Erythropoiesis Abnormalities Contribute to Early-Onset Anemia in Patients with Septic Shock

Erythropoiesis Abnormalities Contribute to Early-Onset
Anemia in Patients with Septic Shock
Yann-Erick Claessens, Michae
¨la Fontenay, Fre´de´ric Pene, Jean-Daniel Chiche, Martine Guesnu, Cyrla Hababou,
Nicole Casadevall, Jean-Franc¸ois Dhainaut, Jean-Paul Mira, and Alain Cariou
Medical Intensive Care Unit, INSERM U567, Department of Emergency Medicine, and Laboratory of Hematology, Ho
ˆpital Cochin;
Faculte´ de Me´decine, Universite´ Paris-Descartes; Laboratory of Hematology, Ho
ˆtel-Dieu, Paris; and Ortho Biotech France, Inc.,
Issy-les-Moulineaux, France
Rationale: The intimate mechanisms of early onset anemia observed
in this population, including ?uid-loading–related hemodilution,
in critically ill patients with septic shock remain unclear.
daily blood withdrawal for routine laboratory analysis (3), and
Objectives: We investigated erythropoiesis abnormalities in this set-
in?ammation with impaired iron metabolism (4). However, these
ting by studying morphologic, functional, and biochemical patterns
mechanisms of anemia during septic shock remain overall insuf-
of erythroid lineage.
?ciently explored as erythropoiesis has never been investigated
Methods: Erythroid lineage in the bone marrow from patients with
in this context.
septic shock who developed early-onset anemia was compared with
Erythropoiesis results in the production of red blood cells
that of healthy control subjects. Survival and proliferation capacities
from the bone marrow pluripotent stem cells. In adults, it is
were quantified in both groups. Biochemical and flow cytometry
tightly regulated by erythropoietin (Epo), a glycoprotein hor-
patterns of apoptosis were dissected by exploring antiapoptotic
mone mainly produced by interstitial cells of the kidney. Epo
(erythropoietin [Epo] receptor–dependent) and proapoptotic
(death receptor–dependent) pathways.
binds itself to a speci?c Epo receptor (EpoR) to activate a
Measurements and Main Results: Erythroid lineage was morphologi-
signaling cascade that supports in vivo survival and proliferation
cally similar in both groups. Apoptosis of glycophorin-A–positive
of mature erythroid progenitors of colony-forming unit ery-
erythroid precursors was increased in patients versus control sub-
throids and their terminal differentiation. These positive effects
jects as assessed by labeling with annexin V (26.1 ? 8.8 vs. 3.1 ?
of Epo are counterbalanced by ligands for the members of the
2.9%, p ? 0.05) or 3–3?-dihexyloxacarbocyanine iodide (55.9 ? 10.5 vs.
tumor necrosis factor (TNF) receptor family, especially Fas li-
19.1 ? 5.4%, p ? 0.05), respectively. This was associated with
gand (FasL), TNF apoptosis-inducing ligand (TRAIL), and
significant overexpression of Fas on erythroid precursors and higher
TNF-? (5–7). An imbalance between survival and death signals
tumor necrosis factor-? plasma levels in patients with septic shock
resulting in impaired red cell production has been previously
vs. control subjects. Moreover, growth capacities of late erythroid
observed in myelodysplasia and rheumatoid arthritis (8, 9). Apo-
progenitors of burst-forming unit erythroids (BFU-Es) at Day 10
ptosis is also of dramatic importance during septic shock (10).
were impaired in the presence of serum from patients with septic
More precisely, the involvement of apoptosis by the Fas pathway
shock as compared with the effect of serum from control subjects
was previously suggested in various lineages during sepsis (10,
(27 ? 12 vs. 109 ? 27 per 105 seeded cells, respectively; p ? 0.001).
11) but had never been studied in the erythroid compartment.
Saturating concentrations of recombinant human Epo (rHuEpo)
To investigate the erythropoiesis abnormalities that could con-
restored growth capacity of patients’ BFU-Es (72 ? 14 per 105 seeded
tribute to the early-onset anemia during septic shock, morpho-
cells) in autologous conditions of serum.
logic and functional studies were conducted in erythroid cells
Conclusions: Early-onset anemia that may be observed in patients
derived from the bone marrow of critically ill patients with septic
with septic shock is associated with defective erythropoiesis related
shock. Some of the results of this study were reported in the
to an excess of apoptosis that can be counterbalanced in vitro by
rHuEpo.
form of an abstract at the 100th meeting of the American
Thoracic Society in 2004 (12).
