloma, of which five were mobilized
with cyclophosphamide (1Æ5 g/m
2
n¼3, 4 g/m
2
n¼2)
and granulocyte colony-stimulating factor (G-CSF), and
three were mobilized with ESHAP (etoposide, cisplatin,
cytarabine and methylprednisolone) and G-CSF as described
(Velasquez et al, 1994; Watts et al, 2000). One patient had
non-Hodgkin’s lymphoma and was mobilized with ESHAP
and G-CSF. The other subject was a normal donor for a
patient requiring an allograft and was mobilized with G-CSF
alone. Aphereses were performed on two successive days
using either the COBE Spectra (COBE Laboratories, Glou-
cester, UK) (n¼2) or the Baxter CS3000 (Baxter Health-
care, Deerfield, IL) (n¼8). Patients were eligible for the
study only if the combined CD34 count in the two
successive harvests was greater than 3 ·10
6
/kg.
Harvest split and cell processing. The first harvest was
diluted in an equal volume of autologous plasma and stored
overnight at room temperature in gas-permeable bags in
accordance with CliniMACS recommendations for harvest
storage, with the cell count not exceeding 2 ·10
8
/ml. The
next day, it was pooled with the second harvest and the
resultant combined harvest was mixed and then split in to
two equal volumes for CD34 processing. Both the Clini-
MACS and Isolex 300i machines were operated according to
their standard operating protocols and no allowance was
made for high CD34 target counts. The CliniMACS machine
was used in conjunction with the COBE 2991 cell washer.
Standard kits and reagents were used according to manu-
facturers’ guidelines. Version 2Æ5 software was used with
the Isolex 300i and version 2Æ1 software with the Clini-
MACS.
Flow cytometry. Flow cytometry was performed using an
EPICS Elite flow cytometer (Beckman-Coulter, High Wyco-
mbe, UK). Harvest and purified product CD34 cell numbers
were determined as described previously (Pollard et al,
1999). Phycoerythrin (PE)-conjugated anti-CD34 and flu-
orescein isothiocyanate (FITC)-conjugated anti-CD45 were
from Becton-Dickenson (San Jose, CA, USA). Harvest and
purified product T- and B-cell contents were determined
using a four-colour flow cytometric method. Briefly, 5 llof
apheresis harvest, diluted 1:20 in AB serum (approximately
1·10
6
cells), or 200 ll of product (also approximately
1·10
6
cells), was incubated for 30 min with PC5-conju-
gated anti-CD3, RD1 conjugated anti-CD4, FITC conjugated
anti-CD8 and energy-coupled dye (ECD) conjugated anti-
CD19 (Beckman-Coulter). Similar numbers of cells were
incubated with appropriate control antibodies. Samples
were then treated with Q-prep (Beckman-Coulter) to lyse red
cells and washed in phosphate-buffered saline (PBS) before
flow cytometry where a minimum of 50 000 events were
recorded. To determine retention of mouse immunoglobulin
on the surface of processed cells, 100 ll of product was
incubated for 15 min with a FITC-conjugated anti-mouse
immunoglobulin (Ig) (Dako, Ely, UK). Apoptotic cells in the
freshly purified products were detected by determining
Annexin V binding using a commercial kit (Annexin-
V-FLUOS Staining kit, Boehringer Mannheim, Germany)
in accordance with manufacturer’s instructions.
