Aplastic Anemia: Evidence for Dysfunctional Bone Marrow

From www.bloodjournal.org by guest on January 26, 2015. For personal use only.
Aplastic Anemia: Evidence for Dysfunctional Bone Marrow Progenitor Cells
and the Corrective Effect of Granulocyte Colony-Stimulating Factor
In Vitro
By John Scopes, Stephen Daly, Richard Atkinson, Sarah E. Ball, Edward C. Gordon-Smith, and Frances M. Gibson
We investigatedthe effects of granulocyte-macrophagecolony-stimulating factor, interleukin-3, stem cell factor, interleukin-6, and granulocyte colony-stimulating factor (GCSF) alone, and in combination, on the clonogenic potential
of normal and aplastic anemia (AA) bone marrow mononuclear cells (BMMC) and CD34+ cells. AA BMMC consistently
produced a significantly lower absolute number of colonies
than normal, but, when account was taken of the reduced
proportion of CD34+cells in AA BM, there was no significant
differencein terms of cloning efficiency (CE).However, when
removed from the influence of accessory cells, the CE of
AA CD34+ cells decreased significantly more than normal,
indicating a defect in their function, either in terms of dependence on accessory cell-derived factors or susceptibility to
cell damage when sorted. Of the factors studied, G-CSF had
the most significant effect on the response of CD34+ cells
from both groups when removed from their accessory cells.
This was particularly true for AA CD34+ cells, whose response to cytokine stimuli containing G-CSF enabled them
to match the response of normal CD34+ cells.
0 1996 by The American Society of Hematology.
that, in combination with Epo, SCF, but not IL-3, stimulated
colony formation to up to 45% of normal values in 1 1 of 19
cases. Maciejewski et al’ investigated the clonogenic potential of 41 acute, untreated AA patients (26 severe and 15
moderate) and 40 hematologically recovered patients by
assaying BM mononuclear cells (BMMC) and FACS-enriched CD34+ cells in a clonogenic culture system using IL3, GM-CSF, SCF, and Epo. The acute AA samples were
found to have reduced clonogenic potential (BMMC and
purified CD34+ cells), whereas purified CD34’ cells from
the hematologically recovered AA patients showed an increased clonogenic potential, compared with the acute patients, and their unsorted BMMC displayed normal clonogenic efficiency.’ In a study of 26 AA patients with a range
of disease severity, we have shown that SCF synergized with
GM-CSF, IL-3, and Epo in clonogenic culture to stimulate
colony formation, in some cases to within the normal range.”
Response to SCF correlated with disease severity, with those
showing the greatest response tending to be less severe, perhaps indicating that a normal response is possible, given the
correct cytokine stimulation, if enough progenitor cells are
In the present study, we investigated the effects of GMCSF, IL-3, G-CSF, SCF, and IL-6, alone and in combination,
on the colony formation of BMMC and purified CD34+ cells
from normal and AA BM. The aims of this study were as
follows. (1) We wanted to investigate the effects of various
cytokine stimuli on cloning efficiency (the proportion of
CD34+ cells giving rise to colonies in clonogenic culture)
TUDIES OF APLASTIC anemia (AA) bone marrow
(BM) in long-term BM culture (LTBMC) have provided evidence for a stem cell defect in AA by showing an
absence of or reduction in the generation of hematopoietic
progenitor cells, despite normal stroma f0rmati0n.l.~Crossover experiments, involving the recharge of irradiated preformed stroma with a second marrow, have permitted the
separate analysis of stromal and hematopoietic stem cell
function and, in two such studies, a stem cell defect has
again been implicated in AA.2.5Other evidence for an intrinsic stem cell defect has come from the isolation of CD34+
marrow cells from patients with AA using immunomagnetic
beads and fluorescence activated cell sorting (FACS). These
cells have been found to be reduced in number and to generate fewer colonies than normal in clonogenic culture.637AA
CD34+ cells grown in LTBMC on normal stroma show an
absence in the generation of progenitor cells, implying a
deficiency in long-term culture initiating cells (LTCIC) believed to be of the phenotype CD34+CD33-.* Using immunophenotyping, we have previously shown that the
CD34TD33- subset is reduced in AA BM, although the
clonogenic potential of these cells has yet to be determined.’
Hematopoiesis involves a cascade of differentiation steps
that is driven and regulated by a group of glycoproteins
known as hematopoietic cytokines.’O,”Through multiple interactions, both negative and positive, they regulate the proliferation and differentiation of hematopoietic stem cells.”