Keywords: anemia; apoptosis; erythropoiesis; erythropoietin; septic
shock
METHODS
Anemia is common in critically ill patients who often require
Patients
packed red blood cell (PRBC) transfusion (1). Patients with
Ten patients were enrolled in a prospective observational study and
severe sepsis or septic shock usually have a hemoglobin level
were compared with healthy control subjects. Eligible patients were
lower than other patients admitted to the intensive care unit
adults (age ? 18 yr) with septic shock related to a community-acquired
(ICU) and are more likely to require transfusion of PRBCs (2).
infection (13) who developed early- onset anemia de?ned by a hemoglo-
A number of mechanisms have been proposed to explain anemia
bin level below 100 g/L within 48 h after ICU admission. Patients could
not be enrolled if any of the following criteria were present: history of
bone marrow disease, chemotherapy or immunosuppressive treatments,
chronic renal failure (de?ned as need for chronic renal replacement
therapy or known serum creatinine level above 180 ?mol/L before the
(Received in original form April 11, 2005; accepted in final form March 30, 2006 )
septic event), infection by HIV or human T-cell lymphotropic virus
Supported by OrthoBiotech France, Inc.
type 1 (HTLV), hemolysis, underlying bleeding disorders, or recent
Correspondence and requests for reprints should be addressed to Alain Cariou,
surgical procedure (during the last 30 d). Data collected included demo-
M.D., Medical Intensive Care Unit, Cochin Hospital, APHP Universite´ Paris-Descartes,
graphics, clinical characteristics including Simpli?ed Acute Physiologic
27 rue du Faubourg Saint-Jacques, F-75679 Paris, Cedex 14, France. E-mail: alain.
Score (SAPS) 2 and logistic organ dysfunction (LOD) severity scores
cariou@cch.ap-hop-paris.fr
(14, 15), biological features, treatments, and outcome. Healthy control
This article has an online supplement, which is accessible from this issue’s table
subjects were patients referred for exploration of thrombophilia, which
of contents at www.atsjournal.org
includes bone marrow aspiration as a routine procedure. These control
subjects were found to have normal hemogram and normal clonogenic
Am J Respir Crit Care Med
Vol 174. pp 51–57, 2006
Originally Published in Press as DOI: 10.1164/rccm.200504-561OC on March 30, 2006
culture study in methyl-cellulose medium and bone marrow pathologic
Internet address: www.atsjournals.org
analysis. The use of these control samples was granted with the agreement

---

52
AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE
VOL 174
2006
of the French Ministry of Health. The study was approved by our institu-
sulfate–polyacrylamide gel electrophoresis and immunoblots were per-
tion’s ethics committee and all subjects or their surrogates gave written,
formed using phospho-ERK 1/2 and phospho-STAT5 from Promega
informed consent.
Life Science (Madison, WI), anti-S473 phospho-AKT from New
England Biolabs (Beverly, MA), and anti-threonine 32 phospho-
Plasma and Serum Dosages of Epo, Interleukin-6, TRAIL,
FKHRL1 from Upstate Biotechnologies (Lake Placid, NY).
FasL, TNF-?, and IFN-?
Immunophenotyping
Plasma and serum were obtained from peripheral blood at time of
medullar aspiration. Cytokine levels were measured at inclusion using
GPA-positive BMMNCs were incubated (20 min, 4?C) in the presence
ELISA with speci?c reagents according to the manufacturer’s instruc-
of anti-Fas ?uorescein isothiocyanate, anti–TNF-? receptor type 1
tions (Quantikine; R&D System, Minneapolis, MN).
(TNFR1) phycoerythrin (PE)-conjugated antibodies (Becton-Dickinson,
Palo Alto, CA). Cells were also incubated with unlabeled primary
Bone Marrow Samples and Cytologic Analysis
antibodies to TRAIL or TRAIL receptors (TRAIL-R1, TRAIL-R2,
TRAIL-R3, TRAIL-R4) and with secondary PE-conjugated antibodies
The mean delay between ICU admission and medullar aspiration was
before analysis. A Coulter Epics XL device (Beckman Coulter, Inc.,
30 ? 12 h. Bone marrow was collected by sternal aspiration in a dried
Miami, FL) was used to quantify ?uorescence. Fas, TNFR1, TRAIL,
sodium heparin–containing bottle and diluted in phosphate-buffered
TRAIL-R1, TRAIL-R2, TRAIL-R3, and TRAIL-R4 expression was
saline (PBS) containing 0.8% bovine serum albumin. Bone marrow
reported as the ratio of mean ?uorescence intensity to the irrelevant
mononuclear cells (BMMNCs) were isolated using Ficoll-Hypaque gra-
control antibody.
dient. Cellular lineages were quanti?ed using light microscopy (?100)
after May-Gru¨nwald-Giemsa staining.
Analysis of Apoptosis
Glycophorin A–positive Cell Separation
Apoptosis was quanti?ed on GPA-positive cells by staining both with
annexin V ?uorescein isothiocyanate (Boehringer Mannhein, Mannh-
Cells that expressed membrane glycophorin A (GPA), a speci?c protein
ein, Germany) and propidium iodide (PI; Sigma-Aldrich, St. Louis,
of mature erythroid elements, were isolated on MIDI-MACS immuno-
MO). Variation in the mitochondrial transmembrane potential (??
magnetic columns among the BMMNCs according to the manufacturer’s
m)
was measured by labeling cells with both 3-3?-dihexyloxacarbocyanine
instructions (Miltenyi Biotech, Bergisch Badgach, Germany [8]). GPA
iodide (DiOC
positivity quanti?ed by ?ow cytometry was above 90% after isolation.