Colony assays. Colony-forming unit content of the har-
vest and product was assessed in semi-solid, methylcellu-
lose-based media (MethoCult H4230; Stem Cell
Technologies, Vancouver, Canada) with 20% added Iscove’s
modified Dulbecco’s medium (IMDM) (Life Technologies,
Paisley, UK), stem cell factor (SCF, 10 ng/ml) (Sigma, Poole,
UK), interleukin 3 (IL-3, 30 ng/ml), granulocyte/monocyte
colony-stimulating factor (GM-CSF, 25 ng/ml) (both from
Sandoz, Berne, Switzerland), G-CSF (25 ng/ml) (Amgen,
Thousand Oaks, CA, USA) and erythropoietin (EPO, 3 U/ml)
(Roche, Basel, Switzerland). Briefly, cells from the apheresed
harvests were suspended in 2Æ5 ml of media at
25 000 cells/ml and were cultured split between four
0Æ5 ml wells in a 24-well plate at 37C and 4% CO
2
. Cells
from the CD34-purified products were suspended in media
at a concentration of 500 cells/ml and were plated out
similarly. Colonies were counted on d 14.
Liquid culture of progenitor cells. CD34
+
-selected cells were
placed in culture in one well of a six-well plate on the day of
processing at a density of 1–2Æ5·10
5
/ml in IMDM, 20%
fetal calf serum (FCS) (Life Technologies) and 1% penicillin
118 M. J. Watts et al
2002 Blackwell Science Ltd, British Journal of Haematology 118: 117–123

and streptomycin (Sigma) at 37C and 5% CO
2
. To generate
erythroid cells, this medium was supplemented with
EPO (2 U/ml), SCF (20 ng/ml) and IL-3 (1 ng/ml). To
generate myeloid cells medium was supplemented with SCF
(20 ng/ml), IL-3 (20 ng/ml) and G-CSF (100 ng/ml).
During culture, cells were maintained at a density less than
1·10
6
/ml by dilution with medium supplemented with
growth factors whenever necessary. Cell counts and
estimations of viability using Trypan blue were performed
on d 0, 2, 4, 6 and 9 of the culture. On d 9 of the culture,
cytospin preparations demonstrated that cells in the eryth-
roid culture were predominantly at early to intermediate
normoblast stage and that cells in the myeloid culture were
predominantly at the myelocyte stage of differentiation
(data not shown).
Transendothelial migration assays. Assays were performed
as described (Yong et al, 1998). Briefly, 3 ·10
5
CD34-
selected cells were placed in a 3-lm Transwell filter (Costar,
Corning, NY) containing a confluent endothelial cell
monolayer. They were allowed to migrate overnight, with
100 ng/ml stromal-derived growth factor (SDF-1, Pepro-
Tech, London, UK) as a chemoattractant. At the end of the
experiment, transmigrated cells were recovered from the
lower compartment and set up in colony assays, as detailed
above. Percentage migration of granulocyte/macrophage-
CFU (CFU-GM) and erythroid burst-forming units (BFU-E)
was calculated from the colony content of aliquots of the cell
suspension originally seeded on to the filters.
RESULTS
Product CD34
+
cell purity and yield
Ten pairs of harvests were available for processing. Each
pair was pooled, gently mixed and then split equally. The
median total nucleated cell count in the resultant 20
harvests was 28 ·10
9
(range 15–56 ·10
9
) and the
median CD34
+
cell content was 1Æ8% (range 0Æ4–6Æ4%),
with no significant difference between matched pairs.
Following processing of one of each matched pair on each
of the two devices, the product obtained from the CliniMACS
machine was found to have significantly higher CD34
+
purity, although the yield of CD34
+
cells from the harvests
was similar for both machines (Table I). It was noted that in
the two harvest pairs in which the absolute CD34
+
cell
count was above 1Æ0·10
9
per individual split harvest, the
yield of CD34
+
cells was low from both machines (27% and
45% for the CliniMACS, and 40% and 42%, respectively, for
the Isolex 300i). The manufacturer’s guidelines for maxi-
mum capture of target CD34
+
cells on the CliniMACS device
(6 ·10
8
cells) was exceeded in these cases. In one other
case, CD34
+
cell yield was very low at 25%. On this
occassion, the CliniMACS column became blocked by DNA
gel half way through the procedure. In all other cases, the
yield of CD34
+
cells was above 40%.
