Вы находитесь на странице: 1из 9

Inflamm. Res.

(2010) 59:339347
DOI 10.1007/s00011-009-0100-0 Inflammation Research
ORIGINAL RESEARCH PAPER

Procalcitonin as an early marker of bacterial infection


in neutropenic febrile children with acute lymphoblastic
leukemia
Maria Hatzistilianou Aleka Rekliti
Fanni Athanassiadou Dorothea Catriu

Received: 31 August 2008 / Revised: 15 July 2009 / Accepted: 22 September 2009 / Published online: 6 October 2009
Birkhauser Verlag, Basel/Switzerland 2009

Abstract response to antibiotic therapy, and early detection of


Objective and design The aim of this study was to assess complications in the infectious process.
the value of procalcitonin (PCT), C-reactive protein (CRP),
tumor necrosis factor-alpha (TNF-a), interleukin (IL)-1b, Keywords CRP  PCT  Cytokines  IL-1b  IL-8 
IL-8, and soluble TNF receptor II (sTNFRII) in early and sTNFRII  ALL  Children
rapid diagnosis of infection in neutropenic children with
acute lymphoblastic leukemia (ALL) and to distinguish
bacterial from viral infections. Introduction
Patients The study included five groups (A, B, C, D, and
E) of children with ALL undergoing intensive chemo- In recent years, intensification of cytostatic chemotherapy
therapy. Groups A and B consisted of neutropenic children for leukemia has resulted in higher remission rates. How-
with bacterial and viral infection, respectively. Groups C ever, this success was accompanied by an increase in the
and D consisted of nonneutropenic children with bacterial incidence of infections due to neutropenia [1]. Infections
and viral infection, respectively. Group E consisted of account for about 70% of fatal complications in acute
children without neutropenia and without fever. leukemia [2]. Since bacterial infections are life-threatening
Methods In all groups, blood samples were collected for neutropenic patients, early diagnosis of these severe
upon admission and then for 7 days on a daily basis. Levels infections would be helpful in management decisions
of CRP, PCT, TNF-a, IL-1b, IL-8, and sTNFRII were regarding antibacterial therapy and hospitalization [3]. In
determined in all blood samples. these patients, bacteremia can occur without any major
Results We found a highly significant difference in PCT complications and, apart from fever, without striking
levels between bacterial and nonbacterial episodes. Sensi- symptoms; however, it can rapidly lead to septic shock and
tivity and specificity of PCT were 94 and 96.5%, death [4].
respectively. In the past decades there have been major advances in
Conclusions Serial measurement of PCT levels on a daily the diagnostics of severe infections [5]. However, there are
basis seems to be helpful for early prediction of severe only a few inflammatory parameters, which are sensitive
bacterial infections, monitoring febrile episodes regarding but not specific markers and of limited validity in neutro-
penic patients. Therefore, various serological markers have
been applied for early diagnosis of infections in neutro-
Responsible Editor: K. Visvanathan.
penic febrile patients with ALL [1, 2, 5, 6].
M. Hatzistilianou (&)  A. Rekliti  F. Athanassiadou  The most common parameter is C-reactive protein
D. Catriu (CRP), an acute-phase protein produced after proinflam-
2nd Department of Paediatrics, matory cytokine release [7]. The CRP concentration rises
Aristotle University of Thessaloniki,
within 24 h and is elevated in almost all cases of inflam-
Agiou Ioannou 23, Kalamaria,
551 32 Thessaloniki, Greece mation and infection, but its reliability as a marker of
e-mail: nontas@topo.auth.gr infection is hampered by very low specificity [7, 8].
340 M. Hatzistilianou et al.