In clinical trials conducted on AA patients using granulocyte colony-stimulating factor (G-CSF), granulocyte-macrophage colony-stimulating factor (GM-CSF), and interleukin3 (IL-3), no long-lasting effect on BM function has been
Generally, patients show a transient increase
in peripheral granulocyte counts, but those with no residual
marrow function rarely benefit.I6
The effects of cytokines on AA BM has been studied in
vitro. In clonogenic culture, colony formation from AA BM
is consistently low. Bagnara et all8 investigated the effect of
stem cell factor (SCF) on BM from AA patients and found
a modest improvement in colony growth when SCF was
added in combination with GM-CSF, IL-3, and erythropoietin (Epo). In similar studies, Wodnar-Filipowicz et all9found
that, in the majority of cases, SCF synergized with GMCSF, IL-3, G-CSF, and Epo to stimulate colony formation
to up to 30% of normal values and Ammo et alZofound
Blood, Vol 87, No 8 (April 15). 1996: pp 3179-3185
From the Division of Haematology, Department of Cellular and
Molecular Sciences, and the Department of Public Health Sciences,
St George’s Hospital Medical School, London, UK.
Submitted September 18, 1995; accepted November 29, 1995.
Supported by the Marrow Environment Fund, the Leukaemia Research Fund, and Action Research.
Address reprint requests to John Scopes, Division of Haematology, Department of Cellular and Molecular Sciences, St George’s
Hospital Medical School, London SW17 ORE, UK.
The publication costs of this article were defrayed in part by page
charge payment. This article must therefore be hereby marked
“advefiisement” in accordance with 18 U.S.C. section 1734 solely to
indicate this fact.
0 1996 by The American Society of Hematology.
From www.bloodjournal.org by guest on January 26, 2015. For personal use only.
Table 1. Clinical Details of Patients With AA
No. of patients
Age lyr)
Sex (M:F)
Disease duration (mo)
On therapy at time of study (yesno)
Neutrophils (lOg/L)
Transfusion dependence (yes:no)
Platelets (10%)
Hb (g/dL)
Previous therapy (yesno)
47 (38-68)
28 (1-99)
0.3 (0.2-0.5)
25 (11-31)
9 (1-24)
0.8 (0.7-2.51
19 (14-23)
9.7 (8.9-10.5)
45 (40-48)
8 (6-11)
3.4 (2.4-7.2)t
49 (30-68)
1 1 (11-11)
1.6 (1.2-2.0)
56 (37-75)
35 (20-76)
63 (2-336)
3.4 (2.6-15)
195 (98-2841
13.6 (9.3-16.2)
Patients were grouped according to severity or response at the time of study. Severity defined by Camitta et aLZ3Figures given are the median
with the range in parentheses. Platelet count and Hb are only given for TI patients.
Abbreviations: VS, very severe; S, severe; NS, nonsevere; NA, not applicable; PRTD, partial response (transfusion dependent); PRTI, partial
response (transfusion independent); CR, complete response.
* One S and one NS patient had evidence of paroxysmal nocturnal hemoglobulinuria (PNH).
t Patient with neutrophil count of 7.2 x 109/Lwas on G-CSFtherapy at the time of the study.
Three patients spontaneously recovered, including 1 patient with evidence of PNH.
and to determine which give the greatest response. ( 2 ) We
wanted to assess the effect of the presence of accessory cells
on the cloning efficiency of BMMC, compared with purified
CD34+ cells. (3) Having previously shown that the number
of hematopoietic progenitors in AA BM is reduced,' we
wanted to investigate whether this reduced population is
also dysfunctional and whether AA BM responds to various
cytokine stimuli in the same way as normal BM to clarify
whether a functional, as well as quantitative, defect is implicated in the disease.
Normal BMspecimens. Normal BM cells (n = 14) were obtained
from iliac crest marrow aspirates of hematologically normal donors
after informed consent. Marrow aspirates were diluted 1:1 in Iscove's
Modification of Dulbecco's medium (IMDM; GIBCO, Paisley, Scotland), supplemented with 100 IU/mL penicillin-streptomycin
(GIBCO) and 10 IU/mL preservative-free heparin (Leo Laboratories,
Ltd, Princes Risborough, Buckinghamshire, UK), and then centrifuged on Ficoll-hypaque (Pharmacia, St Albans, Hertfordshire, UK)
to obtain the BMMC, which were washed twice in the above medium.
AA EM specimens. BM was obtained from patients with AA (n
= 25; 23 patients, 2 on 2 separate occasions) referred to St George's
Hospital, with the diagnosis being made on the basis of strict criteria.22The severity of the disease was made according to the criteria
The BMMC were separated as above. The clinical
of Camitta et
status of the AA patients is shown in Table 1.