Product T- and B-cell content
The products were analysed for T- and B-cell content. As
shown in Table II, the median T-cell content of the
CliniMACS product was significantly lower than that of
the Isolex 300i product. The difference in CD4 cell content
was also statistically significant, although the difference in
CD8 content only bordered on statistical significance. This
reduction in overall T-cell content represented a highly
significant median 0Æ9 log further reduction achieved with
the CliniMACS machine by comparison with the Isolex
300i. There was no statistically significant difference in the
B-cell content of the two products.
CliniMACS cells retain surface mouse Ig
As shown in Table III, and consistent with the mechanism
of cell processing, the iron oxide/dextran conjugated mouse
anti-CD34 antibody used to select the cells on the Clini-
MACS device was retained on the surface of the CD34
+
cells,
whereas this was not the case with cells purified on the
Isolex 300i device.
Viability and clonal expansion
The recovery of CFUs by each device was assessed by
comparing the number of CFUs obtained from the processed
product with the number of CFUs contained within the
unprocessed harvest. As shown in Table IV, median overall
CFU recovery was significantly higher in the Isolex 300i
product compared with the CliniMACS product.
Lineage-specific expansion of cells from the selected
products was additionally assessed in liquid culture medium
in eight of the paired products. Equal numbers of viable cells
from each product, as determined by Trypan blue exclusion,
were placed into liquid culture medium with growth factors
that support either myeloid or erythroid cell expansion. As
shown in Fig 1, in conditions that supported the expansion
of either myeloid or erythroid cells, the Isolex 300i product
generated significantly more cells at all time-points up to
d 9 than cells from the CliniMACS product. On d 9 of the
myeloid culture, mean expansion of cells from the Isolex
300i device was 55-fold (range 9–110) compared with a
35-fold (range 3–80) expansion of cells from the CliniMACS
device (P¼0Æ03). On d 9 of the erythroid expansion, Isolex
300I-processed cells had undergone a mean 155-fold (range
7–324) expansion whereas CliniMACS cells had expanded
73-fold (range 6–203) (P¼0Æ03). This difference is likely to
be explained by the apoptosis of a greater proportion of the
CliniMACS processed cells in the 48 h following the start of
the culture, as compared with the Isolex 300i processed
Table I. Purified product CD34 purity and yield (median and
range).
Machine CD34 purity (%) CD34 yield (%)
Isolex 300i 78 (60–93) 60 (40–84)
CliniMACS 90 (79–99) 64 (24–76)
P-value (t-test) 0Æ004 0Æ85
CD34
+
cell content of the harvest and product was
determined by flow cytometry. CD34
+
cell yield was deter-
mined by dividing the number of CD34
+
cells in the product
by the number of CD34
+
cells in the split harvest.
CliniMACS

(v.2Æ1) vs Isolex

300i (v2Æ5) 119
2002 Blackwell Science Ltd, British Journal of Haematology 118: 117–123

cells. Using Trypan blue exclusion as the determinant of cell
viability, on d 2 of the erythroid culture, compared with
d 0, the percentage viability of the CliniMACS-processed
cells had fallen by a median of 23Æ5% (range 11–60%). This
was a significantly greater drop than the median 13%
(range )8 to 57%) drop in viability observed with the Isolex
300I-processed cells (P¼0Æ04).
Further evidence that the CliniMACS products contained
an increased proportion of terminally damaged cells was
seen when the proportion of apoptotic cells in both products
was assessed using an annexin V binding assay. Cells that
exclude Trypan blue may nevertheless be committed to
apoptosis, and a proportion of these may be identified by
annexin V binding. As shown in Table V, the median
percentage of apoptotic cells in the CliniMACS product was
significantly higher than the Isolex 300i product.
The logarithmic expansion curves for both erythroid and
myeloid cultures were parallel following d 2 (Fig 1), sug-
gesting no difference in the proliferative potential of
surviving progenitor cells. Furthermore, there was no
significant difference found between the products from
either machine when the time and intensity of upregulation
Table II. Purified product T- and B-cell content (median and range).