Interactions among cells of the immune system during 214 years (6.8 3.1 years) and group D consisted of 11
infections are largely controlled by soluble mediators nonneutropenic children with 35 febrile episodes due to
named cytokines, such as TNF-a, IL-1b, IL-8, and others virus infections or of unknown origin, aged 114 years
[9, 10]. Challenge of healthy volunteers with endotoxin is (5.9 2.1 years). Group E consisted of 49 children with-
followed by increased serum levels of TNF-a, IL-1b, and out neutropenia and without fever, aged 114 years
IL-8. Moreover, not only bacterial compounds but also (5.6 2.9 years).
components of Gram-positive bacteria, fungi or viruses are Neutropenia was defined as an absolute neutrophil count
potent cytokine-inducing agents [11, 12]. (ANC) \0.5 9 109 or absolute leukocyte count \1 9 109
TNF-a is secreted by macrophages and other antigen- [20].
presenting cells and provides stimulatory signal for T-cell Fever was defined as axillary temperature of [38.0C
activation [13]. Therefore, TNF-a can promote both over a 6-h observation period or [38.5C once [21].
humoral and cellular immune response. TNF has two According to the classification of the International
receptors. These receptors are found on many cell types, Immunocompromised Host Society (ICHS) [21], febrile
are cleaved from activated cells, and act to control TNF-a episodes in neutropenic cancer patients are (a) microbio-
activity by competing with cell-bound TNF-a receptor. logically or virus documented infection, (b) clinically
Elevated serum levels of the cytokines and receptors in a documented infection and fever of unknown origin (FUO).
disease condition represent activation of the immune sys- The infections of Groups A, B, C, and D were infections of
tem [14]. the upper and lower respiratory system, infections of the
Procalcitonin (PCT) is a novel peptide that consists of urinary tract system, skin infections, and infections of the
116 amino acids with the same sequence as the prohor- gastrointestinal system. The diagnosis of microbiological
mone of calcitonin. PCT was described as a new and infections was based on positive cultures of blood, urine,
innovative parameter of infection in 1993 [15]. PCT is and throat swabs. The diagnosis of urinary tract infection
presumably synthesized in tissues other than thyroid required both symptoms and significant growth of a single
C-cells, which are the source of calcitonin in normal microorganism[105 cfu/ml in urine culture. Therefore, the
physiology [16]. According to recent studies the major urine cultures were quantitative. Bacteremia was defined as
source of PCT seemed to be the liver and different cells, fever with positive blood cultures for bacteria (peripheral
and PCT may thus be considered as an acute-phase protein blood or from the central venous indwelling catheter), with
[17]. Serum levels are very low in healthy individuals or without septic symptoms and with no other clinical signs
(\0.5 ng/ml) and in severe infections can reach up to of infection. Clinically documented infection was defined
1,000 ng/ml without changes in serum calcitonin levels as fever in connection with unambiguous diagnostic signs
[18]. The concentration changes of PCT during infection of localized infection, e.g., pneumonia or skin/soft tissue
are rapid and the PCT molecule is stable and easy to inflammation. FUO was defined as fever [38C over at
determine, making it a potentially useful marker [19, 20]. least 1 h or twice within 12 h without evident cause, or
The aim of this study was to assess the value of markers fever without clinical or microbiological evidence. Fever
of inflammation PCT, CRP, TNF-a, IL-1b, IL-8, and was regarded as nonbacterial if there was FUO or it was in
sTNFRII in early and rapid diagnosis of infection and in association with blood products transfusion and/or che-
distinguishing between bacterial and viral infections, in the motherapy with high-dose cytosine arabinose for the
hope of reducing the number of ineffective antibiotic duration of fever with no other sign of infection or docu-
treatments given. mented virus infection [22, 23].
In all groups (A, B, C, D, and E) blood samples were
collected upon admission and before the start of any anti-
Patients and methods bacterial treatment and then for 7 days on a daily basis. For
all children the same blood tests and cultures were per-
The study included five groups (A, B, C, D, and E) of formed. The blood samples were used for culture,
children with acute leukemia undergoing intensive serological tests, virus antibodies, leukocyte count, and for
chemotherapy. analysis of levels of CRP, PCT, TNF-a, IL-1b, IL-8, and
Group A consisted of 16 neutropenic children with 54 sTNFRII.
febrile episodes due to bacterial infections, aged 1 CRP concentration in serum was determined by
14 years (5.8 2.9 years). Group B consisted of 13 neu- nephelometry.
tropenic children with 32 febrile episodes due to virus PCT concentrations were determined by immunolumi-
infections or of unknown origin. nescence assay using a diagnostic kit (LUMItest
Group C consisted of 15 children without neutropenia Pro-Calcitonin, BRAHMS Diagnostica, Berlin, Germany).
with 51 febrile episodes due to bacterial infections, aged This assay, which is specific for procalcitonin, requires
Procalcitonin as an early marker of bacterial infection 341

20 ll serum and can be completed within 1 h. The detec- The mean values of PCT in children with bacterial
tion limit is 0.1 ng/ml. infection (groups A and C) were higher than in those
Serum TNF-a, IL-1b, IL-8, and sTNFRII were measured without bacterial infection (groups B and D). The differ-
as duplicates by enzyme-linked immunosorbent assay ence was statistically significant (p \ 0.001).
(ELISA) method using commercially available ELISA On the first day the mean value of PCT in leukemic
kits (R&D Systems) according to the manufacturers children with neutropenia and bacterial infection (group A)
guidelines. was lower than those with bacterial infection without
neutropenia (group C). The difference was statistically
significant (p \ 0.05) (Fig. 3).
Statistical analysis All children in groups A and C had PCT concentrations
on admission[2 ng/ml. Based on a cutoff level of 2 ng/ml,
The data of this study were evaluated with descriptive PCT showed sensitivity and specificity of 94% and 96.5%,
statistical methods (as mean values standard error, SE) respectively, in group A, and 96.5% and 97%, respectively,
and using Students t test. Differences of continuous vari- in group C for predicting bacteremia. PCT was, in any
able were assessed with MannWhitney and U Wilcoxon
test.
A
The correlation analysis was carried out with Pearsons 4500 B
4000 C
correlation coefficient. D
3500 E
The diagnostic properties of the test were first investi-