Staining for CD34' cells. BMMC were washed twice in phosphate-buffered saline (PBS; Sigma, Poole, Dorset, UK) supplemented with 1% fetal calf serum (FCS). The cells were then sequentially incubated with (1) human y-globulins (30 pg/106 BMMC;
Sigma) for 10 minutes at 4°C; (2) anti-CD34 antibody (10 pL/106
BMMC; QBEND 10; Immunotech S.A., Marseille, France) for 30
minutes at 4°C. followed by two washes in PBS, supplemented as
above; and (3) fluorescein isothiocyanate (FIE)-conjugated rabbit
F(ab), antimouse IgG (6 &/IO6 BMMC; Dako Ltd, High Wycombe,
Buckinghamshire, UK) for 30 minutes at 4"C, followed by two
washes, as above. The cells were then resuspended at a concentration
of 106/mL in PBS, supplemented as above and kept at 4°C to be
ready for FACS sorting. A negative control was set up in parallel
for each sample, omitting anti-CD34 antibody from the incubation.
FACS sorting of CD34' cells. CD34+ BMMC were purified
using a FACStar Plus cell sorter (Becton Dickinson, UK Ltd, Cowley, Oxford, UK). The whole nucleated cell population, excluding
red blood cells and cell debris, was analyzed and scattergrams were
generated by combining right-angle light scatter with fluorescence.
Regions were drawn around clear-cut populations having low rightangle scatter and high fluorescence. Cells falling within this region
were sorted firstly in Enrich mode at 5,000 to 6,000 celIs/s and
subsequently in Normal-R mode at 200 to 300 cells/s. Cells were
sorted into IMDM supplemented with 10% FCS. The frequency of
CD34' cells was assessed both before and after FACS sorting and
the purity and enrichment of the sorted population were then calculated.
Committed EM progenitor assay. Normal and AA BMMC (1 Os)
or CD34' cells (1.67 X lo3) were cultured in 1 mL IMDM supplemented with 30% FCS, 1% deionized bovine serum albumin (BSA;
Sigma), 1 0-4 m o m mercaptoethanol, and 0.9% methylcellulose
(Stemcell Technologies Inc, Terry Fox Laboratory, Vancouver, British Columbia, Canada), in 35-mm petri dishes. The following recombinant growth factors were added either alone or in comination: 10
ng/mL GM-CSF (1.39 x 10' U/mg; Sandoz Pharma, Camberley,
Surrey, UK), 100 ng/mL IL-3 (4.9 x lo6 Ulmg; Sandoz Pharma),
100 ng/mL SCF (IO5 U/mg; Immunex Corp, Seattle, WA), 100 ng/
mL G-CSF (10' U/mg; Amgen UK Ltd, Cambridge, UK), and 100
ng/mL IL-6 (5.2 X IO' U/mg; Sandoz Pharma). All growth factors
were aglycosolated products made in Escherichia coli, with the exception of SCF, which was glycosolated and made in yeast. Cultures
were also set up in the absence of growth factors. All reagents were
pretested for their ability to support optimal growth. The following
cytokine combinations were studied: (1) control (no cytokines); (2)
IL-6; (3) SCF; (4) GM-CSF; (5) IL-3; (6) G-CSF; (7) IL-3 + IL-6;
(8) IL-3 + SCF; (9) IL-3 + GM-CSF (10) IL-3 + G-CSF; (1 1 ) IL3 + GM-CSF + IL-6; (12) IL-3 + SCF + IL-6; (13) IL-3 + GMCSF + SCF; (14) IL-3 + GM-CSF + G-CSF (15) IL-3 + SCF +
G-CSF; (16) IL-3 + GM-CSF + SCF + IL-6; (17) IL-3 + SCF +
IL-6 + G-CSF; (18) IL-3 + GM-CSF + SCF + G-CSF; (19) IL-3
+ GM-CSF + SCF + G-CSF + IL-6.
Cultures were set up in duplicate and incubated at 37°C in 5%
C02/95% air. Epo at 2 UlmL (Eprex; Cilag Ltd, High Wycombe,
Buckinghamshire, UK) was added to all cultures on day 3. Colonyforming unit-granulocyte-macrophage (CFU-GM), burst-forming
unit-erythroid (BFU-E), and CFU-granulocyte, erythroid, monocyte
(CFU-GEM) were counted on day 14 and their counts were pooled
From www.bloodjournal.org by guest on January 26, 2015. For personal use only.
and expressed as either (1) the absolute colony numbed105 BMMC
or CD34' cells or (2) the cloning efficiency (CE), where CE = (no.
of coloniesll0' cells)/(no. of CD34' cellsll0' cells).
Sruristical analysis. Scatter diagrams of colony numbers and
CEs, one for each combination of environment (unsorted and sorted)
and group (AA and normal) across the 19 cytokine stimuli, were
produced to assess the data visually (data not shown). The initial
levels of CD34' cells were also plotted to compare the two groups
in unsorted and sorted samples (data not shown). As a result of the
visual assessment and consideration of the assumptions required for
analysis of variance, the arcsine root transformation was applied to
the CEs before the analyses.
Unsorted and sorted data were analyzed separately, using splitunit analysis of variance (ANOVA; two factors: group and cytokine
stimulus). Where appropriate, adjustments for nonsphericity were
Multiple comparisons of the mean responses to the cytokine stimuli were performed using the Bonferroni inequality to control for
type I error rates.