6[3]; Sigma-Aldrich) and PI (16). Results were expressed
as percentages of annexin V–positive/PI-negative cells and percentages
Morphology was assessed by light microscopy (?100). This separation
of DiOC
provided erythroid precursors that were tested for ?uorescence-
6(3) low/PI-negative cells.
activated cell sorter analysis and biochemical assays.
Statistical Analysis
Colony Formation Assay
Values are expressed in mean ? SD. In the absence of a normal distribu-
tion, continuous variables of independent samples are compared by
BMMNCs were cultured to quantify clonogenic erythroid progenitors
nonparametric Mann-Whitney’s test. Paired data were analyzed by
that form late burst-forming unit erythroids (BFU-Es) after 10 d of
Wilcoxon rank sum test. p Values of less than 0.05 were considered
culture in semisolid medium. Cells were plated as previously described
statistically signi?cant.
in 0.8% methyl-cellulose medium containing fetal calf serum 20% and
a mixture of cytokines including 2 U/ml recombinant human (rHu)
Epo provided by OrthoBiotech France, Inc. (Issy-les-Moulineaux,
RESULTS
France) (8). The effect of control subjects’ and patients’ sera on BFU-E
formation was also tested. BMMNCs from patients were plated in usual
Patients
semisolid culture medium with 0.8% methyl-cellulose containing 0.5
Between November 2002 and March 2003, 46 patients were
U/ml rHuEpo, 50 ng/ml stem cell factor (SCF), and 40 ng/ml rHu
admitted to the medical ICU for septic shock, of whom 10 (mean
IGF-1 added with 20% of control subjects’ or patients’ sera. In some
age, 51 ? 16 yr) met the inclusion criteria. The reasons for
experiments, addition of a higher concentration of rHuEpo (2 U/ml)
noninclusion were as follows: hemolysis (n ? 1); HIV infection
tested the “rescue” properties of Epo. Late BFU-Es were numbered
(n ? 2); chemotherapy, immunosuppressive treatment, or hemo-
under an inverted microscope at Day 10 by two independent examina-
tors blinded for the sample provenance.
pathy (n ? 14); chronic renal failure (n ? 3); and inability to
obtain informed consent (n ? 16).
Epo-induced Stimulation
Patients’ characteristics are reported in Table 1. Mean SAPS
For the analysis of EpoR signaling, BMMNCs were resuspended and
2 and LOD scores were, respectively, 68.7 and 7.8, highlighting
serum starved for 4 h before stimulation with 10 U/ml rHuEpo for
a severely ill population with a predictable mortality rate higher
10 min before lysis. Protein extracts were separated by sodium dodecyl
than 80%. Mean duration of stay in the ICU was 15 d (ranging
TABLE 1. DEMOGRAPHICS, DIAGNOSIS, MICROBIOLOGICAL IDENTIFICATION, PROGNOSTIC
SCORING SYSTEM, LENGTH OF STAY IN INTENSIVE CARE UNIT, AND OUTCOME FOR THE
10 PATIENTS IN THE STUDY
Patient No.
Sex/Age (yr)
Diagnosis
Bacteria
SAPS 2/LOD
ICU LOS (d )
Outcome
1
F/43
Pneumonia
NA
59/12
2
Died
2
F/25
Pneumonia
Haemophilus influenzae
39/5
21
Died
3
F/61
Endocarditis
Staphylococcus aureus
71/10
2
Died
4
M/56
Pneumonia
Klesiella pneumoniae
63/11
26
Died
5
M/59
Septicemia
Escherichia coli
74/10
3
Died
6
F/62
Pneumonia
Streptococcus pneumoniae
44/6
16
Survived
7
M/42
Peritonitis
Streptococcus pneumoniae
51/4
14
Survived
8
M/47
Pneumonia
Haemophilus influenzae
93/7
56
Survived
9
F/29
Pneumonia
NA
64/7
14
Survived
10
M/85
Pneumonia
NA
129/6
2
Died
Definition of abbreviations: F ? female; ICU ? intensive care unit; LOD ? logistic organ dysfunction system; LOS ? length of
stay; M ? male; NA ? not available; SAPS 2 ? simplified acute physiology score 2.

---

Claessens, Fontenay, Pene, et al.: Erythropoiesis Abnormalities during Sepsis
53
TABLE 2. LABORATORY DATA ON ADMISSION AND RED CELL PACKED TRANSFUSIONS DURING
INTENSIVE CARE UNIT HOSPITALIZATION
Patient No.