Leukocyes
3000
gated by receiver-operating characteristic (ROC) analysis. 2500
This technique summarizes the validity coefficients of a 2000
1500
test and provides an overall index of diagnostic accuracy
1000
(the area under the ROC curve) from a plot of sensitivity 500
against the false-positive rate (1-specificity) for all possible 0
1 2 3 4 5 6 7
cutoff scores. ROC curves were generated by the Statistical
Time (Days)
Program for Social Science version 11 (SPSS). The area
under the ROC curve (AUC) is proportional to the proba- Fig. 1 The mean values of leukocytes in all groups
bility of a correct distinction. An area under the ROC curve
of 0.5 is obtained if the discriminatory ability of a test is no
better than chance. An area of 1.0 corresponds to perfect 20
A
18
discriminatory ability. Various tests for a single patient can 16
B
C
CRP (mg/dl)

be compared with the use of the ROC scale. 14 D


In all statistical tests, values of p \ 0.05 were consid- 12 E
10
ered significant. 8
6
4
2
Results 0
1 2 3 4 5 6 7
Time (Days)
The mean values of leukocytes in all groups during the
7 days are shown in Fig. 1. In febrile patients, the median Fig. 2 Mean values of CRP in all groups
duration of neutropenia following admission was 2.8 days.
The mean values of CRP in all groups all days are
20
shown in Fig. 2. CRP concentrations were elevated in all A
patients. The mean values of CRP in children with bacterial B
15
PCT (ng/ml)

infection (groups A and C) were higher than those without C


bacterial infection (groups B and D). The difference was 10 D
not quite statistically significant on the first day. On the E
second day there was an increase of CRP mean values from 5
9.5% to 17% in groups A, B, C, and D of children with
fever and a continuous, gradual decrease during subsequent 0
1 2 3 4 5 6 7
days (Fig. 2).
Time (Days)
The mean values of PCT in all groups during all days
are shown in Fig. 3. Fig. 3 Mean values of PCT in all groups
342 M. Hatzistilianou et al.

case, a better marker, taking into account its specificity and The mean values of IL-8 in all groups all days are shown
sensitivity. In viral infections PCT concentrations remained in Table 1.
low, with only 4.2% of patients having PCT concentrations Infection was associated with enhanced release of IL-8
[1 ng/ml and a maximum value of 1.94 ng/ml for children in circulation, and increased IL-8 levels in the serum were
of group B and 1.80 ng/ml for children of group D. This closely associated with fever. IL-8 was not detectable in
test, with this cutoff, was clearly better than any other any of the afebrile nonneutropenics (p \ 0.001). In con-
combination. trast, in the majority of patients with febrile neutropenia,
The mean values of PCT in groups A and C decreased elevated IL-8 levels were detectable in the serum. IL-8
on the second day more than 30% while for groups B and serum concentrations were significantly higher in bacterial
D the decrease was less than 10% at the same time point. infections groups than in the non-bacterial-infection
The difference was statistically significant (p \ 0.05). groups.
The discriminatory power of PCT and CRP markers for On the first day the mean value of IL-8 in leukemic
infection in neutropenic patients was evaluated in terms of children with bacterial infection (group A and C) were
the area under ROC curves (AUC). As shown in Fig. 4, higher than those without bacterial infection (group B and
PCT was the best discriminator for bacterial infection on D).
the first day (AUC value of 0.873, whilst CRP was less On the second day there was an increase of IL-8 mean
powerful with a value of 0.702). values from 15% to 35% in groups A, B, C, and D of
The mean values and standard error (SE) of IL-1b in all children with fever and a continuous, gradual decrease
groups during all days are shown in Table 1. during subsequent days (Table 1).
On the first day the mean value of IL-1b in leukemic The discriminatory power of PCT and IL-8 markers for
children with bacterial infection (group A and C) were infection in neutropenic patients was evaluated in terms of
higher than those without bacterial infection (group B and the area under ROC curves (AUC). As shown in Fig. 6,
D). On the second day there was an increase of IL-1b mean PCT was the best discriminator for bacterial infection the
values by more than 25% in all groups of children with first day (AUC value 0.875, whilst IL-8 was less powerful
fever and a continuous, gradual decrease during subsequent with a value of 0.750). The difference was statistically
days (Table 1). significant (p \ 0.01).
The discriminatory power of PCT and IL-1b markers for The mean values of TNF-a in all groups all days are
infection in neutropenic patients was evaluated in terms of shown in Table 2.
the area under ROC curves (AUC). As shown in Fig. 5, TNF-a was excessively produced in the patient group
PCT was the best discriminator for bacterial infection the experiencing febrile episodes. No statistically significant
first day (AUC value 0.875, whilst IL-1b was less powerful differences were demonstrated between patients with neu-
with a value of 0.616). The difference was statistically tropenia and febrile episodes and those without neutropenia
significant (p \ 0.001). and febrile episodes. Therefore, the variability of serum
levels in patients with major infection cannot be explained
on the basis of different degrees of leucopenia. The serum
ROC Curve levels of TNF-a are higher in patients with febrile episodes.
1,0
On the first day the mean value of TNF-a in leukemic
children with bacterial infection (group A and C) were
,8
higher than those without bacterial infection (group B and
Sensitivity