Differences between normal and AA responses to specific cytokine stimuli were analyzed using unpaired t-tests, if the assumptions
for this test were verified, or Wilcoxon tests, if not.
The effect of each cytokine combination on unsorted and sorted
BM for each sample group was analyzed using a within-subject
ANOVA (two factors: environment and cytokine stimulus).
Differences in response between unsorted and sorted samples
(normal and AA separately) were analyzed using paired t-tests or
Wilcoxon tests.
The number of colonies produced increased as the cytokine stimulus moved from one factor to two, three, etc, for
both normal and AA cells, whether sorted or unsorted. Figure
1 shows the mean absolute number of colonies, with 95%
confidence intervals, produced by normal and AA BM in
response to the 19 cytokine combinations, for unsorted and
sorted BM, respectively. Figure 1A shows the much higher
response of normal, compared with AA, BM to the 19 cytokine combinations. Figure 1B shows that, when sorted, the
response of AA BM is much closer to, although still lower
than, that of normal BM. For many combinations, the 95%
confidence intervals of the means of the two groups overlap.
These results are obviously dependent on the initial level of
CD34+ cells in the samples. Mann Whitney U tests showed
a significant difference between normal (mean, 2.16% ?
0.33%; range, 0.9% to 6.0%) and AA (mean, 0.65% t
0.068%; range, 0.2% to 1.5%) CD34+ levels before sorting
(P < .00002). After sorting, there was no significant difference between normal (mean, 88.86% ? 1.35%; range, 80%
to 95%) and AA (mean, 83.56% ? 3.10%; range, 20% to
97% [only 1/25 sorted AA samples had a purity <62%])
CD34' levels (P = .3311). To control for CD34+ levels
when comparing their functional response to cytokine stimuli, results were expressed in terms of CE. Figure 2 shows
the mean transformed CEs, with 95% confidence intervals,
of normal and AA BM, in response to the 19 cytokine combinations, for unsorted and sorted BM, respectively. It shows
that, for both normal and AA, CE is higher in unsorted than
in sorted BM. In the unsorted samples, the AA means are
much closer to those of the normal samples than in the sorted
0 = Nomu1
J - I
o : , , ,
i 2 3 4 5
7 8 s1o111zi~141~isi7iais
Cytokine stimulus
0 = Nom1
i 2 3 4 5
7 8
~ ~ o ~ ~ i z i ~ ~ ~ i ~ ~ s i
Cytokine stimulus
Fig 1. Absolute numbers of colonies produced by IAl unsorted
normal end AA BMMC and (B)sorted normal and AA CD34* cells in
response to the 19 cytokine combinations. Means are shown wkhin
their respective 95% confidence intervals.Statistically significant differences between normaland AA responsesto specific cytokine atimuli sre indicated by an asterisk IP s .01).
Statistical analysis of unsorted BM. Results of the group
by cytokine stimulus ANOVA showed that the interaction
is insignificant (P = .1448). A series of two sample t-tests
comparing the responses in the two groups for each cytokine
stimulus showed very insignificant results, indicating no differences between the groups for any cytokine stimulus, therefore giving no interaction. Moreover, the overall group effect
was insignificant (P = .3306); there was no significant difference between the groups in terms of CE, irrespective of
cytokine stimulus (Fig 2A). Therefore, the CE of unsorted
AA BM is not significantly different from normal.
However, the cytokine stimulus effect was highly significant (P = .OOOl). One or more cytokine stimuli are, therefore, significantly different from one or more others. To
establish which cytokine combinations are significantly different from which, the normal and AA samples were tested
From www.bloodjournal.org by guest on January 26, 2015. For personal use only.
I 2 3 4 5 6 7 0 9 10111213141516171019
Cytokine stimulus
7, 9, 11, 16, and 19 were clearly significantly different,
whereas 6, 10, 13, 14, and 15 were clearly not (Fig 2B).
The cytokine stimulus main effect is highly significant (P =
,0001). To establish which combinations are significantly
different from which for the normal and AA samples separately, further subject by cytokine stimulus ANOVAs and
Bonferroni multiple comparisons were performed (data not
shown; see Discussion).
Statistical comparison of unsorted and sorted BM. Results of the environment by cytokine stimulus ANOVA performed on each sample group separately showed that, for
both normal and AA, the main effects of cytokine stimulus
(normal, P = . m o l ; AA, P = .Owl) and environment (normal, P = ,008; AA, P = .OOOl> and the cytokine stimulus
by environment interaction (normal, P = .0001; AA, P =
.0001) were all highly significant. To clarify the cause of
significance, each stimulus was selected and the response
in the two environments were compared. Paired t-tests and
Wilcoxon tests were used to compare the two groups. If the
level of significance is again taken as P 5 .01, the results
show that, for both normal and AA BM, the unsorted and
sorted samples are not significantly different for stimuli 10,
14, 15, 17, 18, and 19. Also, in the normal subjects, stimulus
6 showed a nonsignificant difference between the two environments (Fig 3).