Machine CD3
+
(%) CD4
+
(%) CD8
+
(%) Log CD3
+
depletion CD19
+
(%)
Isolex 300i 0Æ44 (0Æ06–0Æ76) 0Æ11 (0Æ01–0Æ34) 0Æ10 (0Æ00–0Æ44) 3Æ6(3Æ3–4Æ5) 0Æ03 (0Æ00–0Æ75)
CliniMACS 0Æ06 (0Æ01–0Æ39) 0Æ04 (0Æ00–0Æ18) 0Æ01 (0Æ00–0Æ18) 4Æ5(3Æ8–5Æ6) 0Æ12 (0Æ00–0Æ77)
P-value (t-test) 0Æ003 0Æ02 0Æ06 < 0Æ001 0Æ55
T- and B-cell contents of the product were determined using a four-colour flow cytometric assay. The log CD3
+
cell depletion was determined
by comparing the number of CD3
+
cells in the product with the number of CD3
+
cells in the harvest.
Table III. Purified product cell surface anti-CD34
immunoglobulin retention (median and range).
Machine
Cells retaining bound
mouse immunoglobulin (%)
Isolex 300i 0Æ5 (0–1Æ6)
CliniMACS 96 (67–99)
P-value (t-test) < 0Æ0001
Retention of anti-CD34 immunoglobulin was
determined with a flow cytometric assay utilizing
an anti-mouse Ig FITC-conjugated antibody.
Table IV. Purified product CFU yield (median and range).
Machine
Total CFU
yield (%)
BFU-E
yield (%)
CFU-GM
yield (%)
Isolex 300i 48 (18–77) 39 (10–63) 60 (24–95)
CliniMACS 38 (5–63) 26 (3–43) 53 (7–84)
P-value (t-test) 0Æ035 0Æ002 0Æ27
Colony-forming unit content (both BFU-E and CFU-GM) of har-
vest and product were determined by culture of aliquots of cells in
semi-solid methylcellulose culture. The CFU, BFU-E and CFU-GM
yields were determined by dividing the total number of CFUs in the
product by the total number of CFUs in the split harvest.
Fig 1. Myeloid and erythroid expansion in liquid culture.Equal
numbers of viable CD34
+
-selected progenitors, as assessed by Try-
pan blue dye exclusion, were placed in liquid culture medium that
supported either myeloid or erythroid lineage expansion. Viable
cells were counted on the indicated days and fold expansion over
input numbers was determined for CliniMACS and Isolex 300i
processed cells (mean ± SEM).
120 M. J. Watts et al
2002 Blackwell Science Ltd, British Journal of Haematology 118: 117–123

of the erythroid specific marker Glycophorin A during
erythroid expansion or the myeloid marker CD11b during
myeloid expansion was compared (data not shown).
Transendothelial migration assays
In three cases, the transendothelial migratory capacity of
cells from the CliniMACS and Isolex 300i products was
assessed. There was no significant difference in the trans-
migratory capacity of either CFU-GM or BFU-E derived from
either product. Percentage migration of CFU-GM from the
CliniMACS product was 7Æ7±0Æ5% compared with
10Æ8±3Æ7% for the Isolex 300i (mean ± SEM). Similar
results were obtained for BFU-E migration. This further
suggested that the functional performance of surviving
progenitor cells was equivalent.
DISCUSSION
This study is the first to compare directly the CliniMACS and
Isolex 300i clinical scale CD34
+
selection devices. By pooling
and splitting two PBSC harvests obtained on sequential days,
it was ensured that each device processed identical material.
We found that both devices generally worked reliably and
were straightforward to use. Although the time taken to
process the cells using the CliniMACS machine in conjunc-
tion with the COBE 2991 cell washer was about 1 h faster
than the Isolex 300i (approximately 160 min vs 220 min),
the latter machine had the advantage of being a single
platform device. Additionally, during one procedure the
CliniMACS column became blocked by DNA gel, something
that did not occur with the Isolex 300i machine, probably
due to the use of DNAase.