D).
,5
On the second day there was an increase of TNF-a mean
Source of the Curve values from 33% to 42% in groups A, B, C, and D of
,3
CRP children with fever and a continuous, gradual decrease
0,0 PCT
during subsequent days (Table 2).
0,0 ,3 ,5 ,8 1,0 The discriminatory power of PCT and TNF-a markers
1 - Specificity for infection in neutropenic patients was evaluated in terms
of the area under ROC curves (AUC). As shown in Fig. 7,
Area Under the Curve
PCT was the best discriminator for bacterial infection the
Test Result Variable(s) Area first day (AUC value 0.875, whilst TNF-a was less pow-
PCT ,873 erful with a value of 0.777). The difference was statistically
CRP ,702
significant (p \ 0.05).
Fig. 4 Prediction of bacterial infection: ROC curve, with area under The mean values of sTNFRII in all groups all days are
curve of 0.873 for PCT and 0.702 for CRP shown in Table 2.
Procalcitonin as an early marker of bacterial infection 343

Table 1 Mean values and SE of IL-8 and IL-1b in all groups on all days
Groups Days
1 2 3 4 5 6 7

IL-8 (pg/ml)
A 145.4 12.1 159.3 15.1 134.1 15.4 121.3 10.6 99.1 3.2 79.8 7.7 52.2 6.8
B 58.2 5.4 72.9 7.2 54.6 3.2 49.1 2.1 33.7 1.5 29.3 0.8 24 1.9
C 176.1 10.8 199 13.8 169.3 18.1 142 11.7 112 27.5 88.5 5.3 65.7 3.9
D 42.5 2.3 62.1 3.8 48.5 4.8 44.8 3.3 34.5 2.3 25.9 2.2 24 1.7
E 28 2.8 26.3 1.9 27.7 1.9 22.9 0.9 18.2 0.8 19.1 0.5 15.9 0.6
IL-1b (pg/ml)
A 10.5 1.2 17 2.4 12.8 4 8.4 1.3 5.7 1.4 4.8 1.2 4.6 1.2
B 5.1 14 11.8 2.7 7 0.7 6.9 2.6 5.3 0.3 5 2.2 4.3 1.4
C 15.6 2.1 35.7 3.2 19.6 4 10.1 1.9 8.5 2.2 5.2 0.8 4.0 1.6
D 5.3 0.8 8.3 1.8 7.4 1.2 6.3 1.2 5.8 2.6 5.9 1.6 4.9 1.3
E 5.2 1.9 7.3 0.5 6.5 0.9 6.1 0.9 5.2 0.8 5.1 0.4 4.8 0.8

ROC Curve On the first day the mean value of sTNFRII in children
1,0
with leukemia and with bacterial infection (group A and C)
were higher than those without bacterial infection (group B
,8
and D).
Sensitivity

,5
On the second day there was an increase of sTNFRII
Source of the Curve mean values from 10% to 19% in groups A, B, C and D of
,3 children with fever and a continuous, gradual decrease
IL1B
during subsequent days (Table 2).
0,0 PCT The discriminatory power of PCT and sTNFRII markers
0,0 ,3 ,5 ,8 1,0
for infection in neutropenic patients was evaluated in terms
1 - Specificity
of the area under ROC curves (AUC). As shown in Fig. 7,
Area Under the Curve PCT was the best discriminator for bacterial infection the
Test Result Area
Variable(s)
first day (AUC value 0.875, whilst sTNFRII was less
PCT ,875 powerful with a value of 0.795). The difference was sta-
IL1B ,616 tistically significant (p \ 0.001).
Fig. 5 Prediction of bacterial infection: ROC curve, with area under
curve of 0.875 for PCT and 0.616 for IL-1b
Discussion
ROC Curve
1,0 It is common practice to hospitalize patients with neu-
tropenia due to chemotherapy for as long as the
,8 neutropenia with or without fever persists and to admin-
Sensitivity