Hb Level (g/L)
RCV (fl)
Reticulocytes (/?l)
Epo (mU/ml)
Creatinine (?mol/L)
PRBCs (U )
1
83
83
42,700
19
64
0
2
98
85
48,400
35
75
2
3
97
90
56,900
NA
129
0
4
82
91
19,800
46
397
7
5
89
91
126,000
54
224
0
6
86
89
49,300
42
70
4
7
87
91
NA
15
43
2
8
89
89
51,300
25
222
6
9
88
87
78,700
16
94
2
10
94
88
NA
52
104
0
Definition of abbreviations: Epo ? erythropoientin; Hb ? hemoglobin; NA ? not available; PRBCs ? packed red blood cells;
RCV ? red cell volume.
from 2 to 56 d). Primary infection was pneumonia (n ? 7), peritoni-
shock. These results suggest that early-onset anemia observed
tis (n ? 1), endocarditis (n ? 1), and septicemia (n ? 1). Blood-
during septic shock cannot be explained by a decreased number
stream cultures were positive in four patients and pathogens
of erythroid precursors, abnormal differentiation, or erythropha-
were identi?ed in seven. All patients enrolled in the study devel-
gocytosis secondary to macrophagic activation.
oped multiorgan failure related to sepsis and required supportive
care with vasoactive drugs and mechanical ventilation. An ele-
Apoptosis Is Increased in Bone Marrow Erythroid Precursors
vated serum creatinine concentration was present at inclusion
during Septic Shock
in 4 of 10 patients and 6 patients required renal replacement
Erythroid precursors, which are distinguished from other
therapy after bone marrow and blood sampling. The in-ICU
BMMNCs by their GPA staining and are called GPA? cells,
mortality rate was 60%. Biological data at inclusion plus hemo-
were isolated and studied for apoptosis by ?ow cytometry. The
globin level and transfusion requirement during ICU stay are
percentage of annexin V–positive/PI-negative cells was signi?-
reported in Table 2. All patients exhibited nonregenerative ane-
cantly higher in erythroid precursors derived from patients with
mia and all 5-d survivors (n ? 6) required PRBC transfusion
septic shock as compared with control subjects (26.1 ? 8.8 and
because of a hemoglobin level below 70 g/L according to stan-
3.1 ? 2.9%, respectively; p ? 0.05; Figure 1A). The variation
dard recommendations (17).
of mitochondrial transmembrane potential (??m) was also mea-
sured as an early step of apoptosis. We found that the percentage
Erythroid Lineage Is Not Decreased in Bone Marrow of
of DiOC6(3) low/PI-negative cells was signi?cantly higher in pa-
Patients with Septic Shock and Anemia
tients with septic shock than in control subjects (55.9 ? 10.5 vs.
Morphologic study of the bone marrow aspiration revealed a
19.1 ? 5.4%, respectively; p ? 0.05; Figure 1B). This con?rms
normal distribution of the hematologic lineages in patients with
that there was an increase of cells with features of apoptosis,
septic shock (Table 3). No differences could be detected in the
either collapse of ??m or plasma membrane modi?cations.
erythroid, granulomonocytic, and lymphoid lineages between
Epo Signaling Is Preserved in BMMNCs from Patients with
patients with septic shock and control subjects, and bone marrow
Septic Shock
smears were equally rich in the two groups. Cytologic count of
erythroid compartment in patients with septic shock revealed 7,
Because in vivo survival of the mature erythroid progenitors
22, and 71% of basophilic, polychromatophilic, and acidophilic
mainly depends on the presence of Epo, we quanti?ed the Epo
erythroblasts, respectively. Megakaryocytes were present in all
plasma level and the functionality of its receptor in patients with
bone marrow aspirations. In addition, no features of hemophago-
septic shock and control subjects. Plasma Epo concentration in
cytosis were detected in bone marrow from patients with septic
patients with septic shock did not differ from that of control
subjects (34 ? 15 vs. 38 ? 8 mU/ml, p ? 0.965) and remained
within normal range (Table 2). EpoR appeared to be functional
as demonstrated by analysis of EpoR signaling pathway. Hence
TABLE 3. DISTRIBUTION OF THE HEMATOLOGIC LINEAGES
signals of phospho-ERK 1/2, phospho-STAT5, phospho-Akt and
AFTER CYTOLOGIC EXAMINATION OF THE BONE MARROW
activation of its downstream substrate FKHRL1 were increased
ASPIRATION FROM CONTROL SUBJECTS AND PATIENTS
in BMMNCs from control subjects and patients after Epo stimu-
WITH SEPTIC SHOCK
lation (Figure 2A). This suggests that antiapoptotic signaling is
Patients with
normal in Epo-responsive cells. Then we quanti?ed late ery-
Control Subjects
Septic Shock
throid progenitors giving BFU-Es at Day 10 of semisolid culture
Lineage
(Mean % ? SD)
(Mean % ? SD)
p
from the BMMNCs both in patients and control subjects. The
number of BFU-Es was similar in control subjects and patients
Erythroid lineage
16 ? 3
14 ? 3
0.88
with septic shock (151 ? 65 vs. 174 ? 57 per 105 seeded cells,
Proerythroblast
0
0
0.99
Basophilic erythroblast
2 ? 1
1 ? 1
0.76
respectively; p ? 0.58; Figure 2B). This result shows that the
Polychromatophilic erythroblast
3 ? 1
3 ? 1
0.91
pool of late erythroid progenitors is not quantitatively impaired
Acidophilic erythroblast
12 ? 2
10 ? 2
0.66
in patients’ bone marrows.