We found the CD34
+
purity of the CliniMACS product to
be significantly higher than that of the Isolex 300i product.
This finding concurs with other published studies that
evaluated the devices individually. These have reported
higher median CD34
+
purities in products from the Clini-
MACS device (97% (n¼71) in Schumm et al (1999) and
98% (n¼30) in Despres et al (2000) than from the Isolex
300i device (82% (n¼13) in Stainer et al (1998); 89%
(n¼51) in Martin-Henao et al (2000); 84% (n¼43) in
Hildebrandt et al (2000); and 85% (n¼30) in Croop et al
(2000). The reported median yields of CD34
+
cells have also
been higher for the CliniMACS device (71%, Schumm et al,
1999; 70%, Despres et al, 2000) than for the Isolex 300i
device (50%, Stainer et al, 1998; 58%, Martin-Henao et al,
2000; 51%, Hildebrandt et al, 2000; and 43%, Croop et al,
2000). However, when we compared the devices directly,
the yields were found to be similar for both machines at 60–
64%. This emphasizes the difficulty in directly comparing
machine evaluation studies performed by different centres
on different patient products with different devices utilizing
different software. It was of note that at CD34
+
input levels
between 1Æ0 and 1Æ5·10
9
, both machines suffered a
similar drop in yield, demonstrating similar saturation
points for CD34
+
recovery.
The median product CD34
+
purities obtained in this
study were at the low end of the ranges previously described
for both devices. This may be due to the use of pooled
harvests in this study, with half being stored overnight at
room temperature, as discussed later. Our experience of
processing fresh harvests, which have not been stored
overnight, is similar to that of others. In 55 consecutive
CD34 selection procedures performed during 2000 and
2001, median CD34
+
purity and yield were 95% and 67%,
respectively, for the CliniMACS device (n¼45), and 90%
and 57%, respectively, for the Isolex 300i device (n¼10).
Previous studies also report greater T-cell depletion with
the CliniMACS device (Schumm et al, 1999) than the Isolex
300i device (Stainer et al, 1998). Our data, which also
shows better T-cell depletion with the CliniMACS device,
confirm this finding. Although the CliniMACS device gave
more efficient T-cell depletion, there was no significant
difference in B-cell depletion between the two devices. It is
not clear why this should be, although it is possible that
inherent differences between T- and B-cell surface proteins
and carbohydrates cause different non-specific interactions
with the two columns, thus altering non-specific retention
of the two cell types.
When the functional capacity of the product was
investigated using a liquid culture ex vivo expansion
method, the CliniMACS product was found to generate
approximately half the number of erythroid and two-thirds
of the number of myeloid precursors at all time-points up to
d 9 compared with the Isolex 300i product. Additionally,
recovery of CFUs from the CliniMACS product was 21% less
than from the Isolex 300i product. This is despite more
CD34
+
cells being put in to both the liquid and semi-solid
cultures from the CliniMACS product than the Isolex 300i
product, because the CD34
+
purity of the CliniMACS
product was significantly higher. If correction had been
made for CD34
+
purity when cultures were set up, the
observed difference would have been even greater.
This finding suggests that in the CliniMACS product there
was a greater proportion of myeloid- and erythroid lineage-
specific CD34
+
cells that were either dead or viable with
respect to Trypan blue exclusion but nevertheless committed
to apoptosis. The reason for this is not clear, although several
possibilities arise. First, the CliniMACS device may non-
specifically retain dead cells or cells committed to apoptosis.
It has been reported that dead cells contaminating a harvest
may non-specifically bind the iron oxide/dextran conjugated
Table V. Purified product apoptotic cell content
(immediately post processing) (median and range).