ister broad-spectrum intravenous antibiotics [24]. For this


,5
reason, the main aim of this work was to identify markers
Source of the Curve
,3
for rapid diagnosis of infection and for distinguishing
IL8 between bacterial and viral infections, in the hope of
0,0 PCT reducing the number of ineffective antibiotic treatments
0,0 ,3 ,5 ,8 1,0
given. A diagnostic test to define the high-risk group of
1 - Specificity neutropenic patients with fever would be of great value
Area Under the Curve [25].
Test Result Area Our data support the view of some authors that leuco-
Variable(s)
PCT ,875 cyte count has little value in differentiating the type of
IL8 ,750
infection in all groups.
Fig. 6 Prediction of bacterial infection: ROC curve, with area under A valuable marker of infection in this setting would
curve of 0.875 for PCT and 0.750 for IL-8 be associated with high sensitivity and specificity for
344 M. Hatzistilianou et al.

Table 2 Mean values and SE of sTNFRII and TNF-a in all groups on all days
Groups Days
1 2 3 4 5 6 7

TNF-a (pg/ml)
A 46.1 1.5 79.6 9.7 59.3 4.7 54.4 6.9 43.2 5.1 31.5 4.1 15.2 1.4
B 7.9 1.8 8.9 0.8 6.1 0.3 5.8 0.5 5.4 0.9 4.8 0.3 3.2 0.4
C 82.4 6.9 133.4 11.2 91.3 7.3 77.9 11.3 54.6 5.8 43.0 3.1 29.2 3.8
D 9.2 2.7 10.8 3.3 8.4 0.6 7.4 0.2 6.8 0.4 5.3 0.3 4.3 0.5
E 4.8 0.8 5.1 0.3 4.2 0.5 4.9 0.7 4.7 0.4 4.8 0.6 4.5 0.6
sTNFRII (pg/ml)
A 5666 33 5762 86 4841 23 4830 33 4292 75 3627 74 3333 45
B 832 34 1046 23 813 58 691 47 659 31 586 58 564 56
C 3887 87 5180 87 5148 75 4591 44 4381 75 4281 57 3754 28
D 1626 46 2338 57 2110 63 1195 53 1151 77 1012 54 974 37
E 913 26 1002 24 982 36 909 54 835 76 835 22 592 53

ROC Curve in groups A and C had PCT concentrations on admission


1,0
[2 ng/ml and, based on a cutoff level of 2 ng/ml PCT, the
sensitivity and specificity were 94% and 96.5%, respec-
,8
Source of the Curve tively, in group A, and 96.5% and 97%, respectively, in
Sensitivity

,5 group C for predicting bacteremia. PCT was, in any case, a


TNFA
better marker, taking into account its specificity and sen-
,3 STNFRII sitivity. In this study PCT was the best marker for
identifying severe bacterial infections.
PCT
0,0 In our study, PCT concentrations were significantly
0,0 ,3 ,5 ,8 1,0
lower in neutropenic than in nonneutropenic patients with
1 - Specificity
bacterial infections. Recently it has been demonstrated that
Area Under the Curve immunoreactive cells as well as neutrophils are possible
sources of PCT production [26]. On the other hand, because
Test Result Area
hepatic cells are now also considered to be a source of PCT
Variable(s)
PCT 0.875 production, liver damagefor example, caused by che-
STNFRII 0.795 motherapycould be responsible for reduced PCT levels in
hematological patients [27]. Our data support this hypoth-
TNFA 0.777
esis, because PCT levels were lower in children with
Fig. 7 Prediction of bacterial infection: ROC curve, with area under neutropenia and bacterial infection than in children with
curve of 0.875 for PCT, 0.777 for TNF-a, and 0.795 for sTNFRII bacterial infection without neutropenia.
PCT determination may be very helpful in cases of fever
predicting infection and a half-life short enough to be without any signs of clinical infection. Low levels of PCT
useful in follow-up. support the diagnosis of virus infection or unknown origin.
A rapid and reliable biochemical marker that is highly The duration of antibiotic treatment could possibly be
associated with bacterial infections could improve the care of determined based on serum concentrations of PCT levels. If
febrile and neutropenic children by providing the clinician the levels return to normal once the fever has ceased, anti-
with more definitive evidence of the etiology of fever [25]. biotic treatment can be shorter than in cases with prolonged
The ability to identify patients with high risk for high levels of PCT. Decreasing PCT levels during the febrile
bacterial infection early in their course may lead to course reflected defervescence, clinical improvement, and/
improvement in clinical management and outcome. For or successful antibacterial therapy, independently of etiol-
this reason, in the present study we assessed the value of ogy or site of infection [28].
PCT, CRP, TNF-a, IL-1b, IL-8, and sTNFRII in all groups Serial measurement of PCT serum levels on a daily basis
for seven continuous days. in neutropenic febrile patients with ALL seems to be helpful
We found a highly significant difference in PCT levels for early prediction of severe systemic bacterial infections,
between bacterial and nonbacterial episodes. All children monitoring of all febrile episodes regarding the response to
Procalcitonin as an early marker of bacterial infection 345