Granulocytic lineage
70 ? 5
68 ? 5
0.72
Monocytes
2 ? 2
2 ? 2
0.96
Fas Is Overexpressed at the Membrane of Erythroid
Lymphoid lineage
10 ? 2
9 ? 2
0.76
Precursors during Septic Shock
Results are presented as mean percentage of total bone marrow cell count ?
We tested GPA? cells for the expression of pro- and antiapo-
SD for each cell population.
ptotic proteins to investigate whether a speci?c death receptor

---

54
AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE
VOL 174
2006
Figure 1. Flow cytometry analysis for the quantification of
apoptotic cells. Apoptotic cells were quantified by flow
cytometry using membrane staining with annexin
V–positive/propidium iodide (PI)–negative (A ) and 3–3?-
dihexyloxacarbocyanine iodide (DiOC6[3]) low/PI–negative
(B ). Results are presented as mean ? SD percentage of
apoptotic cells for each staining. A p value ? 0.01 is consid-
ered statistically significant and marked with an asterisk.
C illustrates a representative experiment of the fluores-
cence-activated cell sorter analysis: annexin V (top) and
??m (bottom) in a control (left) and a patient (right).
pathway could be involved in the erythroid compartment
was signi?cantly lower in the presence of serum from patients
apoptosis. Using ?ow cytometry, we quanti?ed the expression
with septic shock (27 ? 12 per 105 seeded cells) as compared
of Fas, the main death domain receptor involved in the negative
with control serum (109 ? 27 per 105 seeded cells, p ? 0.001;
regulation of normal erythropoiesis. The expression of Fas was
Figure 3) suggesting a strong growth inhibitory effect of septic
signi?cantly higher in patients with septic shock than in control
serum. To test the ability of a saturating concentration of Epo
subjects (mean ?uorescence intensity: 3.06 ? 1.20 vs. 1.06 ?
to restore the development of erythroid progenitors in the pres-
0.18, respectively; p ? 0.01; Table 4). In contrast, the membrane
ence of septic serum, 2 U/ml rHuEpo was added to clonogenic
expression of the type 1 TNF-? receptor and of TRAIL and its
cultures. At Day 10, both in the presence of septic and control
speci?c receptors (the death domain–containing TRAIL-R1 and
serum, the number of BFU-Es was signi?cantly increased in the
TRAIL-R2 and the decoy receptors TRAIL-R3 and TRAIL-
presence of an elevated concentration of Epo but remained
R4) were similar in both groups (Table 4).
higher when erythroid progenitors were cultured with control
serum as compared with septic serum (182 ? 15 vs. 72 ? 14,
Serum from Patients with Septic Shock Impairs the
respectively; p ? 0.001), suggesting the presence of an inhibitory
Development of Erythroid Progenitors
factor in the latter group (Figure 3).
To assess the role of serum from patients with septic shock on
To investigate whether septic serum components contributed
the development of late erythroid progenitors, BMMNCs from
to impaired erythroid progenitor development, we measured
several patients with septic shock were plated in semisolid cul-
plasma concentrations of IL-6 as a marker of in?ammation
tures containing 0.5 U/ml rHuEpo and 20% serum from either
and cytokines known to regulate apoptosis including TNF-?,
patients or control subjects. At Day 10, the number of BFU-Es
FasL, TRAIL, and IFN-?. Plasma levels of IL-6 signi?cantly

---

Claessens, Fontenay, Pene, et al.: Erythropoiesis Abnormalities during Sepsis
55
This research presents some limits. The number of studied
patients is low but may be suf?cient for a preliminary study with
well-de?ned characteristics in a speci?c subset of patients with
septic shock as revealed by the homogeneity of bone marrow
studies. In contrast, the nature of the control population (healthy
ambulatory patients) may be criticized but ethical considerations
precluded medullar aspirations in the best control population,
which would have been patients with septic shock who had not
developed anemia. Finally, we did not assess the quanti?cation
of colony-forming unit erythroids that correspond to the most
mature erythroid progenitors that differentiate into erythroid
precursors in vivo.
Anemia is a common feature in critically ill patients with
septic shock. Among the numerous factors that may contribute
to its development, some of them cannot be implicated in our
Figure 2. (A ) EpoR signaling in bone marrow mononuclear cells
population due to the design of the study. Only patients with
(BMMNCs) from patients with septic shock and control subjects. Cells
early-onset anemia were studied, thus excluding frequent blood
were extracted from total bone marrow and were incubated with Epo.