Machine Annexin V-positive cells (%)
Isolex 300i 18 (12–34)
CliniMACS 28 (15–51)
P-value (t-test) 0Æ007
Aliquots of cells from the purified products were
incubated with annexin V-FLUOS in the appro-
priate binding buffer. Apoptotic cells binding
annexin V were subsequently identified using flow
cytometry.
CliniMACS

(v.2Æ1) vs Isolex

300i (v2Æ5) 121
2002 Blackwell Science Ltd, British Journal of Haematology 118: 117–123

antibody used with the CliniMACS machine (Schumm et al,
1999). Although the same non-specific interaction may
occur with the Isolex 300i device, viable CD34
+
cells are
released from the column with PR34
+
stem cell releasing
agent, which may release fewer non-specifically bound
apoptotic cells. Additionally, overnight storage of cells at
room temperature may have contributed to increasing the
proportion of apoptotic cells in the processed harvests,
consequently exacerbating this tendency. Although this is
the recommended method of storage of products to be run
on the CliniMACS device, when the nucleated cell count in a
fresh harvest is greater than 1Æ5·10
8
/ml, as was the case
in eight out of 10 of the overnight-stored harvests in this
study, it has been reported that overnight storage at room
temperature increases the proportion of apoptotic cells,
probably due to reduced product pH (Greer et al, 1997;
Ahmed et al, 2000).
An alternative explanation for the reduced functional
capacity of the CliniMACS product is that the mechanical
stresses involved during the CliniMACS processing may
irreversibly damage a proportion of the cells, which are
unaffected by the alternative mechanisms used by the Isolex
300i device. This explanation has been suggested by others
to explain similar findings of reduced viability in the
CliniMACS product (Fritsch et al, 2000), although possible
causes remain speculative.
Another possibility is that the different monoclonal
antibodies used select different subpopulations of CD34
+
cells, with the 9C5 antibody (used with the Isolex 300i
device) selecting a population with greater clonogenic
capacity than that selected by the QBEnd10 antibody (used
with the CliniMACS device). The existence of differences in
the expression of Class 2 Pasteurella glycoprotease-sensitive
CD34 epitopes and Class 3 cleavage-resistant CD34 epitopes
on subpopulations of haemopoietic progenitor cells has been
suggested before (Steen et al, 1996). The QBEnd10 and 9C5
antibodies, however, both recognize Class 2 CD34 epitopes
(Steen et al, 1996; Martin-Henao et al, 2000) and, although
we cannot rule it out, it seems unlikely that our results are
explained by differential subpopulation selection.
In most situations, the differences demonstrated in this
study in the functional capacity of the products from the
two machines will not matter. Surviving progenitor cells
have equivalent in vitro proliferative and migratory capacity
regardless of the machine used. However, where the
number of CD34
+
cells following selection is predicted to
be marginal (i.e. near to the previously determined thresh-
old for uncomplicated autograft engraftment of 2 ·10
6
/kg
(Watts et al, 1997; Watts et al, 2000) the choice of machine
used to process the harvest could become important.
Additionally, because this study demonstrates that a
proportion of processed CD34
+
cells that are intact and
resistant to Trypan blue are already committed to apoptotic
cell death, it is clear that determination of the number of
CD34
+
cells in the product alone is an insufficient assess-
ment of the functional capacity of the product. Additional
functional assays, such as CFU determination, performed in
parallel with immunophenotyping are essential to assess
fully the product’s suitability for transplantation.
ACKNOWLEDGMENTS
This study was funded in part by a research grant from
Baxter. Additionally, T.C.P.S. is supported by a personal
fellowship from the Medical Research Council (UK), and
A.K., K.Y. and D.C.L. are supported by a co-operative group
grant from the Medical Research Council (UK).
REFERENCES
Ahmed, F., Watts, M.J., Ings, S.J. & Flynn, M. (2000) Poor viability
of freshly apheresed PBSC harvests: implications for overnight
storage before cryopreservation or additional manipulative pro-
cedures. Blood,98, 380a (abstract 1639).