antibiotic therapy, and early detection of complications in Our own data showed that significantly elevated IL-8
the infectious process and to monitor neutropenic ALL serum levels in the group with bacterial infection persist for
children with septic complications [29, 30]. at least 2 days. Furthermore, in children without fever,
CRP is an excellent marker of infection, but it cannot be serum levels were about fourfold lower than the serum
used to distinguish between bacterial and viral infections concentrations measured in children with fever, supporting
[8, 31]. Although CRP levels appeared to be slightly lower the possible role of IL-8 as a diagnostic marker in patients
in patients with nonbacterial infection, the difference was with fever and chemotherapy-induced neutropenia.
not significant. The results indicated that CRP, the most IL-8 is a chemokine responsible for migration of neu-
commonly used marker, was elevated in all conditions trophils and macrophages to the site of inflammation and is
related with febrile neutropenia, including nonbacterial not normally present in high quantities in normal children
episodes. [12]. It has been reported that neutrophils and monocytes
Moreover, because PCT rises earlier than CRP after are the main source of IL-8. The number of these cells is
bacterial stimuli, this test may prove to be more accurate at more than eight- to tenfold lower in children with neutro-
the beginning of infection. Total and differential leukocyte penia due to chemotherapy in comparison with healthy
count, which are commonly used in the decision algorithm, children. Therefore other cell types must be considered as a
performed poorly compared with PCT and CRP, raising the major source of IL-8 in neutropenic patients. It has been
question of their utility [32]. We therefore suggest that this reported that blood fibrocytes and endothelial cells secrete
approach may be abandoned as a routine first screening cytokines such as IL-8. In our series the specificity of IL-8
method. was high but its sensitivity for distinguishing between viral
In comparison with PCT, CRP levels showed a similar and bacterial infections was low.
course, characterized by a more delayed increase and Our study demonstrates that an increase in serum con-
decrease. However, there was wide overlap of CRP levels centrations of TNF-a in response to bacterial infections can
between the different kinds of infections at admission, as be detected also in children with leukemia during severe
well as during the entire febrile episode. In neutropenic chemotherapy-induced neutropenia. Neutropenic patients
children with bacterial infection after an appropriate ther- without infection showed low levels of TNF-a, but serum
apy, PCT and CRP levels decreased in 24 and 48 h, TNF-a increased when the patients developed bacterial
respectively. infections. Thus, even during severe neutropenia due to
The ROC curve illustrates the superior sensitivity and chemotherapy, the cytokine network in patients with ALL
specificity of procalcitonin compared with CRP, as a can be modulated in response to external stimulation [8,
consequence of the wide range of CRP concentrations in all 12]. The increase in serum TNF-a was followed by a
categories of infection. PCT offers better specificity than decrease in all patients where signs of infection resolved
CRP for differentiating between viral and bacterial etiology before hematopoietic reconstitution, whereas a continuous
of febrile episodes in immunosuppressed children. PCT increase in serum TNF-a would be expected if this was an
increases earlier after infectious stimuli than do CRP and early sign of hematopoietic reconstitution.
other inflammatory cytokines. Increased sTNFRII serum levels as a result of bacterial
Our results indicated that IL-1b, the proinflammatory infection were observed in this study and showed a clear
cytokine, was elevated in all conditions related with febrile correlation with the severity of the infection [32]. This
neutropenia, including bacterial and nonbacterial episodes. study documented increased sTNFRII serum levels
IL-1b is produced by cells of the immune system after throughout the febrile episodes even during the period of
immunologic stimuli and is vital to normal immunity. The clinical improvement. PCT was the best parameter because
increase in serum IL-1b paralleled the increase in serum successful antibacterial therapy and clinical improvement
PCT, but more slowly and remained higher even after an was reflected in persistently low levels or a decrease in
appropriate therapy. So, serum PCT is a more useful levels, respectively, of the kind of infection. In contrast
clinical parameter than serum IL-1b during infections, sTNFRII was unsuitable for the assessment of the febrile
because serum PCT shows a relatively larger increase and course because of a high percentage of persistently high
is not affected by bone marrow reconstitution. levels.
This study enhances our understanding of the response In conclusion, in this study we found that PCT was a
to bacterial infection in neutropenic children. The use of a better marker than CRP, IL-8, IL-1b, TNF-a or sTNFRII
combination of markers as opposed to a comparison for distinguishing between bacterial and viral infections in
between markers produced a screening test with extraor- children with ALL who were neutropenic and nonneu-
dinary sensitivity and specificity [12, 32]. This tropenic on the first day of fever. Serial measurements of
demonstrates that PCT and IL-8 are produced in a clini- PCT serum levels on a daily basis in neutropenic and
cally relevant time in response to bacterial infection. nonneutropenic febrile patients with ALL offered the best
346 M. Hatzistilianou et al.