We investigated the effects of a wide range of cytokine
stimuli on the clonogenic potential of normal and AA
BMMC and purified CD34+ cells.
(1) It was found that the clonogenic potential of BM is
greatly influenced by the effects of the exogenous cytokines
studied. For the unsorted samples, there was no significant
difference between normal and AA BM in terms of CE,
irrespective of cytokine stimulus (Fig 2A). The multiple
comparison of mean responses to cytokine stimuli, combining normal and AA data, showed no clear pattern emerging
as to which cytokine stimuli stand out in their effect on
CE but only that, generally, the more cytokines present, the
greater the CE. The cytokine production of accessory cells
clearly confounds an already complex picture of response.
We sorted CD34+ cells to remove them from this influence,
laying bare their functional response to the cytokine stimuli
administered in an attempt to highlight differences in this
response between normal and AA.
(2) On doing so, our first finding was that, for both normal
and AA BM, CE decreases after sorting (Fig 3). For each
cytokine stimulus studied, the CE of sorted BM was lower
than that of unsorted BM. This is to be expected, given that
the maximum stimulus contained only five cytokines, with
accessory cells presumably providing a plethora of others.
However, there were several combinations of cytokines that,
although producing a lower response in the sorted than in
the unsorted samples, did not produce a significantly lower
response; for both normal and AA BM, these were 10, 14,
15, 17, 18, and 19 (Fig 3). These cytokine stimuli comprise
all the 2-, 3-, 4-,and 5-factor combinations containing GCSF. In addition, for normal subjects, G-CSF alone (no.
1 2 3 4 5 6 7 0 910111213141516171019
Cytokine stimulus
Fig 2. Transformed CEs of (AJunsorted normal and AA BMMC
and (6) sorted normal and AA CD34' calls in response to the 19
cytokine combinations.Means are ahown within their respective 95%
confidence intervals. Statistically significant differences batween
normal and AA responses to specific cytokine stimuli are indicated
by an asterisk I f c .01).
for heterogeneity of variance before being collapsed into a
single group, after which a multiple comparison of the mean
responses to the cytokine stimuli was performed using the
Bonferroni inequality to control overall type I error rates
(data not shown; see Discussion).
Statistical analysis of sorted BM. Results of the group
by cytokine stimulus ANOVA showed that the interaction
is significant (P = .0148). Therefore, some cytokine combinations produce significantly different effects in the two
groups. The group main effect is also highly significant (P
= ,0069). After sorting, therefore, the CE of AA CD34+
cells was statistically significantly different from normal. To
establish the cause of this difference, a series of two-sample
Wilcoxon and t-tests of the transformed data were conducted
for each cytokine combination. Given the number of tests,
if P 5 .01 is taken as the level of significance, stimuli 4, 5,
From www.bloodjournal.org by guest on January 26, 2015. For personal use only.
0 = Unsorted
* Sorted
0 = Unsorted
Fig 3. TransformedCEs of normal (A) and AA IB) unsorted BMMC
and sorted CD34' cells in response to the 19 cytokine combinations.
Means are shown within their respective 95% confidence intervals.
Statistically significant differences between the responses of unsorted and sorted BM to specific cytokine stimuli are indicated by an
asterisk (Pc .01).
6) produced a nonsignificant difference between the two
The effect of G-CSF could be interpreted in two ways:
either G-CSF is a major component of the cytokine stimulus
of accessory cells or its effects overlap with those of other
cytokines produced by these cells; G-CSF is an early as well
as late acting ~ y t o k i n e ~and
~ - *is~known to be produced by
LTBMC stromal cells.27 In either case, exogenous G-CSF
clearly plays a key role in substituting for the effects of
accessory cells in stimulating the CE of purified normal and
AA CD34+ populations.
(3) In agreement with previous studies, we found that AA
BM contains significantly less CD34+ cells than normal BM.
Using FACS, we enriched the CD34' cells in BM from
the two groups to a level where they no longer differed
significantly and to a purity that enabled us to investigate
and compare their clonogenic potential.
Unsorted, AA BM produced a significantly lower absolute
number of colonies than did normal BM, regardless of the
cytokine stimulus (Fig IA). However, this difference would
appear to be a reflection of the reduced level of CD34+ cells
in AA BM, because, in terms of CE, there was no significant
difference between normal and AA BM. For some cytokine
combinations, in fact, AA CE was equal to, or higher than,
normal CE (Fig 2A). One might argue from these findings
that the CD34' cells of these AA patients were not dysfunctional, with the reduced number of colonies produced by
their BM simply being a result of their reduced numbers of
CD34' cells. However, it is possible that a dysfunction in
AA CD34' cells could be masked by the influence of compensating accessory cells. Using FACS, we removed CD34+
cells from the influence of accessory cells, increasing their
clonogenic dependence on the administered exogenous cytokine stimuli.