Ardeshna, K.M., Corney, C.P., Ings, S.J., Watts, M.J., Linch, D.C. &
Devereux, S. (2000) A clinically applicable method for the ex vivo
generation of antigen-presenting cells from CD34
+
progenitors.
Vox Sanguis,79, 46–52.
Cagnoni, P.J. & Shpall, E.J. (1996) Mobilization and selection
of CD34-positive hematopoietic progenitors. Blood Reviews,10,
1–7.
Croop, J.M., Cooper, R., Seshadri, R., Fernandez, C., Graves, V.,
Kreissman, S., Smith, F.O., Cornetta, K., Williams, D.A. &
Abonour, R. (2000) Large-scale mobilization and isolation of
CD34
+
cells from normal donors. Bone Marrow Transplanation,
26, 1271–1279.
Despres, D., Flohr, T., Uppenkamp, M., Baldus, M., Hoffmann, M.,
Huber, C. & Derigs, H.G. (2000) CD34
+
cell enrichment for
autologous peripheral blood stem cell transplantation by use of
the CliniMACs device. Journal of Hematotherapy and Stem Cell
Research,9, 557–564.
Devereux, S., Corney, C., Macdonald, C., Watts, M., Sullivan, A.,
Goldstone, A.H., Ward, M., Bank, A. & Linch, D.C. (1998) Fea-
sibility of multidrug resistance (MDR-1) gene transfer in patients
undergoing high-dose therapy and peripheral blood stem cell
transplantation for lymphoma. Gene Therapy,5, 403–408.
Fritsch, G., Scharner, D., Froschl, G., Buchinger, P., Peters, C.,
Matthes, S., Witt, V. & Gadner, H. (2000) Selection of CD34-
positive blood cells for allogeneic transplantation: approaches to
optimize D34-Cell recovery, purity, viability, and T-cell depletion.
Onkologie,23, 449–456.
Gratwohl, A., Passweg, J., Baldomero, H. & Urbano-Ispizua, A.
(2001) Hematopoietic stem cell transplantation activity in
Europe 1999. Bone Marrow Transplantation,27, 899–916.
Greer, M., Hami, L., Zumerchik, P., Green, T., Goodrich, R. & Shpall,
E.J. (1997) Viability and recovery of PBSC products stored at
room temperature is dependent on storage conditions. Blood,90,
290a (abstract 920).
Hildebrandt, M., Serke, S., Meyer, O., Ebell, W. & Salama, A. (2000)
Immunomagnetic selection of CD34
+
cells: factors influencing
component purity and yield. Transfusion,40, 507–512.
Kawano, Y., Takaue, Y., Watanabe, A., Takeda, O., Arai, K., Itoh,
E., Ohno, Y., Teshima, T., Harada, M., Watanabe, T., Okamoto,
Y., Abe, T., Kajiume, T., Matsushita, T., Ikeda, K., Endo, M.,
Kuroda, Y., Asano, S., Tanosaki, R., Yamaguchi, K., Law, P. &
McMannis, J.D. (1998) Partially mismatched pediatric trans-
plants with allogeneic CD34(
+
) blood cells from a related donor.
Blood,92, 3123–3130.
Martin-Henao, G.A., Picon, M., Amill, B., Querol, S., Gonzalez, J.R.,
Martinez, C., Martino, R., Ferra, C., Brunet, S., Granena, A.,
Sierra, J. & Garcia, J. (2000) Isolation of CD34
+
progenitor cells
from peripheral blood by use of an automated immunomagnetic
selection system: factors affecting the results. Transfusion,40,
35–43.
122 M. J. Watts et al
2002 Blackwell Science Ltd, British Journal of Haematology 118: 117–123

Pollard, Y., Watts, M.J., Grant, D., Chavda, N., Linch, D.C. &
Machin, S.J. (1999) Use of the haemopoietic progenitor cell count
of the Sysmex SE-9500 to refine apheresis timing of peripheral
blood stem cells. British Journal of Haematology,106, 538–544.