discrimination of all parameters between mild and serious 13. Nijsten MW, Olinga P, The TH, de Vries EG, Koops HS.
bacterial infections. During treatment, decreasing PCT Procalcitonin behaves as a fast responding acute phase protein
in vivo and in vitro. Crit Care Med. 2000;28:45861.
levels reflected defervescence, rapid or gradual clinical 14. Hoffmann G, Schobersberger W. Procalcitonin as a marker. Acta
improvement, and successful antibacterial therapy, irre- Anaesthesiol Scand. 2003;47(2):23743.
spective of the kind of bacterial infection. Also, PCT seems 15. Gendel D, Bohuon C. Procalcitonin a marker of bacterial infec-
to be helpful for monitoring all febrile episodes regarding tion. Infection. 1997;25:1338.
16. Whang KT, Steinwald PM, White JC. Serum calcitonin precur-
the response to antibiotic therapy, early detection of com- sors in sepsis and systemic inflammation. J Clin Endocrinol
plications in the infectious process, and monitoring Metab. 1998;83:3296301.
neutropenic and nonneutropenic ALL children with septic 17. Emporiadou M, Hatzistilianou M, Haidopoulou K, Aggouridaki
complications. C, Reklity A, Magnisali C, et al. Procalcitonin and inflam-
matory cytokines in children with asthma. Eur J Inflamm.
2005;3(2):7581.
18. Bernard L, Ferriere F, Casassus P, Malas E, Leveque S, Gu-
illevin L. Procalcitonin as an early marker of bacterial
infection in severely neutropenic febrile adults. Clin Infect Dis.
References 1998;27:9147.
19. Hatzistilianou M, Hitoglou S, Gougoustamou D, Rekliti A,
1. Finberg RW, Talcott JA. Fever and neutropenia-how to use a new Tzouvelekis G, Nanas C, et al. Serum procalcitonin, adenosine
treatment strategy. N Engl J Med. 1999;341(5):3628. deaminase and its isoenzymes in the aetiological diagnosis of
2. Crokaert F. Febrile neutropenia in children. Int J Antimicrob pneumonia in children. Int J Immunopathol Pharmacol.
Agent. 2000;16:1737. 2002;15(2):11923.
3. Schimpff SC. Infections in the cancer patients-diagnosis, pre- 20. The Immunocompromised Host Society. The design, analysis,
vention and treatment. In: Mandell GL, Douglas RG, Bennet JE, and reporting of clinical trials on the empirical antibiotic man-
editors. Principles and practice of infectious diseases. New York: agement of the neutropenic patient. J Infect Dis. 1990;161:397
Churchill Livingstone; 1995. p. 2666. 404.
4. Laws HJ, Schneider DT, Janssen G, Wessalowski R, Dilloo D, 21. Alexander SW, Pizzo PA. Current considerations in the man-
Meisel R, et al. Trends in infections in children with malignant agement of fever and neutropenia. In: Remington JS, Swartz MN,
disease in 2000: comparison of data of 1980/81. Pediatr Hematol editors. Current clinical topics in infectious diseases. Oxford:
Oncol. 2000;24(5):34354. Blackwell; 1999. p. 16080.
5. Von Lilienfeld-Toal M, Dietrich MP, Glasmacher A, Lehmann L, 22. Hughes WT, Armstrong D, Bodey GP, Brown AE, Edwards JE,
Breig P, Hahn C, et al. Markers of bacteremia in febrile neutro- Feld R, et al. Guidelines for the use of antimicrobial agents in
penic patients with hematological malignancies: procalcitonin neutropenic patients with unexplained fever. Clin Infect Dis.
and IL-6 are more reliable than C-reactive protein. Eur J Clin 1997;25:5515.
Macrobiol Infect Dis. 2004;23:53944. 23. Penel N, Fournier C, Clisant S, NGuyen M. Causes of fever and
6. Hitoglou S, Hatzistilianou M, Gougoustamou D, Athanassiadou F, value of C-reactive protein and procalcitonin in differentiating
Kotsis A, Catriu D. Adenosine deaminase activity and isozyme infections from paraneoplastic fever. Support Care Cancer.