sampling and renal failure as potential causes. Even if some of
Immunoblotting was used to reveal the activation of pathways down-
our patients (6 of 10) required renal replacement therapy, these
stream from EpoR. (B ) Quantification of burst forming unit erythroids
patients had no previous renal disorder and the renal failure
(BFU-Es) in semisolid clonogenic culture. A total of 2.5 ? 105 BMMNCs
was an acute phenomenon related to the septic shock. Further-
were seeded in a methylcellulose medium containing 10% fetal calf
serum with Epo, stem cell factor (SCF), interleukin (IL)-6, IL-3, and
more, as patients with blood loss (surgical procedure, active
granulocyte-macrophage colony–stimulating factor. BFU-E–derived col-
bleeding) or hemolysis were excluded, erythropoiesis abnormali-
onies were counted at Day 10. Results are presented as mean ? SD of
ties can be suspected as a factor that contributes to early-onset
the BFU-E colonies numbered for 105 seeded cells.
anemia in patients with septic shock. Erythropoiesis is under
control of antiapoptotic effects of Epo but also depends on
proapoptotic signals from Fas and TRAIL pathways (6, 7, 18).
Our results suggest that early-onset anemia in patients with
differed between patients with septic shock and control subjects
septic shock is associated with a defective erythropoiesis related
(1,332 ? 821 vs. 13 ? 11 ng/ml, respectively; p ? 0.001). Also,
to an imbalance between anti- and proapoptotic signals. The
TNF-? levels were higher in plasma of patients with septic shock
patients exhibited inappropriately low plasma Epo concentra-
(31 ? 17 pg/ml) as compared with control subjects (12 ? 9 pg/ml,
tions for the degree of anemia as previously reported (19). This
p ? 0.018). Concentrations of the remaining cytokines involved in
?nding is in accordance with previously published data on low
the apoptotic response did not signi?cantly differ between the two
Epo levels in serum of critically ill patients with sepsis (20). This
groups (Table 5).
inadequacy could partly explain anemia because Epo signaling
is pivotal for survival of late erythroid progenitors in vivo and
in vitro (21, 22), through the phosphatidylinositol 3-kinase path-
DISCUSSION
way (23). Another mechanism involved in early-onset anemia
Erythropoiesis abnormalities have never been studied in the
could be a blunted EpoR signaling. However, Western blot anal-
setting of early-onset anemia in patients with septic shock. Here
ysis of BMMNCs from patients with septic shock stated that this
we show that (1 ) erythroid lineage is not quantitatively de-
pathway was functional because Akt and FKHRL1, downstream
creased in the bone marrow of patients with septic shock with
substrates of phosphatidylinositol 3-kinase, were phosphorylated
early anemia, (2 ) anemia is associated with apoptosis of bone
in response to Epo stimulation, suggesting a functional EpoR.
marrow erythroid precursors despite the presence of a functional
Apoptosis in sepsis can account for organ failure (11) and
EpoR on BMMNCs, (3 ) membrane expression of the death
involves hematopoietic lineages including polymorphonuclear
domain receptor Fas is up-regulated, and (4 ) the number of late
cells (24), lymphocytes, dendritic cells (25, 26), and endothelial
erythroid progenitors is decreased in the presence of serum from
cells (27). Our results suggest that apoptosis is increased in the
patients with septic shock.
erythroid lineage of patients with septic shock compared with
control subjects. Indeed, mean percentages of DiOC6(3) low cells/
PI-negative cells and of annexin V–positive/PI-negative cells
were higher in patients with septic shock than in the control
TABLE 4. GLYCOPHORIN A–POSITIVE ERYTHROID
group. Because apoptosis was rare in the control group, we could
PRECURSORS MEMBRANE EXPRESSION OF Fas, TUMOR
exclude the possibility that GPA? cells were stressed during the
NECROSIS FACTOR (TNF) RECEPTOR 1, TNF APOPTOSIS-
isolation step. The collapse of ??m revealed by low DiOC6(3)
INDUCING LIGAND (TRAIL), AND TRAIL RECEPTORS
labeling has been demonstrated to characterize cells ongoing
Control Subjects
Patients with Septic
early steps of apoptosis preceding DNA fragmentation and to
(Mean ? SD)
Shock (Mean ? SD)
p
be reversible when Bcl-2 is overexpressed (16). By contrast,
annexin V–labeled cells correspond to apoptotic cells with DNA
Fas
1.06 ? 0.18
3.06 ? 1.20
? 0.01
fragmentation and phosphatidylserine exposure, which is a
TNFR1
0.97 ? 0.18
0.93 ? 0.22
0.74
TRAIL
0.93 ? 0.15
1.49 ? 0.70
0.11
phagocytic signal. Because the mean percentage of cells with a
TRAIL-R1
0.82 ? 0.16
1.59 ? 0.87
0.08
drop of ??m was higher than the mean percentage of annexin
TRAIL-R2
1.13 ? 0.18
1.58 ? 1.20
0.49
V–positive cells both in the septic shock and control groups, it
TRAIL-R3
1.11 ? 0.11
1.73 ? 1.14
0.27
is very likely that only some of the GPA? erythroid precursors
TRAIL-R4
1.28 ? 0.53
1.58 ? 0.78
0.48
enter the full apoptotic state. Apoptosis was associated with
Results are presented as mean ? SD of ratio of mean fluorescent intensity to
increased membrane expression of Fas and high TNF-? plasma
isotype antibody for each membrane protein.