Reiffers, J., Cailliot, C., Dazey, B., Attal, M., Caraux, J. & Boiron, J.M.
(1999) Abrogation of post-myeloablative chemotherapy neutro-
penia by ex vivo expanded autologous CD34-positive cells. Lancet,
354, 1092–1093.
Schmitz, N., Linch, D.C., Dreger, P., Goldstone, A.H., Boogaerts,
M.A., Ferrant, A., Demuynck, H.M., Link, H., Zander, A. & Barge,
A. (1996) Randomised trial of filgrastim-mobilised peripheral
blood progenitor cell transplantation versus autologous bone-
marrow transplantation in lymphoma patients. Lancet,347,
353–357.
Schumm, M., Lang, P., Taylor, G., Kuci, S., Klingebiel, T., Buhring,
H.J., Geiselhart, A., Niethammer, D. & Handgretinger, R. (1999)
Isolation of highly purified autologous and allogeneic peripheral
CD34
+
cells using the CliniMACS device. Journal of Hematothera-
py,8, 209–218.
Stainer, C.J., Miflin, G., Anderson, S., Davy, B., McQuaker, I.G. &
Russell, N.H. (1998) A comparison of two different systems for
CD34
+
selection of autologous or allogeneic PBSC collections.
Journal of Hematotherapy,7, 375–383.
Steen, R., Tjonnfjord, G.E., Gaudernack, G., Brinch, L. & Egeland, T.
(1996) Differences in the distribution of CD34 epitopes on normal
and haemopoietic progenitor cells and leukaemic blast cells.
British Journal of Haematology,94, 597–605.
Urbano-Ispizua, A., Rozman, C., Martinez, C., Marin, P., Briones, J.,
Rovira, M., Feliz, P., Viguria, M.C., Merino, A., Sierra, J., Mazzara,
R., Carreras, E. & Montserrat, E. (1997) Rapid engraftment
without significant graft-versus-host disease after allogeneic
transplantation of CD34
+
selected cells from peripheral blood.
Blood,89, 3967–3973.
Velasquez, W.S., McLaughlin, P., Tucker, S., Hagemeister, F.B.,
Swan, F., Rodriguez, M.A., Romaguera, J., Rubenstein, E. &
Cabanillas, F. (1994) ESHAP: an effective chemotherapy regimen
in refractory and relapsing lymphoma: a 4-year follow-up study.
Journal of Clinical Oncology,12, 1169–1176.
Watts, M.J., Sullivan, A.M., Ings, S.J., Leverett, D., Peniket, A.J.,
Perry, A.R., Williams, C.D., Devereux, S., Goldstone, A.H. &
Linch, D.C. (1997) Evaluation of clinical scale CD34
+
cell puri-
fication: experience of 71 immunoaffinity column procedures.
Bone Marrow Transplantation,20, 157–162.
Watts, M.J., Ings, S.J., Leverett, D., MacMillan, A., Devereux, S.,
Goldstone, A.H. & Linch, D.C. (2000) ESHAP and G-CSF is a
superior blood stem cell mobilizing regimen compared to cyclo-
phosphamide 1.5 g/m
2
and G-CSF for pre-treated lymphoma
patients: a matched pairs analysis of 78 patients. British Journal of
Cancer,82, 278–282.
Yong, K.L., Watts, M., Shaun, T.N., Sullivan, A., Ings, S. & Linch,
D.C. (1998) Transmigration of CD34
+
cells across specialized and
non-specialized endothelium requires prior activation by growth
factors and is mediated by PECAM-1 (CD31). Blood,91, 1196–
1205.
CliniMACS

(v.2Æ1) vs Isolex

300i (v2Æ5) 123
2002 Blackwell Science Ltd, British Journal of Haematology 118: 117–123