levels in serum and peripheral blood lymphocytes in childhood 2004;12:5937.
acute lymphoblastic leukemia. Mol Immunol. 1998;35(11 24. Ruokonen E, Nousiainen T, Pulkki K, Takala J. Procalcitonin
12):7549. concentrations in patients with neutropenic fever. Eur J Clin
7. Petrola V, Toikka P, Irjala K, Mertsola J, Ruuskanen O. Microbiol Infect Dis. 1999;18:2838.
Discrepancy between total white blood cell count and serum 25. Schuttrump S, Binder L, Hagemann T, Berkovie D, Trumper L,
C-reactive protein levels in febrile children. Scand J Infect Dis. Binder C. Procalcitonin: a useful discriminator between febrile
2007;39(6):5605. conditions of different origin in hemato-oncological patients?
8. Kalio R, Bloigu A, Surcel HM, Syrjiada H. C-reactive protein and Ann Hematol. 2003;82:98101.
erythrocyte sedimentation rate in differential diagnosis between 26. Hitoglou S, Hatzistilianou M, Gougoustamou D, Rekliti A,
infections and neoplastic fever in patients with solid tumours and Agguridaki Ch, Athanassiadou F, et al. Serum adenosine deam-
lymphomas. Support Care Cancer. 2001;9:1249. inase and procalcitonin concentrations in neutropenic febrile
9. Hartel C, Deutser M, Lehrnbecher T, Schultz C. Current children with acute lymphoblastic leukemia. Clin Exp Med.
approaches for risk stratification of infectious complications in 2005;5:605.
pediatric oncology. Pediatr Blood Cancer. 2007;49(6):76773. 27. Christ-Crain M, Jaccard-Stolz D, Bingisser R, Gencay M, Huber
10. Sheu JN, Chen MC, Lue KH, Sl Cheng, Lee IC, Chen SM, et al. P, Tamm M, et al. Effect of procalcitonin-guided treatment on
Serum and urine levels of interleukin-6 and interleukin-8 in chil- antibiotic use and outcome in lower respiratory tract infections:
dren with acute pyelonephritis. Cytokine. 2006;36(56):27682. cluster-randomised, single-blinded intervention trial. Lancet.
11. Shereen Mohamed EM, Manar Mohamed M, Manar Mohamed I, 2004;363:6005.
Lobna M, Hadir A. The diagnostic value of C-reactive protein, 28. Hatzistilianou M, Rekleity A, Athanassiadou F, Delutils MA,
interleukin-8, and monocyte chemotactic protein in risk stratifi- Conti P, Catriu D. Serial procalcitonin responses in infection of
cation of febrile neutropenic children with hematologic children with secondary immunodeficiency. Clin Invest Med.
malignancies. J Pediatr Hematol Oncol. 2007;29(3):1316. 2007;30(2):7585.
12. Lehrnbecher T, Venzon D, de Haas M, Chanock SJ, Kuel J. 29. Mitaka C. Clinical laboratory differentiation of infectious versus
Assessment of measuring circulating levels of interleukin-6, non-infectious systemic inflammatory response syndrome. Clin
interleukin-8, C-reactive protein, soluble Fc gamma receptor type Chim Acta. 2005;351(12):1721.
III, and mannose-binding protein in febrile children with cancer 30. Chesney J, Metz C, Stavitsky AB, Bacher M, Bacher R.
and neutropenia. Clin Infect Dis. 1999;29:4149. Regulated production of type I collagen and inflammatory
Procalcitonin as an early marker of bacterial infection 347

cytokines by peripheral blood fibrocytes. J Immunol. 1998; 32. Fleischhack G, Kambeck I, Cipic D, Hasan C, Bode U. Pro-
160:41925. calcitonin in paediatric cancer patients: its diagnostic relevance is
31. Harbarth S, Holechova K, Froidevaux C, Pittet D, Ricou B, Grau superior to that of C-reactive protein, interleukin 6, interleukin 8,
G, et al. Diagnostic value of procalcitonin, interleukin-6 and soluble interleukin 2 receptor and soluble necrosis factor receptor
interleukin-8 in critically III patients admitted with suspected II. Br J Haematol. 2000;111:1093102.
sepsis. J Resp Crit Care Med. 2001;164(3):396402.