levels. Physiologically, Fas allows the down-regulation of early

---

56
AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE
VOL 174
2006
Figure 3. Results of semisolid clonogenic cul-
tures: BFU-E formation using serum from control
subjects and from patients with septic shock, in
the presence of 0.5 or 2 U/ml rHuEpo. Compari-
sons between control and septic serums were
performed using a Mann-Whitney test. Compar-
isons between the two Epo conditions (0.5 and
2 U/ml) in the control and septic serums were
performed using a Wilcoxon rank sum test.
erythroid progenitors by late erythroid progenitors that express
We hypothesize that the relatively low level of plasma Epo in
FasL (7). In various conditions, including hematologic diseases,
patients with septic shock may represent an insuf?cient antiapo-
up-regulation of Fas-related apoptosis may explain anemia (8,
ptotic signal delivered to the erythroid lineage.
28). Anemia related to in?ammatory disorders like rheumatoid
Recent clinical data also support our observations. Critically
arthritis has also been explained by an increased apoptosis of
ill patients treated with rHuEpo required less transfusion than
the erythroid lineage linked to TNF-? (9). Indeed, TNF-? is
a control group (33) with dose regimen of rHuEpo allowing
overexpressed in the acute phase of severe sepsis (reviewed
plasma concentration comparable to those used in culture me-
in Reference 29). This cytokine is supposed to down-regulate
dium (34). Unfortunately, the subgroup analysis did not speci?-
erythropoiesis (30, 31) as it positively regulates the expression
cally evaluate the subset of patients with septic shock. Our results
of Fas by a transcriptional mechanism. Thus, overexpression of
support the need for further clinical research on the bene?cial
Fas on erythroid precursors might be related to the high level
effects of high doses of rHuEpo in patients with septic shock.
of TNF-?. Soluble FasL plasma concentration, which is usually
elevated in sepsis (32), was not increased in our septic popula-
Conflict of Interest Statement : Y.-E.C. does not have a financial relationship with
tion. Moreover FasL was not overexpressed in GPA? erythroid
a commercial entity that has an interest in the subject of this manuscript. M.F.
does not have a financial relationship with a commercial entity that has an interest
precursors lysates of patients with septic shock (data not shown).
in the subject of this manuscript. F.P. does not have a financial relationship with
The accurate mechanisms of erythroid progenitor apoptosis re-
a commercial entity that has an interest in the subject of this manuscript. J.-D.C.
main unclear in our study. The study of erythroid precursors
does not have a financial relationship with a commercial entity that has an interest
isolated from TNF-? null mice and lpr mice (which encodes a
in the subject of this manuscript. M.G. does not have a financial relationship with
a commercial entity that has an interest in the subject of this manuscript. C.H. is
nonfunctional Fas) subjected to septic shock may help to clarify
an employee of Janssen Cilag division of Ortho Biotech. N.C. received €2,000 in
the role of these signaling pathways in the genesis of septic anemia.
2003 and €2,000 in 2004 for speaking at conferences and participating in advisory
Even if proapoptotic factors still need to be precisely identi-
boards for Ortho Biotech. J.-F.D. does not have a financial relationship with a
commercial entity that has an interest in the subject of this manuscript. J.-P.M.
?ed, we have clearly shown the negative effect of septic serum
does not have a financial relationship with a commercial entity that has an interest
on late erythroid progenitor proliferation. Interestingly, addition
in the subject of this manuscript. A.C. has participated as a speaker in scientific
of high-concentration rHuEpo partially allows the in vitro devel-
meetings organized and financed by Orthobiotech France, Inc., and received
opment of these erythroid progenitors even in the presence of
honoraria from their firm. Orthobiotech France, Inc., had no involvement in the
analysis and interpretation of the data, in the writing of the report, or in the
septic serum. This result is in accordance with a previous study
decision to submit the paper for publication.
reporting that Fas cross-linking–induced apoptosis in erythro-
blasts was antagonized by Epo in a dose-dependent manner (7).
Contributors: The study was designed by Y.E. Claessens and A. Cariou. The experi-
ments were realized by Y. E. Claessens, M. Guesnu, and N. Casadevall. C. Hababou
contributed to the collection of the data. Y. E. Claessens, A. Cariou, F. Pene, M.
Fontenay, and J. P. Mira produced the manuscript. J. D. Chiche and J. F. Dhainaut
participated in its critical revision. Y. E. Claessens had full access to all the data
TABLE 5. PLASMA CONCENTRATION OF THE DIFFERENT
in the study and had final responsibility for the decision to submit for publication.
CYTOKINES TESTED IN SERUM AND PLASMA FROM
PATIENTS WITH SEPTIC SHOCK AND CONTROL SUBJECTS
References
Control Subjects
Patients with Septic
1. Vincent JL, Baron JF, Reinhart K, Gattinoni L, Thijs L, Webb A, Meier-
Cytokine
(Mean ? SD)
Shock (Mean ? SD)
p
Hellmann A, Nollet G, Peres-Bota D. Anemia and blood transfusion in
critically ill patients. ABC (Anemia and Blood Transfusion in Critical
Epo
(mU/ml)
38 ? 8
34 ? 15
0.965
Care) Investigators. JAMA 2002;288:1499–1507.