Although the CE of both normal and AA BM decreased
significantly when sorted, the CE of AA BM decreased significantly more than did that of normal BM (Fig 2B). Not
only do AA CD34' cells have a lower CE than normal
after sorting, but their pattern of response to the cytokine
combinations also differs from normal. This could be explained in two ways: (1) the mechanics of cell sorting might
in some way damage CD34+ cells, perhaps by decreasing
cytokine receptor expression, thereby lowering CE (AA
CD34+ cells may be more "fragile," more susceptible to
this kind of damage); or, (2) as suggested above, AA CD34'
cells might be more dependent on the stimulation and control
of accessory cells than normal CD34+ cells. In either case,
it appears that AA CD34' cells differ from normal CD34'
cells and are in some way dysfunctional. This was examined
further in a series of two-sample Wilcoxon and f-tests, which
showed that normal and AA sorted samples are clearly significantly different for stimuli 4, 5, 7, 9, 11, 16, and 19, but
not significantly different for 6, 10, 13, 14, and 15 (Fig 2B).
G-CSF again seems to play a key role here, because it is
present in 6, 10, 14, and 15, implying that it has an equally
powerful effect on the CE of AA CD34' cells as it does on
normal CD34+ cells. The fact that stimulus 19 contains GCSF but does not stimulate normal and AA CD34' cells
equally may not be contradictory, because this is the combination containing the most cytokines and the effect of GCSF is, therefore, less noticeable. IL-3 + GM-CSF + SCF
(no. 13) also appears to stimulate both populations equally.
To establish which cytokine combinations are significantly different from which, for the normal and AA samples
separately, multiple comparisons of the mean responses were
conducted. Generally, normal CD34+ cells showed a much
wider range of response to the stimuli than AA CD34' cells,
implying that they are more responsive to the synergistic
effects of the cytokines. However, there were some instances
in which the AA CD34' samples show significant differences, whereas the normal CD34' cells do not. Thus, for
AA, but not normal, (1) the effect of G-CSF is significantly
greater than that of IL-3 + IL-6, IL-3 + GM-CSF, and IL3 + GM-CSF + IL-6 and (2) the effect of IL-3 + GM-CSF
From www.bloodjournal.org by guest on January 26, 2015. For personal use only.
+ SCF is significantly greater than that of both IL-3 + IL6 and IL-3 + GM-CSF. It is clear from Fig 2B that, of the
five cytokines studied, G-CSF has by far the greatest overall
effect on AA, as well as normal, CD34+ cells. Moreover, in
contrast to normal, IL-3 seems to have little effect on AA
CD34' cells, either alone or in combination with IL-6 and/
or GM-CSF. However, in combination with GM-CSF + SCF
(stimulus no. 13), IL-3 does have an effect; this was the only
cytokine combination not containing G-CSF that stimulted
AA CD34' cells to the same extent as normal CD34' cells
(see above).
As already stated, in agreement with other groups, we
found that AA EM contains significantly less CD34' cells
than normal BM. There was a correlation between disease
severity and proportion of CD34+ cells: all 3 severe AA
patients had levels less than the AA mean (0.65% 5
0.068%);of those with complete response (CR), 7 of 11 had
levels greater than the mean, whereas the remaining 4 were
just below it ( 2 with 0.6% and 2 with 0.5%). Of the 25
samples analyzed, 8 were from patients who had received
no previous therapy, including 3 on presentation ( 2 nonsevere and 1 severe; Table 1). There was also a degree of
correlation between disease severity and response to cytokine stimuli, in agreement with previous work.' All 3 severe
AA samples fell below the normal CE 95% confidence interval range, both before and after sorting. Indeed, 2 of 3 did
not show any response, regardless of cytokine stimulus. Of
the 9 samples falling within the normal range, 6 were in CR
and 3 were nonsevere. Only I in CR was out of the normal
range before and after sorting. However, there were 4 in CR
and 2 in partial response (PR) that, although falling within
the normal range before sorting, were outside the normal
range after sorting. This may support the previously reported
abnormal in vitro growth of EM from patients who have
responded to treatment.' The remaining samples (2 nonsevere and 3 in PR) tended to fall outside the normal range
both before and after sorting. In a few cases, insufficient
numbers of cells were obtained after sorting to allow the
analysis of response to all cytokine combinations.
In summary, we found that the reduced CD34+ compartment of AA BM is, in addition, dysfunctional. G-CSF appears to compensate for this dysfunction to some extent
and there may be other compensating factors produced by
accessory cells, resulting in the insignificant difference between normal and AA CE before sorting. This dependence
on accessory cells may account for the significant decrease
in AA CE seen after sorting, which may also be partly, or
wholly, attributable to the fragilitylinstability of AA CD34'
cells compared with normal CD34' cells.