IL-6
(pg/ml)
13 ? 11
1332 ? 821
? 0.001
TNF-?
(ng/ml)
12 ? 9
31 ? 17
0.018
2. Nguyen BV, Bota DP, Melot C, Vincent JL. Time course of hemoglobin
IFN-?
(pg/ml)
14 ? 11
21 ? 20
0.365
concentrations in nonbleeding intensive care unit patients. Crit Care
FasL
(pg/ml)
59 ? 9
53 ? 46
0.792
Med 2003;31:406–410.
TRAIL
(pg/ml)
56 ? 39
38 ? 58
0.504
3. Burnum JF. Medical vampires. N Engl J Med 1986;314:1250–1251.
4. von Ahsen N, Muller C, Serke S, Frei U, Eckardt KU. Important role
Definition of abbreviations: IL ? interleukin; TNF ? tumor necrosis factor;
of nondiagnostic blood loss and blunted erythropoietic response in
TRAIL ? TNF apoptosis-inducing ligand.
the anemia of medical intensive care patients. Crit Care Med 1999;27:
Results are presented as mean ? SD for each cytokine.
2630–2639.

---

Claessens, Fontenay, Pene, et al.: Erythropoiesis Abnormalities during Sepsis
57
5. Rebel VI, Hartnett S, Hill GR, Lazo-Kallanian SB, Ferrara JL, Sieff CA.
van de Wiel A. Response of erythropoiesis and iron metabolism to
Essential role for the p55 tumor necrosis factor receptor in regulating
recombinant human erythropoietin in intensive care unit patients. Crit
hematopoiesis at a stem cell level. J Exp Med 1999;190:1493–1504.
Care Med 2000;28:2773–2778.
6. De Maria R, Testa U, Luchetti L, Zeuner A, Stassi G, Pelosi E, Riccioni
20. Rogiers P, Zhang H, Leeman M, Nagler J, Neels H, Melot C, Vincent
R, Felli N, Samoggia P, Peschle C. Apoptotic role of Fas/Fas ligand
JL. Erythropoietin response is blunted in critically ill patients. Intensive
system in the regulation of erythropoiesis. Blood 1999;93:796–803.
Care Med 1997;23:159–162.
7. Zamai L, Secchiero P, Pierpaoli S, Bassini A, Papa S, Alnemri ES,
21. Zon LI. Developmental biology of hematopoiesis. Blood 1995;86:2876–
Guidotti L, Vitale M, Zauli G. TNF-related apoptosis-inducing ligand
2891.
(TRAIL) as a negative regulator of normal human erythropoiesis.
22. Wu H, Klingmuller U, Acurio A, Hsiao JG, Lodish HF. Functional
Blood 2000;95:3716–3724.
interaction of erythropoietin and stem cell factor receptors is essential
8. Claessens YE, Bouscary D, Dupont JM, Picard F, Melle J, Gisselbrecht
for erythroid colony formation. Proc Natl Acad Sci USA 1997;94:
S, Lacombe C, Dreyfus F, Mayeux P, Fontenay-Roupie M. In vitro
1806–1810.
proliferation and differentiation of erythroid progenitors from patients
23. Bouscary D, Pene F, Claessens YE, Muller O, Chretien S, Fontenay-
with myelodysplastic syndromes: evidence for Fas-dependent apopto-
Roupie M, Gisselbrecht S, Mayeux P, Lacombe C. Critical role for PI
sis. Blood 2002;99:1594–1601.
3-kinase in the control of erythropoietin-induced erythroid progenitor
9. Papadaki HA, Kritikos HD, Valatas V, Boumpas DT, Eliopoulos GD.
proliferation. Blood 2003;101:3436–3443.
Anemia of chronic disease in rheumatoid arthritis is associated with
24. Adrie C, Bachelet M, Vayssier-Taussat M, Russo-Marie F, Bouchaert I,
increased apoptosis of bone marrow erythroid cells: improvement fol-
Adib-Conquy M, Cavaillon JM, Pinsky MR, Dhainaut JF, Polla BS.
lowing anti-tumor necrosis factor-alpha antibody therapy. Blood 2002;
Mitochondrial membrane potential and apoptosis peripheral blood
100:474–482.
monocytes in severe human sepsis. Am J Respir Crit Care Med 2001;
10. Hotchkiss RS, Karl IE. The pathophysiology and treatment of sepsis.
164:389–395.
N Engl J Med 2003;348:138–150.
25. Hotchkiss RS, Tinsley KW, Swanson PE, Schmieg RE Jr, Hui JJ, Chang
11. Papathanassoglou ED, Moynihan JA, Vermillion DL, McDermott MP,
KC, Osborne DF, Freeman BD, Cobb JP, Buchman TG, et al. Sepsis-
Ac