G-CSF makes a significant contribution to the response
of CD34' cells from both groups when removed from the
influence of accessory cells. This seems particularly true of
AA CD34' cells, because it enables them to match the response of normal CD34' cells when sorted. A recent report
suggests that the administration of G-CSF with ALG and
steroids in the treatment of SAA patients improves their
response rate and survival; our data may support the hypothesis that this is due to the mobilizing effect of G-CSF on
early progenitors.28Our data also provide evidence that AA
CD34' cells may be dysfunctional in their response to IL3. There appears to be a correlation between disease severity
not only with the proportion of CD34+ cells in EM, but
also with the dysfunctionality of these cells, lending further
support for a stem cell functional defect in this disease.
Whether this dysfunctionality is due to an intrinsic abnormality or to a difference in the functional state of the AA CD34'
cells is unclear, although the corrective effect of G-CSF may
support the latter. The question of whether the observed
dysfunctionality applies to the whole stem cell compartment
or to a specific subpopulation will be addressed in assays
for LTCIC and single-cell cultures.
The authors thank Sandoz Pharma, Amgen UK, Immunex Corp,
and Cilag Ltd for their kind gifts of cytokines; Dr Nick Philpott for
taking BM samples from normal donors; and the hematology clinical
staff of St George's Hospital for providing samples from AA patients. Permission to seek informed consent for the use of BM samples was obtained from the St George's Hospital Research Ethics
Committee for both AA patients and normal subjects.
1. Juneja HS, Lee S, Gardner FH: Human long-term bone marrow
cultures in aplastic anaemia. Int J Cell Cloning 7:129, 1989
2. Marsh JCW, Chang J, Testa NG, Hows JM, Dexter TM: The
hematopoietic defect in aplastic anemia assessed by long-term marrow culture. Blood 76:1748, 1990
3. Gibson FM, Gordon-Smith EC: Long-term culture of aplastic
anaemia bone marrow. Br J Haematol 75:421, 1990
4. Bacigalupo A, Figari 0, Tong J, Piaggio G, Miceli S, Frassoni
F, Caciagli P, Badolati G, Marmont AM: Long-term marrow culture
in patients with aplastic anemia compared with marrow transplant
recipients and normal controls. Exp Hematol 20425, 1992
5. Hotta T, Kat0 T, Maeda H, Yamao H, Yamada H, Saito H:
Functional changes in marrow stromal cells in aplastic anemia. Acta
Hematol 74:65, 1985
6. Marsh JCW, Chang J, Testa NG, Hows JM, Dexter TM: In
vitro assessment of marrow "stem cell" and stromal cell function
in aplastic anaemia. Br J Haematol 78:258, 1991
7. Maciejewski JP, Anderson S, Katevas P, Young NS: Phenotypic and functional analysis of bone marrow progenitor cell compartment in bone marrow failure. Br J Haematol 87:227, 1994
8. Andrews RG, Singer JW,Bernstein ID: Precursors of colonyforming cells in humans can be distinguished from colony-forming
cells by expression of the CD33 and CD34 antigens and light scatter
properties. J Exp Med 169:1721, 1989
9. Scopes J, Bagnara M, Gordon-Smith EC, Ball SE, Gibson FM:
Haemopoietic progenitor cells are reduced in aplastic anaemia. Br
J Haematol 86:427, 1994
10. Dexter TM, Spooncer E: Growth and differentiation in the
hemopoietic system. Ann Rev Cell Biol 3:423, 1987
11. Antin JH, Smith BR, Holmes W, Rosenthal DS: Phase JAI
study of recombinant human granulocyte-macrophage colony-stimulating factor in aplastic anemia and myelodysplastic syndrome.
Blood 72:705, 1988
12. Bagby Gc,McCall R, Layman DL: Regulation of colonystimulating activity production. Interactions of fibroblasts, mononuclear phagocytes and lactofemn. J Clin Invest 71:340, 1983
13. Kojima S , Fukuda M, Miyajima Y, Matsuyama T, Honbe K:
Treatment of aplastic anemia in children with recombinant human
granulocyte colony-stimulating factor. Blood 77:937, 1991
14. Vadhan-Raj S, Buesher S, Broxmeyer HE, LeMaistre A,
From www.bloodjournal.org by guest on January 26, 2015. For personal use only.
Lepe-Zuniga JL, Ventura G, Jeha S, Honvitz LJ, Trujillo JM, Gillis
S, Hittelman WN, Gutterman JU: Stimulation of myelopoiesis in
patients with aplastic anemia by recombinant human granulocytemacrophage colony-stimulating factor. N Eng J Med 319:1628, 1988
15. Ganser A, Lindemann A, Seipelt G, Ottmann OG, Eder M,
Falk S, H e r "
F, Kaltwasser JP, Meusers P, Klausmann M,
Frisch J, Schulz G, Mertelsmann R, Hoelzer D: Effects of recombinant human interleukin-3 in aplastic anemia. Blood 76:1287, 1990
16. Nissen C, Tichelli A, Gratwohl A, Speck B, Milne A, GordonSmith EC, Schaedelin J: Failure of recombinant human granulocytemacrophage colony-stimulating factor therapy in aplastic anemia
patients with very severe neutropenia. Blood 72:2045, 1988
17. Henmann F, Lindemann A, Raghavachar A, Heimpel H, Mertelsmann R: In vivo recruitment of GM-CSF-response myelpoietic
progenitor cells by interleukin-3 in aplastic anemia. Leukemia 4:67 1,
18. Bagnara GP, Strippoli P, Bonsi L, Brizzi MF, Avanzi GC,
Timeus F, Ramenghi U, Piaggio G, Tong J, Podesta M, Paolucci G,
Pegoraro L, Gabutti V, Bacigalupo A: Effect of stem cell factor on
colony growth from acquired and constitutional (Fanconi) aplastic
anemia. Blood 80:382, 1992
19. Wodnar-Filipowicz A, Tichelli A, Zsebo KM, Speck B, Nissen C: Stem cell factor stimulates the in vitro growth of bone marrow
cells from aplastic anemia patients. Blood 79:3196, 1992
20. Amano Y, Koike K, Nakahata T: Stem cell factor enhances
the growth of primitive erythroid progenitors to a greater extent
than interleukin-3 in patients with aplastic anaemia. Br J Haematol
85:663, 1993
21. Gibson FM, Scopes J, Daly S, Ball SE, Gordon-Smith EC:
In vitro response of normal and aplastic anemia bone marrow to mast
cell growth factor and in combination with granulocyte-macrophage
colony-stimulating factor and interleukin-3. Exp Hematol 22:302,
22. International Agranulocytosis and Aplastic Anemia Study:
Incidence of aplastic anemia: The relevance of diagnostic criteria.
Blood 70:1718, 1987
23. Camitta BM, Rappeport JM, Parkman R, Nathan DG: Selection of patients for bone marrow transplantation in severe aplastic
anemia. Blood 45:355, 1975
24. Tavassoli M: Hematopoietic survival factors. Exp Hematol
21:1511, 1993
25. Ottmann OG, Abboud M, Welte K, Souza LM, Pelus LM:
Stimulation of human hematopoietic progenitor cell proliferation and
differentiation by recombinant human interleukin-3. Comparison and
interactions with recombinant human granulocyte-macrophage and
granulocyte colony-stimulating factors. Exp Hematol 17: 191, 1989
26. Lemoli RM, Gulati SC, Strife A, Lambek C, Perez A,
Clarkson BD: Proliferative response of human acute myeloid leukemia cells and normal marrow enriched progenitor cells to human
recombinant growth factors IL-3, GM-CSF and G-CSF alone and
in combination. Leukemia 5386, 1991
27. Eaves CJ, Cashman JD, Kay RJ, Dougherty GJ, Otsuka T,
Gaboury LA, Hogge DE, Lansdorp PM, Eaves AC, Humphries R K
Mechanisms that regulate the cell cycle status of very primitive
hematopoietic cells in long-term human marrow cultures. 11. Analysis of positive and negative regulators produced by stromal cells
within the adherent layer. Blood 78:1 IO, 1991
28. Bacigalupo A, Broccia G, Corda G, Arcese W, Carotenuto
M, Gallamini A, Locatelli F, Mori PG, Saracco P, Todeschini G,
Coser P, Iacopino P, van Lint MT, Gluckman E, for the European
Group for Blood and Marrow Transplantation (EBMT) Working
Party on SAA: Antilymphocyte globulin, cyclosporin and granulocyte colony-stimulating factor in patients with acquired severe aplastic anemia (SAA): A pilot study of the EBMT SAA working party.
Blood 85:1348, 1995
From www.bloodjournal.org by guest on January 26, 2015. For personal use only.
1996 87: 3179-3185
Aplastic anemia: evidence for dysfunctional bone marrow progenitor
cells and the corrective effect of granulocyte colony-stimulating
factor in vitro
J Scopes, S Daly, R Atkinson, SE Ball, EC Gordon-Smith and FM Gibson
Updated information and services can be found at:
Articles on similar topics can be found in the following Blood collections
Information about reproducing this article in parts or in its entirety may be found online at:
Information about ordering reprints may be found online at:
Information about subscriptions and ASH membership may be found online at:
Blood (print ISSN 0006-4971, online ISSN 1528-0020), is published weekly by the American
Society of Hematology, 2021 L St, NW, Suite 900, Washington DC 20036.
Copyright 2011 by The American Society of Hematology; all rights reserved.